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 although 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 too big 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 Note that the Xen shared info page, if configured, shall always be assumed
375 to be dirty. KVM will not explicitly mark it such.
377 4.9 KVM_SET_MEMORY_ALIAS
378 ------------------------
383 :Parameters: struct kvm_memory_alias (in)
384 :Returns: 0 (success), -1 (error)
386 This ioctl is obsolete and has been removed.
396 :Returns: 0 on success, -1 on error
400 ======= ==============================================================
401 EINTR an unmasked signal is pending
402 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
403 instructions from device memory (arm64)
404 ENOSYS data abort outside memslots with no syndrome info and
405 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
406 EPERM SVE feature set but not finalized (arm64)
407 ======= ==============================================================
409 This ioctl is used to run a guest virtual cpu. While there are no
410 explicit parameters, there is an implicit parameter block that can be
411 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
412 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
413 kvm_run' (see below).
420 :Architectures: all except ARM, arm64
422 :Parameters: struct kvm_regs (out)
423 :Returns: 0 on success, -1 on error
425 Reads the general purpose registers from the vcpu.
431 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
432 __u64 rax, rbx, rcx, rdx;
433 __u64 rsi, rdi, rsp, rbp;
434 __u64 r8, r9, r10, r11;
435 __u64 r12, r13, r14, r15;
441 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
453 :Architectures: all except ARM, arm64
455 :Parameters: struct kvm_regs (in)
456 :Returns: 0 on success, -1 on error
458 Writes the general purpose registers into the vcpu.
460 See KVM_GET_REGS for the data structure.
467 :Architectures: x86, ppc
469 :Parameters: struct kvm_sregs (out)
470 :Returns: 0 on success, -1 on error
472 Reads special registers from the vcpu.
478 struct kvm_segment cs, ds, es, fs, gs, ss;
479 struct kvm_segment tr, ldt;
480 struct kvm_dtable gdt, idt;
481 __u64 cr0, cr2, cr3, cr4, cr8;
484 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
487 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
489 interrupt_bitmap is a bitmap of pending external interrupts. At most
490 one bit may be set. This interrupt has been acknowledged by the APIC
491 but not yet injected into the cpu core.
498 :Architectures: x86, ppc
500 :Parameters: struct kvm_sregs (in)
501 :Returns: 0 on success, -1 on error
503 Writes special registers into the vcpu. See KVM_GET_SREGS for the
513 :Parameters: struct kvm_translation (in/out)
514 :Returns: 0 on success, -1 on error
516 Translates a virtual address according to the vcpu's current address
521 struct kvm_translation {
523 __u64 linear_address;
526 __u64 physical_address;
538 :Architectures: x86, ppc, mips, riscv
540 :Parameters: struct kvm_interrupt (in)
541 :Returns: 0 on success, negative on failure.
543 Queues a hardware interrupt vector to be injected.
547 /* for KVM_INTERRUPT */
548 struct kvm_interrupt {
558 ========= ===================================
560 -EEXIST if an interrupt is already enqueued
561 -EINVAL the irq number is invalid
562 -ENXIO if the PIC is in the kernel
563 -EFAULT if the pointer is invalid
564 ========= ===================================
566 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
567 ioctl is useful if the in-kernel PIC is not used.
572 Queues an external interrupt to be injected. This ioctl is overleaded
573 with 3 different irq values:
577 This injects an edge type external interrupt into the guest once it's ready
578 to receive interrupts. When injected, the interrupt is done.
580 b) KVM_INTERRUPT_UNSET
582 This unsets any pending interrupt.
584 Only available with KVM_CAP_PPC_UNSET_IRQ.
586 c) KVM_INTERRUPT_SET_LEVEL
588 This injects a level type external interrupt into the guest context. The
589 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
592 Only available with KVM_CAP_PPC_IRQ_LEVEL.
594 Note that any value for 'irq' other than the ones stated above is invalid
595 and incurs unexpected behavior.
597 This is an asynchronous vcpu ioctl and can be invoked from any thread.
602 Queues an external interrupt to be injected into the virtual CPU. A negative
603 interrupt number dequeues the interrupt.
605 This is an asynchronous vcpu ioctl and can be invoked from any thread.
610 Queues an external interrupt to be injected into the virutal CPU. This ioctl
611 is overloaded with 2 different irq values:
615 This sets external interrupt for a virtual CPU and it will receive
618 b) KVM_INTERRUPT_UNSET
620 This clears pending external interrupt for a virtual CPU.
622 This is an asynchronous vcpu ioctl and can be invoked from any thread.
632 :Returns: -1 on error
634 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
640 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
642 :Type: system ioctl, vcpu ioctl
643 :Parameters: struct kvm_msrs (in/out)
644 :Returns: number of msrs successfully returned;
647 When used as a system ioctl:
648 Reads the values of MSR-based features that are available for the VM. This
649 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
650 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
653 When used as a vcpu ioctl:
654 Reads model-specific registers from the vcpu. Supported msr indices can
655 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
660 __u32 nmsrs; /* number of msrs in entries */
663 struct kvm_msr_entry entries[0];
666 struct kvm_msr_entry {
672 Application code should set the 'nmsrs' member (which indicates the
673 size of the entries array) and the 'index' member of each array entry.
674 kvm will fill in the 'data' member.
683 :Parameters: struct kvm_msrs (in)
684 :Returns: number of msrs successfully set (see below), -1 on error
686 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
689 Application code should set the 'nmsrs' member (which indicates the
690 size of the entries array), and the 'index' and 'data' members of each
693 It tries to set the MSRs in array entries[] one by one. If setting an MSR
694 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
695 by KVM, etc..., it stops processing the MSR list and returns the number of
696 MSRs that have been set successfully.
705 :Parameters: struct kvm_cpuid (in)
706 :Returns: 0 on success, -1 on error
708 Defines the vcpu responses to the cpuid instruction. Applications
709 should use the KVM_SET_CPUID2 ioctl if available.
712 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID
713 configuration (if there is) is not corrupted. Userspace can get a copy
714 of the resulting CPUID configuration through KVM_GET_CPUID2 in case.
715 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model
716 after running the guest, may cause guest instability.
717 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc...
718 may cause guest instability.
722 struct kvm_cpuid_entry {
731 /* for KVM_SET_CPUID */
735 struct kvm_cpuid_entry entries[0];
739 4.21 KVM_SET_SIGNAL_MASK
740 ------------------------
745 :Parameters: struct kvm_signal_mask (in)
746 :Returns: 0 on success, -1 on error
748 Defines which signals are blocked during execution of KVM_RUN. This
749 signal mask temporarily overrides the threads signal mask. Any
750 unblocked signal received (except SIGKILL and SIGSTOP, which retain
751 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
753 Note the signal will only be delivered if not blocked by the original
758 /* for KVM_SET_SIGNAL_MASK */
759 struct kvm_signal_mask {
771 :Parameters: struct kvm_fpu (out)
772 :Returns: 0 on success, -1 on error
774 Reads the floating point state from the vcpu.
778 /* for KVM_GET_FPU and KVM_SET_FPU */
783 __u8 ftwx; /* in fxsave format */
800 :Parameters: struct kvm_fpu (in)
801 :Returns: 0 on success, -1 on error
803 Writes the floating point state to the vcpu.
807 /* for KVM_GET_FPU and KVM_SET_FPU */
812 __u8 ftwx; /* in fxsave format */
823 4.24 KVM_CREATE_IRQCHIP
824 -----------------------
826 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
827 :Architectures: x86, ARM, arm64, s390
830 :Returns: 0 on success, -1 on error
832 Creates an interrupt controller model in the kernel.
833 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
834 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
835 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
836 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
837 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
838 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
839 On s390, a dummy irq routing table is created.
841 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
842 before KVM_CREATE_IRQCHIP can be used.
848 :Capability: KVM_CAP_IRQCHIP
849 :Architectures: x86, arm, arm64
851 :Parameters: struct kvm_irq_level
852 :Returns: 0 on success, -1 on error
854 Sets the level of a GSI input to the interrupt controller model in the kernel.
855 On some architectures it is required that an interrupt controller model has
856 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
857 interrupts require the level to be set to 1 and then back to 0.
859 On real hardware, interrupt pins can be active-low or active-high. This
860 does not matter for the level field of struct kvm_irq_level: 1 always
861 means active (asserted), 0 means inactive (deasserted).
863 x86 allows the operating system to program the interrupt polarity
864 (active-low/active-high) for level-triggered interrupts, and KVM used
865 to consider the polarity. However, due to bitrot in the handling of
866 active-low interrupts, the above convention is now valid on x86 too.
867 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
868 should not present interrupts to the guest as active-low unless this
869 capability is present (or unless it is not using the in-kernel irqchip,
873 ARM/arm64 can signal an interrupt either at the CPU level, or at the
874 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
875 use PPIs designated for specific cpus. The irq field is interpreted
878 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
879 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
881 The irq_type field has the following values:
884 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
886 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
887 (the vcpu_index field is ignored)
889 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
891 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
893 In both cases, level is used to assert/deassert the line.
895 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
896 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
899 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
900 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
901 be used for a userspace interrupt controller.
905 struct kvm_irq_level {
908 __s32 status; /* not used for KVM_IRQ_LEVEL */
910 __u32 level; /* 0 or 1 */
917 :Capability: KVM_CAP_IRQCHIP
920 :Parameters: struct kvm_irqchip (in/out)
921 :Returns: 0 on success, -1 on error
923 Reads the state of a kernel interrupt controller created with
924 KVM_CREATE_IRQCHIP into 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;
942 :Capability: KVM_CAP_IRQCHIP
945 :Parameters: struct kvm_irqchip (in)
946 :Returns: 0 on success, -1 on error
948 Sets the state of a kernel interrupt controller created with
949 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
954 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
957 char dummy[512]; /* reserving space */
958 struct kvm_pic_state pic;
959 struct kvm_ioapic_state ioapic;
964 4.28 KVM_XEN_HVM_CONFIG
965 -----------------------
967 :Capability: KVM_CAP_XEN_HVM
970 :Parameters: struct kvm_xen_hvm_config (in)
971 :Returns: 0 on success, -1 on error
973 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
974 page, and provides the starting address and size of the hypercall
975 blobs in userspace. When the guest writes the MSR, kvm copies one
976 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
981 struct kvm_xen_hvm_config {
991 If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the
992 KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl.
993 This requests KVM to generate the contents of the hypercall page
994 automatically; hypercalls will be intercepted and passed to userspace
995 through KVM_EXIT_XEN. In this case, all of the blob size and address
998 No other flags are currently valid in the struct kvm_xen_hvm_config.
1003 :Capability: KVM_CAP_ADJUST_CLOCK
1006 :Parameters: struct kvm_clock_data (out)
1007 :Returns: 0 on success, -1 on error
1009 Gets the current timestamp of kvmclock as seen by the current guest. In
1010 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
1013 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
1014 set of bits that KVM can return in struct kvm_clock_data's flag member.
1016 The following flags are defined:
1018 KVM_CLOCK_TSC_STABLE
1019 If set, the returned value is the exact kvmclock
1020 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called.
1021 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant
1022 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try
1023 to make all VCPUs follow this clock, but the exact value read by each
1024 VCPU could differ, because the host TSC is not stable.
1027 If set, the `realtime` field in the kvm_clock_data
1028 structure is populated with the value of the host's real time
1029 clocksource at the instant when KVM_GET_CLOCK was called. If clear,
1030 the `realtime` field does not contain a value.
1033 If set, the `host_tsc` field in the kvm_clock_data
1034 structure is populated with the value of the host's timestamp counter (TSC)
1035 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field
1036 does not contain a value.
1040 struct kvm_clock_data {
1041 __u64 clock; /* kvmclock current value */
1053 :Capability: KVM_CAP_ADJUST_CLOCK
1056 :Parameters: struct kvm_clock_data (in)
1057 :Returns: 0 on success, -1 on error
1059 Sets the current timestamp of kvmclock to the value specified in its parameter.
1060 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1063 The following flags can be passed:
1066 If set, KVM will compare the value of the `realtime` field
1067 with the value of the host's real time clocksource at the instant when
1068 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final
1069 kvmclock value that will be provided to guests.
1071 Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored.
1075 struct kvm_clock_data {
1076 __u64 clock; /* kvmclock current value */
1085 4.31 KVM_GET_VCPU_EVENTS
1086 ------------------------
1088 :Capability: KVM_CAP_VCPU_EVENTS
1089 :Extended by: KVM_CAP_INTR_SHADOW
1090 :Architectures: x86, arm, arm64
1092 :Parameters: struct kvm_vcpu_event (out)
1093 :Returns: 0 on success, -1 on error
1098 Gets currently pending exceptions, interrupts, and NMIs as well as related
1103 struct kvm_vcpu_events {
1107 __u8 has_error_code;
1128 __u8 smm_inside_nmi;
1132 __u8 exception_has_payload;
1133 __u64 exception_payload;
1136 The following bits are defined in the flags field:
1138 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1139 interrupt.shadow contains a valid state.
1141 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1144 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1145 exception_has_payload, exception_payload, and exception.pending
1146 fields contain a valid state. This bit will be set whenever
1147 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1152 If the guest accesses a device that is being emulated by the host kernel in
1153 such a way that a real device would generate a physical SError, KVM may make
1154 a virtual SError pending for that VCPU. This system error interrupt remains
1155 pending until the guest takes the exception by unmasking PSTATE.A.
1157 Running the VCPU may cause it to take a pending SError, or make an access that
1158 causes an SError to become pending. The event's description is only valid while
1159 the VPCU is not running.
1161 This API provides a way to read and write the pending 'event' state that is not
1162 visible to the guest. To save, restore or migrate a VCPU the struct representing
1163 the state can be read then written using this GET/SET API, along with the other
1164 guest-visible registers. It is not possible to 'cancel' an SError that has been
1167 A device being emulated in user-space may also wish to generate an SError. To do
1168 this the events structure can be populated by user-space. The current state
1169 should be read first, to ensure no existing SError is pending. If an existing
1170 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1171 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1172 Serviceability (RAS) Specification").
1174 SError exceptions always have an ESR value. Some CPUs have the ability to
1175 specify what the virtual SError's ESR value should be. These systems will
1176 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1177 always have a non-zero value when read, and the agent making an SError pending
1178 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1179 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1180 with exception.has_esr as zero, KVM will choose an ESR.
1182 Specifying exception.has_esr on a system that does not support it will return
1183 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1184 will return -EINVAL.
1186 It is not possible to read back a pending external abort (injected via
1187 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1188 directly to the virtual CPU).
1192 struct kvm_vcpu_events {
1194 __u8 serror_pending;
1195 __u8 serror_has_esr;
1196 __u8 ext_dabt_pending;
1197 /* Align it to 8 bytes */
1204 4.32 KVM_SET_VCPU_EVENTS
1205 ------------------------
1207 :Capability: KVM_CAP_VCPU_EVENTS
1208 :Extended by: KVM_CAP_INTR_SHADOW
1209 :Architectures: x86, arm, arm64
1211 :Parameters: struct kvm_vcpu_event (in)
1212 :Returns: 0 on success, -1 on error
1217 Set pending exceptions, interrupts, and NMIs as well as related states of the
1220 See KVM_GET_VCPU_EVENTS for the data structure.
1222 Fields that may be modified asynchronously by running VCPUs can be excluded
1223 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1224 smi.pending. Keep the corresponding bits in the flags field cleared to
1225 suppress overwriting the current in-kernel state. The bits are:
1227 =============================== ==================================
1228 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1229 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1230 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1231 =============================== ==================================
1233 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1234 the flags field to signal that interrupt.shadow contains a valid state and
1235 shall be written into the VCPU.
1237 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1239 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1240 can be set in the flags field to signal that the
1241 exception_has_payload, exception_payload, and exception.pending fields
1242 contain a valid state and shall be written into the VCPU.
1247 User space may need to inject several types of events to the guest.
1249 Set the pending SError exception state for this VCPU. It is not possible to
1250 'cancel' an Serror that has been made pending.
1252 If the guest performed an access to I/O memory which could not be handled by
1253 userspace, for example because of missing instruction syndrome decode
1254 information or because there is no device mapped at the accessed IPA, then
1255 userspace can ask the kernel to inject an external abort using the address
1256 from the exiting fault on the VCPU. It is a programming error to set
1257 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1258 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1259 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1260 how userspace reports accesses for the above cases to guests, across different
1261 userspace implementations. Nevertheless, userspace can still emulate all Arm
1262 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1264 See KVM_GET_VCPU_EVENTS for the data structure.
1267 4.33 KVM_GET_DEBUGREGS
1268 ----------------------
1270 :Capability: KVM_CAP_DEBUGREGS
1273 :Parameters: struct kvm_debugregs (out)
1274 :Returns: 0 on success, -1 on error
1276 Reads debug registers from the vcpu.
1280 struct kvm_debugregs {
1289 4.34 KVM_SET_DEBUGREGS
1290 ----------------------
1292 :Capability: KVM_CAP_DEBUGREGS
1295 :Parameters: struct kvm_debugregs (in)
1296 :Returns: 0 on success, -1 on error
1298 Writes debug registers into the vcpu.
1300 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1301 yet and must be cleared on entry.
1304 4.35 KVM_SET_USER_MEMORY_REGION
1305 -------------------------------
1307 :Capability: KVM_CAP_USER_MEMORY
1310 :Parameters: struct kvm_userspace_memory_region (in)
1311 :Returns: 0 on success, -1 on error
1315 struct kvm_userspace_memory_region {
1318 __u64 guest_phys_addr;
1319 __u64 memory_size; /* bytes */
1320 __u64 userspace_addr; /* start of the userspace allocated memory */
1323 /* for kvm_memory_region::flags */
1324 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1325 #define KVM_MEM_READONLY (1UL << 1)
1327 This ioctl allows the user to create, modify or delete a guest physical
1328 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1329 should be less than the maximum number of user memory slots supported per
1330 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1331 Slots may not overlap in guest physical address space.
1333 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1334 specifies the address space which is being modified. They must be
1335 less than the value that KVM_CHECK_EXTENSION returns for the
1336 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1337 are unrelated; the restriction on overlapping slots only applies within
1340 Deleting a slot is done by passing zero for memory_size. When changing
1341 an existing slot, it may be moved in the guest physical memory space,
1342 or its flags may be modified, but it may not be resized.
1344 Memory for the region is taken starting at the address denoted by the
1345 field userspace_addr, which must point at user addressable memory for
1346 the entire memory slot size. Any object may back this memory, including
1347 anonymous memory, ordinary files, and hugetlbfs.
1349 On architectures that support a form of address tagging, userspace_addr must
1350 be an untagged address.
1352 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1353 be identical. This allows large pages in the guest to be backed by large
1356 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1357 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1358 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1359 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1360 to make a new slot read-only. In this case, writes to this memory will be
1361 posted to userspace as KVM_EXIT_MMIO exits.
1363 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1364 the memory region are automatically reflected into the guest. For example, an
1365 mmap() that affects the region will be made visible immediately. Another
1366 example is madvise(MADV_DROP).
1368 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1369 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1370 allocation and is deprecated.
1373 4.36 KVM_SET_TSS_ADDR
1374 ---------------------
1376 :Capability: KVM_CAP_SET_TSS_ADDR
1379 :Parameters: unsigned long tss_address (in)
1380 :Returns: 0 on success, -1 on error
1382 This ioctl defines the physical address of a three-page region in the guest
1383 physical address space. The region must be within the first 4GB of the
1384 guest physical address space and must not conflict with any memory slot
1385 or any mmio address. The guest may malfunction if it accesses this memory
1388 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1389 because of a quirk in the virtualization implementation (see the internals
1390 documentation when it pops into existence).
1396 :Capability: KVM_CAP_ENABLE_CAP
1397 :Architectures: mips, ppc, s390
1399 :Parameters: struct kvm_enable_cap (in)
1400 :Returns: 0 on success; -1 on error
1402 :Capability: KVM_CAP_ENABLE_CAP_VM
1405 :Parameters: struct kvm_enable_cap (in)
1406 :Returns: 0 on success; -1 on error
1410 Not all extensions are enabled by default. Using this ioctl the application
1411 can enable an extension, making it available to the guest.
1413 On systems that do not support this ioctl, it always fails. On systems that
1414 do support it, it only works for extensions that are supported for enablement.
1416 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1421 struct kvm_enable_cap {
1425 The capability that is supposed to get enabled.
1431 A bitfield indicating future enhancements. Has to be 0 for now.
1437 Arguments for enabling a feature. If a feature needs initial values to
1438 function properly, this is the place to put them.
1445 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1446 for vm-wide capabilities.
1448 4.38 KVM_GET_MP_STATE
1449 ---------------------
1451 :Capability: KVM_CAP_MP_STATE
1452 :Architectures: x86, s390, arm, arm64, riscv
1454 :Parameters: struct kvm_mp_state (out)
1455 :Returns: 0 on success; -1 on error
1459 struct kvm_mp_state {
1463 Returns the vcpu's current "multiprocessing state" (though also valid on
1464 uniprocessor guests).
1466 Possible values are:
1468 ========================== ===============================================
1469 KVM_MP_STATE_RUNNABLE the vcpu is currently running
1470 [x86,arm/arm64,riscv]
1471 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1472 which has not yet received an INIT signal [x86]
1473 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1474 now ready for a SIPI [x86]
1475 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1476 is waiting for an interrupt [x86]
1477 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1478 accessible via KVM_GET_VCPU_EVENTS) [x86]
1479 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64,riscv]
1480 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1481 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1483 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1485 ========================== ===============================================
1487 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1488 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1489 these architectures.
1491 For arm/arm64/riscv:
1492 ^^^^^^^^^^^^^^^^^^^^
1494 The only states that are valid are KVM_MP_STATE_STOPPED and
1495 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1497 4.39 KVM_SET_MP_STATE
1498 ---------------------
1500 :Capability: KVM_CAP_MP_STATE
1501 :Architectures: x86, s390, arm, arm64, riscv
1503 :Parameters: struct kvm_mp_state (in)
1504 :Returns: 0 on success; -1 on error
1506 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1509 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1510 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1511 these architectures.
1513 For arm/arm64/riscv:
1514 ^^^^^^^^^^^^^^^^^^^^
1516 The only states that are valid are KVM_MP_STATE_STOPPED and
1517 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1519 4.40 KVM_SET_IDENTITY_MAP_ADDR
1520 ------------------------------
1522 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1525 :Parameters: unsigned long identity (in)
1526 :Returns: 0 on success, -1 on error
1528 This ioctl defines the physical address of a one-page region in the guest
1529 physical address space. The region must be within the first 4GB of the
1530 guest physical address space and must not conflict with any memory slot
1531 or any mmio address. The guest may malfunction if it accesses this memory
1534 Setting the address to 0 will result in resetting the address to its default
1537 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1538 because of a quirk in the virtualization implementation (see the internals
1539 documentation when it pops into existence).
1541 Fails if any VCPU has already been created.
1543 4.41 KVM_SET_BOOT_CPU_ID
1544 ------------------------
1546 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1549 :Parameters: unsigned long vcpu_id
1550 :Returns: 0 on success, -1 on error
1552 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1553 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1554 is vcpu 0. This ioctl has to be called before vcpu creation,
1555 otherwise it will return EBUSY error.
1561 :Capability: KVM_CAP_XSAVE
1564 :Parameters: struct kvm_xsave (out)
1565 :Returns: 0 on success, -1 on error
1575 This ioctl would copy current vcpu's xsave struct to the userspace.
1581 :Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2
1584 :Parameters: struct kvm_xsave (in)
1585 :Returns: 0 on success, -1 on error
1595 This ioctl would copy userspace's xsave struct to the kernel. It copies
1596 as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2),
1597 when invoked on the vm file descriptor. The size value returned by
1598 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
1599 Currently, it is only greater than 4096 if a dynamic feature has been
1600 enabled with ``arch_prctl()``, but this may change in the future.
1602 The offsets of the state save areas in struct kvm_xsave follow the
1603 contents of CPUID leaf 0xD on the host.
1609 :Capability: KVM_CAP_XCRS
1612 :Parameters: struct kvm_xcrs (out)
1613 :Returns: 0 on success, -1 on error
1626 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1630 This ioctl would copy current vcpu's xcrs to the userspace.
1636 :Capability: KVM_CAP_XCRS
1639 :Parameters: struct kvm_xcrs (in)
1640 :Returns: 0 on success, -1 on error
1653 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1657 This ioctl would set vcpu's xcr to the value userspace specified.
1660 4.46 KVM_GET_SUPPORTED_CPUID
1661 ----------------------------
1663 :Capability: KVM_CAP_EXT_CPUID
1666 :Parameters: struct kvm_cpuid2 (in/out)
1667 :Returns: 0 on success, -1 on error
1674 struct kvm_cpuid_entry2 entries[0];
1677 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1678 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1679 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1681 struct kvm_cpuid_entry2 {
1692 This ioctl returns x86 cpuid features which are supported by both the
1693 hardware and kvm in its default configuration. Userspace can use the
1694 information returned by this ioctl to construct cpuid information (for
1695 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1696 userspace capabilities, and with user requirements (for example, the
1697 user may wish to constrain cpuid to emulate older hardware, or for
1698 feature consistency across a cluster).
1700 Dynamically-enabled feature bits need to be requested with
1701 ``arch_prctl()`` before calling this ioctl. Feature bits that have not
1702 been requested are excluded from the result.
1704 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1705 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1706 its default configuration. If userspace enables such capabilities, it
1707 is responsible for modifying the results of this ioctl appropriately.
1709 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1710 with the 'nent' field indicating the number of entries in the variable-size
1711 array 'entries'. If the number of entries is too low to describe the cpu
1712 capabilities, an error (E2BIG) is returned. If the number is too high,
1713 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1714 number is just right, the 'nent' field is adjusted to the number of valid
1715 entries in the 'entries' array, which is then filled.
1717 The entries returned are the host cpuid as returned by the cpuid instruction,
1718 with unknown or unsupported features masked out. Some features (for example,
1719 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1720 emulate them efficiently. The fields in each entry are defined as follows:
1723 the eax value used to obtain the entry
1726 the ecx value used to obtain the entry (for entries that are
1730 an OR of zero or more of the following:
1732 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1733 if the index field is valid
1736 the values returned by the cpuid instruction for
1737 this function/index combination
1739 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1740 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1741 support. Instead it is reported via::
1743 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1745 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1746 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1749 4.47 KVM_PPC_GET_PVINFO
1750 -----------------------
1752 :Capability: KVM_CAP_PPC_GET_PVINFO
1755 :Parameters: struct kvm_ppc_pvinfo (out)
1756 :Returns: 0 on success, !0 on error
1760 struct kvm_ppc_pvinfo {
1766 This ioctl fetches PV specific information that need to be passed to the guest
1767 using the device tree or other means from vm context.
1769 The hcall array defines 4 instructions that make up a hypercall.
1771 If any additional field gets added to this structure later on, a bit for that
1772 additional piece of information will be set in the flags bitmap.
1774 The flags bitmap is defined as::
1776 /* the host supports the ePAPR idle hcall
1777 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1779 4.52 KVM_SET_GSI_ROUTING
1780 ------------------------
1782 :Capability: KVM_CAP_IRQ_ROUTING
1783 :Architectures: x86 s390 arm arm64
1785 :Parameters: struct kvm_irq_routing (in)
1786 :Returns: 0 on success, -1 on error
1788 Sets the GSI routing table entries, overwriting any previously set entries.
1790 On arm/arm64, GSI routing has the following limitation:
1792 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1796 struct kvm_irq_routing {
1799 struct kvm_irq_routing_entry entries[0];
1802 No flags are specified so far, the corresponding field must be set to zero.
1806 struct kvm_irq_routing_entry {
1812 struct kvm_irq_routing_irqchip irqchip;
1813 struct kvm_irq_routing_msi msi;
1814 struct kvm_irq_routing_s390_adapter adapter;
1815 struct kvm_irq_routing_hv_sint hv_sint;
1816 struct kvm_irq_routing_xen_evtchn xen_evtchn;
1821 /* gsi routing entry types */
1822 #define KVM_IRQ_ROUTING_IRQCHIP 1
1823 #define KVM_IRQ_ROUTING_MSI 2
1824 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1825 #define KVM_IRQ_ROUTING_HV_SINT 4
1826 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5
1830 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1831 type, specifies that the devid field contains a valid value. The per-VM
1832 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1833 the device ID. If this capability is not available, userspace should
1834 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1839 struct kvm_irq_routing_irqchip {
1844 struct kvm_irq_routing_msi {
1854 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1855 for the device that wrote the MSI message. For PCI, this is usually a
1856 BFD identifier in the lower 16 bits.
1858 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1859 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1860 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1861 address_hi must be zero.
1865 struct kvm_irq_routing_s390_adapter {
1869 __u32 summary_offset;
1873 struct kvm_irq_routing_hv_sint {
1878 struct kvm_irq_routing_xen_evtchn {
1885 When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit
1886 in its indication of supported features, routing to Xen event channels
1887 is supported. Although the priority field is present, only the value
1888 KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by
1889 2 level event channels. FIFO event channel support may be added in
1893 4.55 KVM_SET_TSC_KHZ
1894 --------------------
1896 :Capability: KVM_CAP_TSC_CONTROL
1899 :Parameters: virtual tsc_khz
1900 :Returns: 0 on success, -1 on error
1902 Specifies the tsc frequency for the virtual machine. The unit of the
1906 4.56 KVM_GET_TSC_KHZ
1907 --------------------
1909 :Capability: KVM_CAP_GET_TSC_KHZ
1913 :Returns: virtual tsc-khz on success, negative value on error
1915 Returns the tsc frequency of the guest. The unit of the return value is
1916 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1923 :Capability: KVM_CAP_IRQCHIP
1926 :Parameters: struct kvm_lapic_state (out)
1927 :Returns: 0 on success, -1 on error
1931 #define KVM_APIC_REG_SIZE 0x400
1932 struct kvm_lapic_state {
1933 char regs[KVM_APIC_REG_SIZE];
1936 Reads the Local APIC registers and copies them into the input argument. The
1937 data format and layout are the same as documented in the architecture manual.
1939 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1940 enabled, then the format of APIC_ID register depends on the APIC mode
1941 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1942 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1943 which is stored in bits 31-24 of the APIC register, or equivalently in
1944 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1945 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1947 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1948 always uses xAPIC format.
1954 :Capability: KVM_CAP_IRQCHIP
1957 :Parameters: struct kvm_lapic_state (in)
1958 :Returns: 0 on success, -1 on error
1962 #define KVM_APIC_REG_SIZE 0x400
1963 struct kvm_lapic_state {
1964 char regs[KVM_APIC_REG_SIZE];
1967 Copies the input argument into the Local APIC registers. The data format
1968 and layout are the same as documented in the architecture manual.
1970 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1971 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1972 See the note in KVM_GET_LAPIC.
1978 :Capability: KVM_CAP_IOEVENTFD
1981 :Parameters: struct kvm_ioeventfd (in)
1982 :Returns: 0 on success, !0 on error
1984 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1985 within the guest. A guest write in the registered address will signal the
1986 provided event instead of triggering an exit.
1990 struct kvm_ioeventfd {
1992 __u64 addr; /* legal pio/mmio address */
1993 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1999 For the special case of virtio-ccw devices on s390, the ioevent is matched
2000 to a subchannel/virtqueue tuple instead.
2002 The following flags are defined::
2004 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
2005 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
2006 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
2007 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2008 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2010 If datamatch flag is set, the event will be signaled only if the written value
2011 to the registered address is equal to datamatch in struct kvm_ioeventfd.
2013 For virtio-ccw devices, addr contains the subchannel id and datamatch the
2016 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2017 the kernel will ignore the length of guest write and may get a faster vmexit.
2018 The speedup may only apply to specific architectures, but the ioeventfd will
2024 :Capability: KVM_CAP_SW_TLB
2027 :Parameters: struct kvm_dirty_tlb (in)
2028 :Returns: 0 on success, -1 on error
2032 struct kvm_dirty_tlb {
2037 This must be called whenever userspace has changed an entry in the shared
2038 TLB, prior to calling KVM_RUN on the associated vcpu.
2040 The "bitmap" field is the userspace address of an array. This array
2041 consists of a number of bits, equal to the total number of TLB entries as
2042 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
2043 nearest multiple of 64.
2045 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2048 The array is little-endian: the bit 0 is the least significant bit of the
2049 first byte, bit 8 is the least significant bit of the second byte, etc.
2050 This avoids any complications with differing word sizes.
2052 The "num_dirty" field is a performance hint for KVM to determine whether it
2053 should skip processing the bitmap and just invalidate everything. It must
2054 be set to the number of set bits in the bitmap.
2057 4.62 KVM_CREATE_SPAPR_TCE
2058 -------------------------
2060 :Capability: KVM_CAP_SPAPR_TCE
2061 :Architectures: powerpc
2063 :Parameters: struct kvm_create_spapr_tce (in)
2064 :Returns: file descriptor for manipulating the created TCE table
2066 This creates a virtual TCE (translation control entry) table, which
2067 is an IOMMU for PAPR-style virtual I/O. It is used to translate
2068 logical addresses used in virtual I/O into guest physical addresses,
2069 and provides a scatter/gather capability for PAPR virtual I/O.
2073 /* for KVM_CAP_SPAPR_TCE */
2074 struct kvm_create_spapr_tce {
2079 The liobn field gives the logical IO bus number for which to create a
2080 TCE table. The window_size field specifies the size of the DMA window
2081 which this TCE table will translate - the table will contain one 64
2082 bit TCE entry for every 4kiB of the DMA window.
2084 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2085 table has been created using this ioctl(), the kernel will handle it
2086 in real mode, updating the TCE table. H_PUT_TCE calls for other
2087 liobns will cause a vm exit and must be handled by userspace.
2089 The return value is a file descriptor which can be passed to mmap(2)
2090 to map the created TCE table into userspace. This lets userspace read
2091 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2092 userspace update the TCE table directly which is useful in some
2096 4.63 KVM_ALLOCATE_RMA
2097 ---------------------
2099 :Capability: KVM_CAP_PPC_RMA
2100 :Architectures: powerpc
2102 :Parameters: struct kvm_allocate_rma (out)
2103 :Returns: file descriptor for mapping the allocated RMA
2105 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2106 time by the kernel. An RMA is a physically-contiguous, aligned region
2107 of memory used on older POWER processors to provide the memory which
2108 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2109 POWER processors support a set of sizes for the RMA that usually
2110 includes 64MB, 128MB, 256MB and some larger powers of two.
2114 /* for KVM_ALLOCATE_RMA */
2115 struct kvm_allocate_rma {
2119 The return value is a file descriptor which can be passed to mmap(2)
2120 to map the allocated RMA into userspace. The mapped area can then be
2121 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2122 RMA for a virtual machine. The size of the RMA in bytes (which is
2123 fixed at host kernel boot time) is returned in the rma_size field of
2124 the argument structure.
2126 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2127 is supported; 2 if the processor requires all virtual machines to have
2128 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2129 because it supports the Virtual RMA (VRMA) facility.
2135 :Capability: KVM_CAP_USER_NMI
2139 :Returns: 0 on success, -1 on error
2141 Queues an NMI on the thread's vcpu. Note this is well defined only
2142 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2143 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2144 has been called, this interface is completely emulated within the kernel.
2146 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2147 following algorithm:
2150 - read the local APIC's state (KVM_GET_LAPIC)
2151 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2152 - if so, issue KVM_NMI
2155 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2159 4.65 KVM_S390_UCAS_MAP
2160 ----------------------
2162 :Capability: KVM_CAP_S390_UCONTROL
2163 :Architectures: s390
2165 :Parameters: struct kvm_s390_ucas_mapping (in)
2166 :Returns: 0 in case of success
2168 The parameter is defined like this::
2170 struct kvm_s390_ucas_mapping {
2176 This ioctl maps the memory at "user_addr" with the length "length" to
2177 the vcpu's address space starting at "vcpu_addr". All parameters need to
2178 be aligned by 1 megabyte.
2181 4.66 KVM_S390_UCAS_UNMAP
2182 ------------------------
2184 :Capability: KVM_CAP_S390_UCONTROL
2185 :Architectures: s390
2187 :Parameters: struct kvm_s390_ucas_mapping (in)
2188 :Returns: 0 in case of success
2190 The parameter is defined like this::
2192 struct kvm_s390_ucas_mapping {
2198 This ioctl unmaps the memory in the vcpu's address space starting at
2199 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2200 All parameters need to be aligned by 1 megabyte.
2203 4.67 KVM_S390_VCPU_FAULT
2204 ------------------------
2206 :Capability: KVM_CAP_S390_UCONTROL
2207 :Architectures: s390
2209 :Parameters: vcpu absolute address (in)
2210 :Returns: 0 in case of success
2212 This call creates a page table entry on the virtual cpu's address space
2213 (for user controlled virtual machines) or the virtual machine's address
2214 space (for regular virtual machines). This only works for minor faults,
2215 thus it's recommended to access subject memory page via the user page
2216 table upfront. This is useful to handle validity intercepts for user
2217 controlled virtual machines to fault in the virtual cpu's lowcore pages
2218 prior to calling the KVM_RUN ioctl.
2221 4.68 KVM_SET_ONE_REG
2222 --------------------
2224 :Capability: KVM_CAP_ONE_REG
2227 :Parameters: struct kvm_one_reg (in)
2228 :Returns: 0 on success, negative value on failure
2232 ====== ============================================================
2233 ENOENT no such register
2234 EINVAL invalid register ID, or no such register or used with VMs in
2235 protected virtualization mode on s390
2236 EPERM (arm64) register access not allowed before vcpu finalization
2237 ====== ============================================================
2239 (These error codes are indicative only: do not rely on a specific error
2240 code being returned in a specific situation.)
2244 struct kvm_one_reg {
2249 Using this ioctl, a single vcpu register can be set to a specific value
2250 defined by user space with the passed in struct kvm_one_reg, where id
2251 refers to the register identifier as described below and addr is a pointer
2252 to a variable with the respective size. There can be architecture agnostic
2253 and architecture specific registers. Each have their own range of operation
2254 and their own constants and width. To keep track of the implemented
2255 registers, find a list below:
2257 ======= =============================== ============
2258 Arch Register Width (bits)
2259 ======= =============================== ============
2260 PPC KVM_REG_PPC_HIOR 64
2261 PPC KVM_REG_PPC_IAC1 64
2262 PPC KVM_REG_PPC_IAC2 64
2263 PPC KVM_REG_PPC_IAC3 64
2264 PPC KVM_REG_PPC_IAC4 64
2265 PPC KVM_REG_PPC_DAC1 64
2266 PPC KVM_REG_PPC_DAC2 64
2267 PPC KVM_REG_PPC_DABR 64
2268 PPC KVM_REG_PPC_DSCR 64
2269 PPC KVM_REG_PPC_PURR 64
2270 PPC KVM_REG_PPC_SPURR 64
2271 PPC KVM_REG_PPC_DAR 64
2272 PPC KVM_REG_PPC_DSISR 32
2273 PPC KVM_REG_PPC_AMR 64
2274 PPC KVM_REG_PPC_UAMOR 64
2275 PPC KVM_REG_PPC_MMCR0 64
2276 PPC KVM_REG_PPC_MMCR1 64
2277 PPC KVM_REG_PPC_MMCRA 64
2278 PPC KVM_REG_PPC_MMCR2 64
2279 PPC KVM_REG_PPC_MMCRS 64
2280 PPC KVM_REG_PPC_MMCR3 64
2281 PPC KVM_REG_PPC_SIAR 64
2282 PPC KVM_REG_PPC_SDAR 64
2283 PPC KVM_REG_PPC_SIER 64
2284 PPC KVM_REG_PPC_SIER2 64
2285 PPC KVM_REG_PPC_SIER3 64
2286 PPC KVM_REG_PPC_PMC1 32
2287 PPC KVM_REG_PPC_PMC2 32
2288 PPC KVM_REG_PPC_PMC3 32
2289 PPC KVM_REG_PPC_PMC4 32
2290 PPC KVM_REG_PPC_PMC5 32
2291 PPC KVM_REG_PPC_PMC6 32
2292 PPC KVM_REG_PPC_PMC7 32
2293 PPC KVM_REG_PPC_PMC8 32
2294 PPC KVM_REG_PPC_FPR0 64
2296 PPC KVM_REG_PPC_FPR31 64
2297 PPC KVM_REG_PPC_VR0 128
2299 PPC KVM_REG_PPC_VR31 128
2300 PPC KVM_REG_PPC_VSR0 128
2302 PPC KVM_REG_PPC_VSR31 128
2303 PPC KVM_REG_PPC_FPSCR 64
2304 PPC KVM_REG_PPC_VSCR 32
2305 PPC KVM_REG_PPC_VPA_ADDR 64
2306 PPC KVM_REG_PPC_VPA_SLB 128
2307 PPC KVM_REG_PPC_VPA_DTL 128
2308 PPC KVM_REG_PPC_EPCR 32
2309 PPC KVM_REG_PPC_EPR 32
2310 PPC KVM_REG_PPC_TCR 32
2311 PPC KVM_REG_PPC_TSR 32
2312 PPC KVM_REG_PPC_OR_TSR 32
2313 PPC KVM_REG_PPC_CLEAR_TSR 32
2314 PPC KVM_REG_PPC_MAS0 32
2315 PPC KVM_REG_PPC_MAS1 32
2316 PPC KVM_REG_PPC_MAS2 64
2317 PPC KVM_REG_PPC_MAS7_3 64
2318 PPC KVM_REG_PPC_MAS4 32
2319 PPC KVM_REG_PPC_MAS6 32
2320 PPC KVM_REG_PPC_MMUCFG 32
2321 PPC KVM_REG_PPC_TLB0CFG 32
2322 PPC KVM_REG_PPC_TLB1CFG 32
2323 PPC KVM_REG_PPC_TLB2CFG 32
2324 PPC KVM_REG_PPC_TLB3CFG 32
2325 PPC KVM_REG_PPC_TLB0PS 32
2326 PPC KVM_REG_PPC_TLB1PS 32
2327 PPC KVM_REG_PPC_TLB2PS 32
2328 PPC KVM_REG_PPC_TLB3PS 32
2329 PPC KVM_REG_PPC_EPTCFG 32
2330 PPC KVM_REG_PPC_ICP_STATE 64
2331 PPC KVM_REG_PPC_VP_STATE 128
2332 PPC KVM_REG_PPC_TB_OFFSET 64
2333 PPC KVM_REG_PPC_SPMC1 32
2334 PPC KVM_REG_PPC_SPMC2 32
2335 PPC KVM_REG_PPC_IAMR 64
2336 PPC KVM_REG_PPC_TFHAR 64
2337 PPC KVM_REG_PPC_TFIAR 64
2338 PPC KVM_REG_PPC_TEXASR 64
2339 PPC KVM_REG_PPC_FSCR 64
2340 PPC KVM_REG_PPC_PSPB 32
2341 PPC KVM_REG_PPC_EBBHR 64
2342 PPC KVM_REG_PPC_EBBRR 64
2343 PPC KVM_REG_PPC_BESCR 64
2344 PPC KVM_REG_PPC_TAR 64
2345 PPC KVM_REG_PPC_DPDES 64
2346 PPC KVM_REG_PPC_DAWR 64
2347 PPC KVM_REG_PPC_DAWRX 64
2348 PPC KVM_REG_PPC_CIABR 64
2349 PPC KVM_REG_PPC_IC 64
2350 PPC KVM_REG_PPC_VTB 64
2351 PPC KVM_REG_PPC_CSIGR 64
2352 PPC KVM_REG_PPC_TACR 64
2353 PPC KVM_REG_PPC_TCSCR 64
2354 PPC KVM_REG_PPC_PID 64
2355 PPC KVM_REG_PPC_ACOP 64
2356 PPC KVM_REG_PPC_VRSAVE 32
2357 PPC KVM_REG_PPC_LPCR 32
2358 PPC KVM_REG_PPC_LPCR_64 64
2359 PPC KVM_REG_PPC_PPR 64
2360 PPC KVM_REG_PPC_ARCH_COMPAT 32
2361 PPC KVM_REG_PPC_DABRX 32
2362 PPC KVM_REG_PPC_WORT 64
2363 PPC KVM_REG_PPC_SPRG9 64
2364 PPC KVM_REG_PPC_DBSR 32
2365 PPC KVM_REG_PPC_TIDR 64
2366 PPC KVM_REG_PPC_PSSCR 64
2367 PPC KVM_REG_PPC_DEC_EXPIRY 64
2368 PPC KVM_REG_PPC_PTCR 64
2369 PPC KVM_REG_PPC_DAWR1 64
2370 PPC KVM_REG_PPC_DAWRX1 64
2371 PPC KVM_REG_PPC_TM_GPR0 64
2373 PPC KVM_REG_PPC_TM_GPR31 64
2374 PPC KVM_REG_PPC_TM_VSR0 128
2376 PPC KVM_REG_PPC_TM_VSR63 128
2377 PPC KVM_REG_PPC_TM_CR 64
2378 PPC KVM_REG_PPC_TM_LR 64
2379 PPC KVM_REG_PPC_TM_CTR 64
2380 PPC KVM_REG_PPC_TM_FPSCR 64
2381 PPC KVM_REG_PPC_TM_AMR 64
2382 PPC KVM_REG_PPC_TM_PPR 64
2383 PPC KVM_REG_PPC_TM_VRSAVE 64
2384 PPC KVM_REG_PPC_TM_VSCR 32
2385 PPC KVM_REG_PPC_TM_DSCR 64
2386 PPC KVM_REG_PPC_TM_TAR 64
2387 PPC KVM_REG_PPC_TM_XER 64
2389 MIPS KVM_REG_MIPS_R0 64
2391 MIPS KVM_REG_MIPS_R31 64
2392 MIPS KVM_REG_MIPS_HI 64
2393 MIPS KVM_REG_MIPS_LO 64
2394 MIPS KVM_REG_MIPS_PC 64
2395 MIPS KVM_REG_MIPS_CP0_INDEX 32
2396 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2397 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2398 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2399 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2400 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2401 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2402 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2403 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2404 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2405 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2406 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2407 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2408 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2409 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2410 MIPS KVM_REG_MIPS_CP0_WIRED 32
2411 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2412 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2413 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2414 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2415 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2416 MIPS KVM_REG_MIPS_CP0_COUNT 32
2417 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2418 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2419 MIPS KVM_REG_MIPS_CP0_STATUS 32
2420 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2421 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2422 MIPS KVM_REG_MIPS_CP0_EPC 64
2423 MIPS KVM_REG_MIPS_CP0_PRID 32
2424 MIPS KVM_REG_MIPS_CP0_EBASE 64
2425 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2426 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2427 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2428 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2429 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2430 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2431 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2432 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2433 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2434 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2435 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2436 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2437 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2438 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2439 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2440 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2441 MIPS KVM_REG_MIPS_COUNT_CTL 64
2442 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2443 MIPS KVM_REG_MIPS_COUNT_HZ 64
2444 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2445 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2446 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2447 MIPS KVM_REG_MIPS_FCR_IR 32
2448 MIPS KVM_REG_MIPS_FCR_CSR 32
2449 MIPS KVM_REG_MIPS_MSA_IR 32
2450 MIPS KVM_REG_MIPS_MSA_CSR 32
2451 ======= =============================== ============
2453 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2454 is the register group type, or coprocessor number:
2456 ARM core registers have the following id bit patterns::
2458 0x4020 0000 0010 <index into the kvm_regs struct:16>
2460 ARM 32-bit CP15 registers have the following id bit patterns::
2462 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2464 ARM 64-bit CP15 registers have the following id bit patterns::
2466 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2468 ARM CCSIDR registers are demultiplexed by CSSELR value::
2470 0x4020 0000 0011 00 <csselr:8>
2472 ARM 32-bit VFP control registers have the following id bit patterns::
2474 0x4020 0000 0012 1 <regno:12>
2476 ARM 64-bit FP registers have the following id bit patterns::
2478 0x4030 0000 0012 0 <regno:12>
2480 ARM firmware pseudo-registers have the following bit pattern::
2482 0x4030 0000 0014 <regno:16>
2485 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2486 that is the register group type, or coprocessor number:
2488 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2489 that the size of the access is variable, as the kvm_regs structure
2490 contains elements ranging from 32 to 128 bits. The index is a 32bit
2491 value in the kvm_regs structure seen as a 32bit array::
2493 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2497 ======================= ========= ===== =======================================
2498 Encoding Register Bits kvm_regs member
2499 ======================= ========= ===== =======================================
2500 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2501 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2503 0x6030 0000 0010 003c X30 64 regs.regs[30]
2504 0x6030 0000 0010 003e SP 64 regs.sp
2505 0x6030 0000 0010 0040 PC 64 regs.pc
2506 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2507 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2508 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2509 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2510 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2511 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2512 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2513 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2514 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2515 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2517 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2518 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2519 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2520 ======================= ========= ===== =======================================
2522 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2525 The equivalent register content can be accessed via bits [127:0] of
2526 the corresponding SVE Zn registers instead for vcpus that have SVE
2527 enabled (see below).
2529 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2531 0x6020 0000 0011 00 <csselr:8>
2533 arm64 system registers have the following id bit patterns::
2535 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2539 Two system register IDs do not follow the specified pattern. These
2540 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2541 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2542 two had their values accidentally swapped, which means TIMER_CVAL is
2543 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2544 derived from the register encoding for CNTV_CVAL_EL0. As this is
2545 API, it must remain this way.
2547 arm64 firmware pseudo-registers have the following bit pattern::
2549 0x6030 0000 0014 <regno:16>
2551 arm64 SVE registers have the following bit patterns::
2553 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2554 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2555 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2556 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2558 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2559 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2560 quadwords: see [2]_ below.
2562 These registers are only accessible on vcpus for which SVE is enabled.
2563 See KVM_ARM_VCPU_INIT for details.
2565 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2566 accessible until the vcpu's SVE configuration has been finalized
2567 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2568 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2570 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2571 lengths supported by the vcpu to be discovered and configured by
2572 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2573 or KVM_SET_ONE_REG, the value of this register is of type
2574 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2577 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2579 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2580 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2581 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2582 /* Vector length vq * 16 bytes supported */
2584 /* Vector length vq * 16 bytes not supported */
2586 .. [2] The maximum value vq for which the above condition is true is
2587 max_vq. This is the maximum vector length available to the guest on
2588 this vcpu, and determines which register slices are visible through
2589 this ioctl interface.
2591 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2594 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2595 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2598 Userspace may subsequently modify it if desired until the vcpu's SVE
2599 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2601 Apart from simply removing all vector lengths from the host set that
2602 exceed some value, support for arbitrarily chosen sets of vector lengths
2603 is hardware-dependent and may not be available. Attempting to configure
2604 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2607 After the vcpu's SVE configuration is finalized, further attempts to
2608 write this register will fail with EPERM.
2611 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2612 the register group type:
2614 MIPS core registers (see above) have the following id bit patterns::
2616 0x7030 0000 0000 <reg:16>
2618 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2619 patterns depending on whether they're 32-bit or 64-bit registers::
2621 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2622 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2624 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2625 versions of the EntryLo registers regardless of the word size of the host
2626 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2627 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2628 the PFNX field starting at bit 30.
2630 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2633 0x7030 0000 0001 01 <reg:8>
2635 MIPS KVM control registers (see above) have the following id bit patterns::
2637 0x7030 0000 0002 <reg:16>
2639 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2640 id bit patterns depending on the size of the register being accessed. They are
2641 always accessed according to the current guest FPU mode (Status.FR and
2642 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2643 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2644 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2645 overlap the FPU registers::
2647 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2648 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2649 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2651 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2652 following id bit patterns::
2654 0x7020 0000 0003 01 <0:3> <reg:5>
2656 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2657 following id bit patterns::
2659 0x7020 0000 0003 02 <0:3> <reg:5>
2661 RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of
2662 that is the register group type.
2664 RISC-V config registers are meant for configuring a Guest VCPU and it has
2665 the following id bit patterns::
2667 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host)
2668 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host)
2670 Following are the RISC-V config registers:
2672 ======================= ========= =============================================
2673 Encoding Register Description
2674 ======================= ========= =============================================
2675 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU
2676 ======================= ========= =============================================
2678 The isa config register can be read anytime but can only be written before
2679 a Guest VCPU runs. It will have ISA feature bits matching underlying host
2682 RISC-V core registers represent the general excution state of a Guest VCPU
2683 and it has the following id bit patterns::
2685 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host)
2686 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host)
2688 Following are the RISC-V core registers:
2690 ======================= ========= =============================================
2691 Encoding Register Description
2692 ======================= ========= =============================================
2693 0x80x0 0000 0200 0000 regs.pc Program counter
2694 0x80x0 0000 0200 0001 regs.ra Return address
2695 0x80x0 0000 0200 0002 regs.sp Stack pointer
2696 0x80x0 0000 0200 0003 regs.gp Global pointer
2697 0x80x0 0000 0200 0004 regs.tp Task pointer
2698 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0
2699 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1
2700 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2
2701 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0
2702 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1
2703 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0
2704 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1
2705 0x80x0 0000 0200 000c regs.a2 Function argument 2
2706 0x80x0 0000 0200 000d regs.a3 Function argument 3
2707 0x80x0 0000 0200 000e regs.a4 Function argument 4
2708 0x80x0 0000 0200 000f regs.a5 Function argument 5
2709 0x80x0 0000 0200 0010 regs.a6 Function argument 6
2710 0x80x0 0000 0200 0011 regs.a7 Function argument 7
2711 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2
2712 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3
2713 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4
2714 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5
2715 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6
2716 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7
2717 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8
2718 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9
2719 0x80x0 0000 0200 001a regs.s10 Callee saved register 10
2720 0x80x0 0000 0200 001b regs.s11 Callee saved register 11
2721 0x80x0 0000 0200 001c regs.t3 Caller saved register 3
2722 0x80x0 0000 0200 001d regs.t4 Caller saved register 4
2723 0x80x0 0000 0200 001e regs.t5 Caller saved register 5
2724 0x80x0 0000 0200 001f regs.t6 Caller saved register 6
2725 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode)
2726 ======================= ========= =============================================
2728 RISC-V csr registers represent the supervisor mode control/status registers
2729 of a Guest VCPU and it has the following id bit patterns::
2731 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host)
2732 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host)
2734 Following are the RISC-V csr registers:
2736 ======================= ========= =============================================
2737 Encoding Register Description
2738 ======================= ========= =============================================
2739 0x80x0 0000 0300 0000 sstatus Supervisor status
2740 0x80x0 0000 0300 0001 sie Supervisor interrupt enable
2741 0x80x0 0000 0300 0002 stvec Supervisor trap vector base
2742 0x80x0 0000 0300 0003 sscratch Supervisor scratch register
2743 0x80x0 0000 0300 0004 sepc Supervisor exception program counter
2744 0x80x0 0000 0300 0005 scause Supervisor trap cause
2745 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction
2746 0x80x0 0000 0300 0007 sip Supervisor interrupt pending
2747 0x80x0 0000 0300 0008 satp Supervisor address translation and protection
2748 ======================= ========= =============================================
2750 RISC-V timer registers represent the timer state of a Guest VCPU and it has
2751 the following id bit patterns::
2753 0x8030 0000 04 <index into the kvm_riscv_timer struct:24>
2755 Following are the RISC-V timer registers:
2757 ======================= ========= =============================================
2758 Encoding Register Description
2759 ======================= ========= =============================================
2760 0x8030 0000 0400 0000 frequency Time base frequency (read-only)
2761 0x8030 0000 0400 0001 time Time value visible to Guest
2762 0x8030 0000 0400 0002 compare Time compare programmed by Guest
2763 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF)
2764 ======================= ========= =============================================
2766 RISC-V F-extension registers represent the single precision floating point
2767 state of a Guest VCPU and it has the following id bit patterns::
2769 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24>
2771 Following are the RISC-V F-extension registers:
2773 ======================= ========= =============================================
2774 Encoding Register Description
2775 ======================= ========= =============================================
2776 0x8020 0000 0500 0000 f[0] Floating point register 0
2778 0x8020 0000 0500 001f f[31] Floating point register 31
2779 0x8020 0000 0500 0020 fcsr Floating point control and status register
2780 ======================= ========= =============================================
2782 RISC-V D-extension registers represent the double precision floating point
2783 state of a Guest VCPU and it has the following id bit patterns::
2785 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr)
2786 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr)
2788 Following are the RISC-V D-extension registers:
2790 ======================= ========= =============================================
2791 Encoding Register Description
2792 ======================= ========= =============================================
2793 0x8030 0000 0600 0000 f[0] Floating point register 0
2795 0x8030 0000 0600 001f f[31] Floating point register 31
2796 0x8020 0000 0600 0020 fcsr Floating point control and status register
2797 ======================= ========= =============================================
2800 4.69 KVM_GET_ONE_REG
2801 --------------------
2803 :Capability: KVM_CAP_ONE_REG
2806 :Parameters: struct kvm_one_reg (in and out)
2807 :Returns: 0 on success, negative value on failure
2811 ======== ============================================================
2812 ENOENT no such register
2813 EINVAL invalid register ID, or no such register or used with VMs in
2814 protected virtualization mode on s390
2815 EPERM (arm64) register access not allowed before vcpu finalization
2816 ======== ============================================================
2818 (These error codes are indicative only: do not rely on a specific error
2819 code being returned in a specific situation.)
2821 This ioctl allows to receive the value of a single register implemented
2822 in a vcpu. The register to read is indicated by the "id" field of the
2823 kvm_one_reg struct passed in. On success, the register value can be found
2824 at the memory location pointed to by "addr".
2826 The list of registers accessible using this interface is identical to the
2830 4.70 KVM_KVMCLOCK_CTRL
2831 ----------------------
2833 :Capability: KVM_CAP_KVMCLOCK_CTRL
2834 :Architectures: Any that implement pvclocks (currently x86 only)
2837 :Returns: 0 on success, -1 on error
2839 This ioctl sets a flag accessible to the guest indicating that the specified
2840 vCPU has been paused by the host userspace.
2842 The host will set a flag in the pvclock structure that is checked from the
2843 soft lockup watchdog. The flag is part of the pvclock structure that is
2844 shared between guest and host, specifically the second bit of the flags
2845 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2846 the host and read/cleared exclusively by the guest. The guest operation of
2847 checking and clearing the flag must be an atomic operation so
2848 load-link/store-conditional, or equivalent must be used. There are two cases
2849 where the guest will clear the flag: when the soft lockup watchdog timer resets
2850 itself or when a soft lockup is detected. This ioctl can be called any time
2851 after pausing the vcpu, but before it is resumed.
2857 :Capability: KVM_CAP_SIGNAL_MSI
2858 :Architectures: x86 arm arm64
2860 :Parameters: struct kvm_msi (in)
2861 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2863 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2878 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2879 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2880 the device ID. If this capability is not available, userspace
2881 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2883 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2884 for the device that wrote the MSI message. For PCI, this is usually a
2885 BFD identifier in the lower 16 bits.
2887 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2888 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2889 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2890 address_hi must be zero.
2893 4.71 KVM_CREATE_PIT2
2894 --------------------
2896 :Capability: KVM_CAP_PIT2
2899 :Parameters: struct kvm_pit_config (in)
2900 :Returns: 0 on success, -1 on error
2902 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2903 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2904 parameters have to be passed::
2906 struct kvm_pit_config {
2913 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2915 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2916 exists, this thread will have a name of the following pattern::
2918 kvm-pit/<owner-process-pid>
2920 When running a guest with elevated priorities, the scheduling parameters of
2921 this thread may have to be adjusted accordingly.
2923 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2929 :Capability: KVM_CAP_PIT_STATE2
2932 :Parameters: struct kvm_pit_state2 (out)
2933 :Returns: 0 on success, -1 on error
2935 Retrieves the state of the in-kernel PIT model. Only valid after
2936 KVM_CREATE_PIT2. The state is returned in the following structure::
2938 struct kvm_pit_state2 {
2939 struct kvm_pit_channel_state channels[3];
2946 /* disable PIT in HPET legacy mode */
2947 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2949 This IOCTL replaces the obsolete KVM_GET_PIT.
2955 :Capability: KVM_CAP_PIT_STATE2
2958 :Parameters: struct kvm_pit_state2 (in)
2959 :Returns: 0 on success, -1 on error
2961 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2962 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2964 This IOCTL replaces the obsolete KVM_SET_PIT.
2967 4.74 KVM_PPC_GET_SMMU_INFO
2968 --------------------------
2970 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2971 :Architectures: powerpc
2974 :Returns: 0 on success, -1 on error
2976 This populates and returns a structure describing the features of
2977 the "Server" class MMU emulation supported by KVM.
2978 This can in turn be used by userspace to generate the appropriate
2979 device-tree properties for the guest operating system.
2981 The structure contains some global information, followed by an
2982 array of supported segment page sizes::
2984 struct kvm_ppc_smmu_info {
2988 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2991 The supported flags are:
2993 - KVM_PPC_PAGE_SIZES_REAL:
2994 When that flag is set, guest page sizes must "fit" the backing
2995 store page sizes. When not set, any page size in the list can
2996 be used regardless of how they are backed by userspace.
2998 - KVM_PPC_1T_SEGMENTS
2999 The emulated MMU supports 1T segments in addition to the
3003 This flag indicates that HPT guests are not supported by KVM,
3004 thus all guests must use radix MMU mode.
3006 The "slb_size" field indicates how many SLB entries are supported
3008 The "sps" array contains 8 entries indicating the supported base
3009 page sizes for a segment in increasing order. Each entry is defined
3012 struct kvm_ppc_one_seg_page_size {
3013 __u32 page_shift; /* Base page shift of segment (or 0) */
3014 __u32 slb_enc; /* SLB encoding for BookS */
3015 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3018 An entry with a "page_shift" of 0 is unused. Because the array is
3019 organized in increasing order, a lookup can stop when encoutering
3022 The "slb_enc" field provides the encoding to use in the SLB for the
3023 page size. The bits are in positions such as the value can directly
3024 be OR'ed into the "vsid" argument of the slbmte instruction.
3026 The "enc" array is a list which for each of those segment base page
3027 size provides the list of supported actual page sizes (which can be
3028 only larger or equal to the base page size), along with the
3029 corresponding encoding in the hash PTE. Similarly, the array is
3030 8 entries sorted by increasing sizes and an entry with a "0" shift
3031 is an empty entry and a terminator::
3033 struct kvm_ppc_one_page_size {
3034 __u32 page_shift; /* Page shift (or 0) */
3035 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
3038 The "pte_enc" field provides a value that can OR'ed into the hash
3039 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3040 into the hash PTE second double word).
3045 :Capability: KVM_CAP_IRQFD
3046 :Architectures: x86 s390 arm arm64
3048 :Parameters: struct kvm_irqfd (in)
3049 :Returns: 0 on success, -1 on error
3051 Allows setting an eventfd to directly trigger a guest interrupt.
3052 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3053 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
3054 an event is triggered on the eventfd, an interrupt is injected into
3055 the guest using the specified gsi pin. The irqfd is removed using
3056 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3059 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3060 mechanism allowing emulation of level-triggered, irqfd-based
3061 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3062 additional eventfd in the kvm_irqfd.resamplefd field. When operating
3063 in resample mode, posting of an interrupt through kvm_irq.fd asserts
3064 the specified gsi in the irqchip. When the irqchip is resampled, such
3065 as from an EOI, the gsi is de-asserted and the user is notified via
3066 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
3067 the interrupt if the device making use of it still requires service.
3068 Note that closing the resamplefd is not sufficient to disable the
3069 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3070 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3072 On arm/arm64, gsi routing being supported, the following can happen:
3074 - in case no routing entry is associated to this gsi, injection fails
3075 - in case the gsi is associated to an irqchip routing entry,
3076 irqchip.pin + 32 corresponds to the injected SPI ID.
3077 - in case the gsi is associated to an MSI routing entry, the MSI
3078 message and device ID are translated into an LPI (support restricted
3079 to GICv3 ITS in-kernel emulation).
3081 4.76 KVM_PPC_ALLOCATE_HTAB
3082 --------------------------
3084 :Capability: KVM_CAP_PPC_ALLOC_HTAB
3085 :Architectures: powerpc
3087 :Parameters: Pointer to u32 containing hash table order (in/out)
3088 :Returns: 0 on success, -1 on error
3090 This requests the host kernel to allocate an MMU hash table for a
3091 guest using the PAPR paravirtualization interface. This only does
3092 anything if the kernel is configured to use the Book 3S HV style of
3093 virtualization. Otherwise the capability doesn't exist and the ioctl
3094 returns an ENOTTY error. The rest of this description assumes Book 3S
3097 There must be no vcpus running when this ioctl is called; if there
3098 are, it will do nothing and return an EBUSY error.
3100 The parameter is a pointer to a 32-bit unsigned integer variable
3101 containing the order (log base 2) of the desired size of the hash
3102 table, which must be between 18 and 46. On successful return from the
3103 ioctl, the value will not be changed by the kernel.
3105 If no hash table has been allocated when any vcpu is asked to run
3106 (with the KVM_RUN ioctl), the host kernel will allocate a
3107 default-sized hash table (16 MB).
3109 If this ioctl is called when a hash table has already been allocated,
3110 with a different order from the existing hash table, the existing hash
3111 table will be freed and a new one allocated. If this is ioctl is
3112 called when a hash table has already been allocated of the same order
3113 as specified, the kernel will clear out the existing hash table (zero
3114 all HPTEs). In either case, if the guest is using the virtualized
3115 real-mode area (VRMA) facility, the kernel will re-create the VMRA
3116 HPTEs on the next KVM_RUN of any vcpu.
3118 4.77 KVM_S390_INTERRUPT
3119 -----------------------
3122 :Architectures: s390
3123 :Type: vm ioctl, vcpu ioctl
3124 :Parameters: struct kvm_s390_interrupt (in)
3125 :Returns: 0 on success, -1 on error
3127 Allows to inject an interrupt to the guest. Interrupts can be floating
3128 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3130 Interrupt parameters are passed via kvm_s390_interrupt::
3132 struct kvm_s390_interrupt {
3138 type can be one of the following:
3140 KVM_S390_SIGP_STOP (vcpu)
3141 - sigp stop; optional flags in parm
3142 KVM_S390_PROGRAM_INT (vcpu)
3143 - program check; code in parm
3144 KVM_S390_SIGP_SET_PREFIX (vcpu)
3145 - sigp set prefix; prefix address in parm
3146 KVM_S390_RESTART (vcpu)
3148 KVM_S390_INT_CLOCK_COMP (vcpu)
3149 - clock comparator interrupt
3150 KVM_S390_INT_CPU_TIMER (vcpu)
3151 - CPU timer interrupt
3152 KVM_S390_INT_VIRTIO (vm)
3153 - virtio external interrupt; external interrupt
3154 parameters in parm and parm64
3155 KVM_S390_INT_SERVICE (vm)
3156 - sclp external interrupt; sclp parameter in parm
3157 KVM_S390_INT_EMERGENCY (vcpu)
3158 - sigp emergency; source cpu in parm
3159 KVM_S390_INT_EXTERNAL_CALL (vcpu)
3160 - sigp external call; source cpu in parm
3161 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3162 - compound value to indicate an
3163 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3164 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3165 interruption subclass)
3166 KVM_S390_MCHK (vm, vcpu)
3167 - machine check interrupt; cr 14 bits in parm, machine check interrupt
3168 code in parm64 (note that machine checks needing further payload are not
3169 supported by this ioctl)
3171 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3173 4.78 KVM_PPC_GET_HTAB_FD
3174 ------------------------
3176 :Capability: KVM_CAP_PPC_HTAB_FD
3177 :Architectures: powerpc
3179 :Parameters: Pointer to struct kvm_get_htab_fd (in)
3180 :Returns: file descriptor number (>= 0) on success, -1 on error
3182 This returns a file descriptor that can be used either to read out the
3183 entries in the guest's hashed page table (HPT), or to write entries to
3184 initialize the HPT. The returned fd can only be written to if the
3185 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3186 can only be read if that bit is clear. The argument struct looks like
3189 /* For KVM_PPC_GET_HTAB_FD */
3190 struct kvm_get_htab_fd {
3196 /* Values for kvm_get_htab_fd.flags */
3197 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
3198 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
3200 The 'start_index' field gives the index in the HPT of the entry at
3201 which to start reading. It is ignored when writing.
3203 Reads on the fd will initially supply information about all
3204 "interesting" HPT entries. Interesting entries are those with the
3205 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3206 all entries. When the end of the HPT is reached, the read() will
3207 return. If read() is called again on the fd, it will start again from
3208 the beginning of the HPT, but will only return HPT entries that have
3209 changed since they were last read.
3211 Data read or written is structured as a header (8 bytes) followed by a
3212 series of valid HPT entries (16 bytes) each. The header indicates how
3213 many valid HPT entries there are and how many invalid entries follow
3214 the valid entries. The invalid entries are not represented explicitly
3215 in the stream. The header format is::
3217 struct kvm_get_htab_header {
3223 Writes to the fd create HPT entries starting at the index given in the
3224 header; first 'n_valid' valid entries with contents from the data
3225 written, then 'n_invalid' invalid entries, invalidating any previously
3226 valid entries found.
3228 4.79 KVM_CREATE_DEVICE
3229 ----------------------
3231 :Capability: KVM_CAP_DEVICE_CTRL
3233 :Parameters: struct kvm_create_device (in/out)
3234 :Returns: 0 on success, -1 on error
3238 ====== =======================================================
3239 ENODEV The device type is unknown or unsupported
3240 EEXIST Device already created, and this type of device may not
3241 be instantiated multiple times
3242 ====== =======================================================
3244 Other error conditions may be defined by individual device types or
3245 have their standard meanings.
3247 Creates an emulated device in the kernel. The file descriptor returned
3248 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3250 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3251 device type is supported (not necessarily whether it can be created
3254 Individual devices should not define flags. Attributes should be used
3255 for specifying any behavior that is not implied by the device type
3260 struct kvm_create_device {
3261 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3262 __u32 fd; /* out: device handle */
3263 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3266 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3267 --------------------------------------------
3269 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3270 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3271 :Type: device ioctl, vm ioctl, vcpu ioctl
3272 :Parameters: struct kvm_device_attr
3273 :Returns: 0 on success, -1 on error
3277 ===== =============================================================
3278 ENXIO The group or attribute is unknown/unsupported for this device
3279 or hardware support is missing.
3280 EPERM The attribute cannot (currently) be accessed this way
3281 (e.g. read-only attribute, or attribute that only makes
3282 sense when the device is in a different state)
3283 ===== =============================================================
3285 Other error conditions may be defined by individual device types.
3287 Gets/sets a specified piece of device configuration and/or state. The
3288 semantics are device-specific. See individual device documentation in
3289 the "devices" directory. As with ONE_REG, the size of the data
3290 transferred is defined by the particular attribute.
3294 struct kvm_device_attr {
3295 __u32 flags; /* no flags currently defined */
3296 __u32 group; /* device-defined */
3297 __u64 attr; /* group-defined */
3298 __u64 addr; /* userspace address of attr data */
3301 4.81 KVM_HAS_DEVICE_ATTR
3302 ------------------------
3304 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3305 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3306 :Type: device ioctl, vm ioctl, vcpu ioctl
3307 :Parameters: struct kvm_device_attr
3308 :Returns: 0 on success, -1 on error
3312 ===== =============================================================
3313 ENXIO The group or attribute is unknown/unsupported for this device
3314 or hardware support is missing.
3315 ===== =============================================================
3317 Tests whether a device supports a particular attribute. A successful
3318 return indicates the attribute is implemented. It does not necessarily
3319 indicate that the attribute can be read or written in the device's
3320 current state. "addr" is ignored.
3322 4.82 KVM_ARM_VCPU_INIT
3323 ----------------------
3326 :Architectures: arm, arm64
3328 :Parameters: struct kvm_vcpu_init (in)
3329 :Returns: 0 on success; -1 on error
3333 ====== =================================================================
3334 EINVAL the target is unknown, or the combination of features is invalid.
3335 ENOENT a features bit specified is unknown.
3336 ====== =================================================================
3338 This tells KVM what type of CPU to present to the guest, and what
3339 optional features it should have. This will cause a reset of the cpu
3340 registers to their initial values. If this is not called, KVM_RUN will
3341 return ENOEXEC for that vcpu.
3343 The initial values are defined as:
3345 * AArch64: EL1h, D, A, I and F bits set. All other bits
3347 * AArch32: SVC, A, I and F bits set. All other bits are
3349 - General Purpose registers, including PC and SP: set to 0
3350 - FPSIMD/NEON registers: set to 0
3351 - SVE registers: set to 0
3352 - System registers: Reset to their architecturally defined
3353 values as for a warm reset to EL1 (resp. SVC)
3355 Note that because some registers reflect machine topology, all vcpus
3356 should be created before this ioctl is invoked.
3358 Userspace can call this function multiple times for a given vcpu, including
3359 after the vcpu has been run. This will reset the vcpu to its initial
3360 state. All calls to this function after the initial call must use the same
3361 target and same set of feature flags, otherwise EINVAL will be returned.
3365 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3366 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3367 and execute guest code when KVM_RUN is called.
3368 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3369 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3370 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3371 backward compatible with v0.2) for the CPU.
3372 Depends on KVM_CAP_ARM_PSCI_0_2.
3373 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3374 Depends on KVM_CAP_ARM_PMU_V3.
3376 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3378 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3379 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3380 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3381 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3384 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3386 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3387 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3388 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3389 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3392 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3393 Depends on KVM_CAP_ARM_SVE.
3394 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3396 * After KVM_ARM_VCPU_INIT:
3398 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3399 initial value of this pseudo-register indicates the best set of
3400 vector lengths possible for a vcpu on this host.
3402 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3404 - KVM_RUN and KVM_GET_REG_LIST are not available;
3406 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3407 the scalable archietctural SVE registers
3408 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3409 KVM_REG_ARM64_SVE_FFR;
3411 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3412 KVM_SET_ONE_REG, to modify the set of vector lengths available
3415 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3417 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3418 no longer be written using KVM_SET_ONE_REG.
3420 4.83 KVM_ARM_PREFERRED_TARGET
3421 -----------------------------
3424 :Architectures: arm, arm64
3426 :Parameters: struct kvm_vcpu_init (out)
3427 :Returns: 0 on success; -1 on error
3431 ====== ==========================================
3432 ENODEV no preferred target available for the host
3433 ====== ==========================================
3435 This queries KVM for preferred CPU target type which can be emulated
3436 by KVM on underlying host.
3438 The ioctl returns struct kvm_vcpu_init instance containing information
3439 about preferred CPU target type and recommended features for it. The
3440 kvm_vcpu_init->features bitmap returned will have feature bits set if
3441 the preferred target recommends setting these features, but this is
3444 The information returned by this ioctl can be used to prepare an instance
3445 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3446 VCPU matching underlying host.
3449 4.84 KVM_GET_REG_LIST
3450 ---------------------
3453 :Architectures: arm, arm64, mips
3455 :Parameters: struct kvm_reg_list (in/out)
3456 :Returns: 0 on success; -1 on error
3460 ===== ==============================================================
3461 E2BIG the reg index list is too big to fit in the array specified by
3462 the user (the number required will be written into n).
3463 ===== ==============================================================
3467 struct kvm_reg_list {
3468 __u64 n; /* number of registers in reg[] */
3472 This ioctl returns the guest registers that are supported for the
3473 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3476 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3477 -----------------------------------------
3479 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3480 :Architectures: arm, arm64
3482 :Parameters: struct kvm_arm_device_address (in)
3483 :Returns: 0 on success, -1 on error
3487 ====== ============================================
3488 ENODEV The device id is unknown
3489 ENXIO Device not supported on current system
3490 EEXIST Address already set
3491 E2BIG Address outside guest physical address space
3492 EBUSY Address overlaps with other device range
3493 ====== ============================================
3497 struct kvm_arm_device_addr {
3502 Specify a device address in the guest's physical address space where guests
3503 can access emulated or directly exposed devices, which the host kernel needs
3504 to know about. The id field is an architecture specific identifier for a
3507 ARM/arm64 divides the id field into two parts, a device id and an
3508 address type id specific to the individual device::
3510 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3511 field: | 0x00000000 | device id | addr type id |
3513 ARM/arm64 currently only require this when using the in-kernel GIC
3514 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3515 as the device id. When setting the base address for the guest's
3516 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3517 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3518 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3519 base addresses will return -EEXIST.
3521 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3522 should be used instead.
3525 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3526 ------------------------------
3528 :Capability: KVM_CAP_PPC_RTAS
3531 :Parameters: struct kvm_rtas_token_args
3532 :Returns: 0 on success, -1 on error
3534 Defines a token value for a RTAS (Run Time Abstraction Services)
3535 service in order to allow it to be handled in the kernel. The
3536 argument struct gives the name of the service, which must be the name
3537 of a service that has a kernel-side implementation. If the token
3538 value is non-zero, it will be associated with that service, and
3539 subsequent RTAS calls by the guest specifying that token will be
3540 handled by the kernel. If the token value is 0, then any token
3541 associated with the service will be forgotten, and subsequent RTAS
3542 calls by the guest for that service will be passed to userspace to be
3545 4.87 KVM_SET_GUEST_DEBUG
3546 ------------------------
3548 :Capability: KVM_CAP_SET_GUEST_DEBUG
3549 :Architectures: x86, s390, ppc, arm64
3551 :Parameters: struct kvm_guest_debug (in)
3552 :Returns: 0 on success; -1 on error
3556 struct kvm_guest_debug {
3559 struct kvm_guest_debug_arch arch;
3562 Set up the processor specific debug registers and configure vcpu for
3563 handling guest debug events. There are two parts to the structure, the
3564 first a control bitfield indicates the type of debug events to handle
3565 when running. Common control bits are:
3567 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3568 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3570 The top 16 bits of the control field are architecture specific control
3571 flags which can include the following:
3573 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3574 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390]
3575 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64]
3576 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3577 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3578 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3579 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86]
3581 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3582 are enabled in memory so we need to ensure breakpoint exceptions are
3583 correctly trapped and the KVM run loop exits at the breakpoint and not
3584 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3585 we need to ensure the guest vCPUs architecture specific registers are
3586 updated to the correct (supplied) values.
3588 The second part of the structure is architecture specific and
3589 typically contains a set of debug registers.
3591 For arm64 the number of debug registers is implementation defined and
3592 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3593 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3594 indicating the number of supported registers.
3596 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3597 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3599 Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3600 supported KVM_GUESTDBG_* bits in the control field.
3602 When debug events exit the main run loop with the reason
3603 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3604 structure containing architecture specific debug information.
3606 4.88 KVM_GET_EMULATED_CPUID
3607 ---------------------------
3609 :Capability: KVM_CAP_EXT_EMUL_CPUID
3612 :Parameters: struct kvm_cpuid2 (in/out)
3613 :Returns: 0 on success, -1 on error
3620 struct kvm_cpuid_entry2 entries[0];
3623 The member 'flags' is used for passing flags from userspace.
3627 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3628 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3629 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3631 struct kvm_cpuid_entry2 {
3642 This ioctl returns x86 cpuid features which are emulated by
3643 kvm.Userspace can use the information returned by this ioctl to query
3644 which features are emulated by kvm instead of being present natively.
3646 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3647 structure with the 'nent' field indicating the number of entries in
3648 the variable-size array 'entries'. If the number of entries is too low
3649 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3650 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3651 is returned. If the number is just right, the 'nent' field is adjusted
3652 to the number of valid entries in the 'entries' array, which is then
3655 The entries returned are the set CPUID bits of the respective features
3656 which kvm emulates, as returned by the CPUID instruction, with unknown
3657 or unsupported feature bits cleared.
3659 Features like x2apic, for example, may not be present in the host cpu
3660 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3661 emulated efficiently and thus not included here.
3663 The fields in each entry are defined as follows:
3666 the eax value used to obtain the entry
3668 the ecx value used to obtain the entry (for entries that are
3671 an OR of zero or more of the following:
3673 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3674 if the index field is valid
3678 the values returned by the cpuid instruction for
3679 this function/index combination
3681 4.89 KVM_S390_MEM_OP
3682 --------------------
3684 :Capability: KVM_CAP_S390_MEM_OP
3685 :Architectures: s390
3687 :Parameters: struct kvm_s390_mem_op (in)
3688 :Returns: = 0 on success,
3689 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3690 > 0 if an exception occurred while walking the page tables
3692 Read or write data from/to the logical (virtual) memory of a VCPU.
3694 Parameters are specified via the following structure::
3696 struct kvm_s390_mem_op {
3697 __u64 gaddr; /* the guest address */
3698 __u64 flags; /* flags */
3699 __u32 size; /* amount of bytes */
3700 __u32 op; /* type of operation */
3701 __u64 buf; /* buffer in userspace */
3702 __u8 ar; /* the access register number */
3703 __u8 reserved[31]; /* should be set to 0 */
3706 The type of operation is specified in the "op" field. It is either
3707 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3708 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3709 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3710 whether the corresponding memory access would create an access exception
3711 (without touching the data in the memory at the destination). In case an
3712 access exception occurred while walking the MMU tables of the guest, the
3713 ioctl returns a positive error number to indicate the type of exception.
3714 This exception is also raised directly at the corresponding VCPU if the
3715 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3717 The start address of the memory region has to be specified in the "gaddr"
3718 field, and the length of the region in the "size" field (which must not
3719 be 0). The maximum value for "size" can be obtained by checking the
3720 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3721 userspace application where the read data should be written to for
3722 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3723 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3724 is specified, "buf" is unused and can be NULL. "ar" designates the access
3725 register number to be used; the valid range is 0..15.
3727 The "reserved" field is meant for future extensions. It is not used by
3728 KVM with the currently defined set of flags.
3730 4.90 KVM_S390_GET_SKEYS
3731 -----------------------
3733 :Capability: KVM_CAP_S390_SKEYS
3734 :Architectures: s390
3736 :Parameters: struct kvm_s390_skeys
3737 :Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage
3738 keys, negative value on error
3740 This ioctl is used to get guest storage key values on the s390
3741 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3743 struct kvm_s390_skeys {
3746 __u64 skeydata_addr;
3751 The start_gfn field is the number of the first guest frame whose storage keys
3754 The count field is the number of consecutive frames (starting from start_gfn)
3755 whose storage keys to get. The count field must be at least 1 and the maximum
3756 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3757 will cause the ioctl to return -EINVAL.
3759 The skeydata_addr field is the address to a buffer large enough to hold count
3760 bytes. This buffer will be filled with storage key data by the ioctl.
3762 4.91 KVM_S390_SET_SKEYS
3763 -----------------------
3765 :Capability: KVM_CAP_S390_SKEYS
3766 :Architectures: s390
3768 :Parameters: struct kvm_s390_skeys
3769 :Returns: 0 on success, negative value on error
3771 This ioctl is used to set guest storage key values on the s390
3772 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3773 See section on KVM_S390_GET_SKEYS for struct definition.
3775 The start_gfn field is the number of the first guest frame whose storage keys
3778 The count field is the number of consecutive frames (starting from start_gfn)
3779 whose storage keys to get. The count field must be at least 1 and the maximum
3780 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3781 will cause the ioctl to return -EINVAL.
3783 The skeydata_addr field is the address to a buffer containing count bytes of
3784 storage keys. Each byte in the buffer will be set as the storage key for a
3785 single frame starting at start_gfn for count frames.
3787 Note: If any architecturally invalid key value is found in the given data then
3788 the ioctl will return -EINVAL.
3793 :Capability: KVM_CAP_S390_INJECT_IRQ
3794 :Architectures: s390
3796 :Parameters: struct kvm_s390_irq (in)
3797 :Returns: 0 on success, -1 on error
3802 ====== =================================================================
3803 EINVAL interrupt type is invalid
3804 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3805 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3806 than the maximum of VCPUs
3807 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3808 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3809 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3811 ====== =================================================================
3813 Allows to inject an interrupt to the guest.
3815 Using struct kvm_s390_irq as a parameter allows
3816 to inject additional payload which is not
3817 possible via KVM_S390_INTERRUPT.
3819 Interrupt parameters are passed via kvm_s390_irq::
3821 struct kvm_s390_irq {
3824 struct kvm_s390_io_info io;
3825 struct kvm_s390_ext_info ext;
3826 struct kvm_s390_pgm_info pgm;
3827 struct kvm_s390_emerg_info emerg;
3828 struct kvm_s390_extcall_info extcall;
3829 struct kvm_s390_prefix_info prefix;
3830 struct kvm_s390_stop_info stop;
3831 struct kvm_s390_mchk_info mchk;
3836 type can be one of the following:
3838 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3839 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3840 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3841 - KVM_S390_RESTART - restart; no parameters
3842 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3843 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3844 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3845 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3846 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3848 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3850 4.94 KVM_S390_GET_IRQ_STATE
3851 ---------------------------
3853 :Capability: KVM_CAP_S390_IRQ_STATE
3854 :Architectures: s390
3856 :Parameters: struct kvm_s390_irq_state (out)
3857 :Returns: >= number of bytes copied into buffer,
3858 -EINVAL if buffer size is 0,
3859 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3860 -EFAULT if the buffer address was invalid
3862 This ioctl allows userspace to retrieve the complete state of all currently
3863 pending interrupts in a single buffer. Use cases include migration
3864 and introspection. The parameter structure contains the address of a
3865 userspace buffer and its length::
3867 struct kvm_s390_irq_state {
3869 __u32 flags; /* will stay unused for compatibility reasons */
3871 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3874 Userspace passes in the above struct and for each pending interrupt a
3875 struct kvm_s390_irq is copied to the provided buffer.
3877 The structure contains a flags and a reserved field for future extensions. As
3878 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3879 reserved, these fields can not be used in the future without breaking
3882 If -ENOBUFS is returned the buffer provided was too small and userspace
3883 may retry with a bigger buffer.
3885 4.95 KVM_S390_SET_IRQ_STATE
3886 ---------------------------
3888 :Capability: KVM_CAP_S390_IRQ_STATE
3889 :Architectures: s390
3891 :Parameters: struct kvm_s390_irq_state (in)
3892 :Returns: 0 on success,
3893 -EFAULT if the buffer address was invalid,
3894 -EINVAL for an invalid buffer length (see below),
3895 -EBUSY if there were already interrupts pending,
3896 errors occurring when actually injecting the
3897 interrupt. See KVM_S390_IRQ.
3899 This ioctl allows userspace to set the complete state of all cpu-local
3900 interrupts currently pending for the vcpu. It is intended for restoring
3901 interrupt state after a migration. The input parameter is a userspace buffer
3902 containing a struct kvm_s390_irq_state::
3904 struct kvm_s390_irq_state {
3906 __u32 flags; /* will stay unused for compatibility reasons */
3908 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3911 The restrictions for flags and reserved apply as well.
3912 (see KVM_S390_GET_IRQ_STATE)
3914 The userspace memory referenced by buf contains a struct kvm_s390_irq
3915 for each interrupt to be injected into the guest.
3916 If one of the interrupts could not be injected for some reason the
3919 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3920 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3921 which is the maximum number of possibly pending cpu-local interrupts.
3926 :Capability: KVM_CAP_X86_SMM
3930 :Returns: 0 on success, -1 on error
3932 Queues an SMI on the thread's vcpu.
3934 4.97 KVM_X86_SET_MSR_FILTER
3935 ----------------------------
3937 :Capability: KVM_X86_SET_MSR_FILTER
3940 :Parameters: struct kvm_msr_filter
3941 :Returns: 0 on success, < 0 on error
3945 struct kvm_msr_filter_range {
3946 #define KVM_MSR_FILTER_READ (1 << 0)
3947 #define KVM_MSR_FILTER_WRITE (1 << 1)
3949 __u32 nmsrs; /* number of msrs in bitmap */
3950 __u32 base; /* MSR index the bitmap starts at */
3951 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
3954 #define KVM_MSR_FILTER_MAX_RANGES 16
3955 struct kvm_msr_filter {
3956 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
3957 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
3959 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
3962 flags values for ``struct kvm_msr_filter_range``:
3964 ``KVM_MSR_FILTER_READ``
3966 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
3967 indicates that a read should immediately fail, while a 1 indicates that
3968 a read for a particular MSR should be handled regardless of the default
3971 ``KVM_MSR_FILTER_WRITE``
3973 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
3974 indicates that a write should immediately fail, while a 1 indicates that
3975 a write for a particular MSR should be handled regardless of the default
3978 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
3980 Filter both read and write accesses to MSRs using the given bitmap. A 0
3981 in the bitmap indicates that both reads and writes should immediately fail,
3982 while a 1 indicates that reads and writes for a particular MSR are not
3983 filtered by this range.
3985 flags values for ``struct kvm_msr_filter``:
3987 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
3989 If no filter range matches an MSR index that is getting accessed, KVM will
3990 fall back to allowing access to the MSR.
3992 ``KVM_MSR_FILTER_DEFAULT_DENY``
3994 If no filter range matches an MSR index that is getting accessed, KVM will
3995 fall back to rejecting access to the MSR. In this mode, all MSRs that should
3996 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
3998 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
3999 specify whether a certain MSR access should be explicitly filtered for or not.
4001 If this ioctl has never been invoked, MSR accesses are not guarded and the
4002 default KVM in-kernel emulation behavior is fully preserved.
4004 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4005 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4008 As soon as the filtering is in place, every MSR access is processed through
4009 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4010 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4011 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4014 If a bit is within one of the defined ranges, read and write accesses are
4015 guarded by the bitmap's value for the MSR index if the kind of access
4016 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4017 cover this particular access, the behavior is determined by the flags
4018 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4019 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4021 Each bitmap range specifies a range of MSRs to potentially allow access on.
4022 The range goes from MSR index [base .. base+nmsrs]. The flags field
4023 indicates whether reads, writes or both reads and writes are filtered
4024 by setting a 1 bit in the bitmap for the corresponding MSR index.
4026 If an MSR access is not permitted through the filtering, it generates a
4027 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4028 allows user space to deflect and potentially handle various MSR accesses
4031 If a vCPU is in running state while this ioctl is invoked, the vCPU may
4032 experience inconsistent filtering behavior on MSR accesses.
4034 4.98 KVM_CREATE_SPAPR_TCE_64
4035 ----------------------------
4037 :Capability: KVM_CAP_SPAPR_TCE_64
4038 :Architectures: powerpc
4040 :Parameters: struct kvm_create_spapr_tce_64 (in)
4041 :Returns: file descriptor for manipulating the created TCE table
4043 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4044 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4046 This capability uses extended struct in ioctl interface::
4048 /* for KVM_CAP_SPAPR_TCE_64 */
4049 struct kvm_create_spapr_tce_64 {
4053 __u64 offset; /* in pages */
4054 __u64 size; /* in pages */
4057 The aim of extension is to support an additional bigger DMA window with
4058 a variable page size.
4059 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4060 a bus offset of the corresponding DMA window, @size and @offset are numbers
4063 @flags are not used at the moment.
4065 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4067 4.99 KVM_REINJECT_CONTROL
4068 -------------------------
4070 :Capability: KVM_CAP_REINJECT_CONTROL
4073 :Parameters: struct kvm_reinject_control (in)
4074 :Returns: 0 on success,
4075 -EFAULT if struct kvm_reinject_control cannot be read,
4076 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4078 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
4079 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4080 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
4081 interrupt whenever there isn't a pending interrupt from i8254.
4082 !reinject mode injects an interrupt as soon as a tick arrives.
4086 struct kvm_reinject_control {
4091 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4092 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4094 4.100 KVM_PPC_CONFIGURE_V3_MMU
4095 ------------------------------
4097 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
4100 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
4101 :Returns: 0 on success,
4102 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4103 -EINVAL if the configuration is invalid
4105 This ioctl controls whether the guest will use radix or HPT (hashed
4106 page table) translation, and sets the pointer to the process table for
4111 struct kvm_ppc_mmuv3_cfg {
4113 __u64 process_table;
4116 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4117 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
4118 to use radix tree translation, and if clear, to use HPT translation.
4119 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4120 to be able to use the global TLB and SLB invalidation instructions;
4121 if clear, the guest may not use these instructions.
4123 The process_table field specifies the address and size of the guest
4124 process table, which is in the guest's space. This field is formatted
4125 as the second doubleword of the partition table entry, as defined in
4126 the Power ISA V3.00, Book III section 5.7.6.1.
4128 4.101 KVM_PPC_GET_RMMU_INFO
4129 ---------------------------
4131 :Capability: KVM_CAP_PPC_RADIX_MMU
4134 :Parameters: struct kvm_ppc_rmmu_info (out)
4135 :Returns: 0 on success,
4136 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4137 -EINVAL if no useful information can be returned
4139 This ioctl returns a structure containing two things: (a) a list
4140 containing supported radix tree geometries, and (b) a list that maps
4141 page sizes to put in the "AP" (actual page size) field for the tlbie
4142 (TLB invalidate entry) instruction.
4146 struct kvm_ppc_rmmu_info {
4147 struct kvm_ppc_radix_geom {
4152 __u32 ap_encodings[8];
4155 The geometries[] field gives up to 8 supported geometries for the
4156 radix page table, in terms of the log base 2 of the smallest page
4157 size, and the number of bits indexed at each level of the tree, from
4158 the PTE level up to the PGD level in that order. Any unused entries
4159 will have 0 in the page_shift field.
4161 The ap_encodings gives the supported page sizes and their AP field
4162 encodings, encoded with the AP value in the top 3 bits and the log
4163 base 2 of the page size in the bottom 6 bits.
4165 4.102 KVM_PPC_RESIZE_HPT_PREPARE
4166 --------------------------------
4168 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4169 :Architectures: powerpc
4171 :Parameters: struct kvm_ppc_resize_hpt (in)
4172 :Returns: 0 on successful completion,
4173 >0 if a new HPT is being prepared, the value is an estimated
4174 number of milliseconds until preparation is complete,
4175 -EFAULT if struct kvm_reinject_control cannot be read,
4176 -EINVAL if the supplied shift or flags are invalid,
4177 -ENOMEM if unable to allocate the new HPT,
4179 Used to implement the PAPR extension for runtime resizing of a guest's
4180 Hashed Page Table (HPT). Specifically this starts, stops or monitors
4181 the preparation of a new potential HPT for the guest, essentially
4182 implementing the H_RESIZE_HPT_PREPARE hypercall.
4186 struct kvm_ppc_resize_hpt {
4192 If called with shift > 0 when there is no pending HPT for the guest,
4193 this begins preparation of a new pending HPT of size 2^(shift) bytes.
4194 It then returns a positive integer with the estimated number of
4195 milliseconds until preparation is complete.
4197 If called when there is a pending HPT whose size does not match that
4198 requested in the parameters, discards the existing pending HPT and
4199 creates a new one as above.
4201 If called when there is a pending HPT of the size requested, will:
4203 * If preparation of the pending HPT is already complete, return 0
4204 * If preparation of the pending HPT has failed, return an error
4205 code, then discard the pending HPT.
4206 * If preparation of the pending HPT is still in progress, return an
4207 estimated number of milliseconds until preparation is complete.
4209 If called with shift == 0, discards any currently pending HPT and
4210 returns 0 (i.e. cancels any in-progress preparation).
4212 flags is reserved for future expansion, currently setting any bits in
4213 flags will result in an -EINVAL.
4215 Normally this will be called repeatedly with the same parameters until
4216 it returns <= 0. The first call will initiate preparation, subsequent
4217 ones will monitor preparation until it completes or fails.
4219 4.103 KVM_PPC_RESIZE_HPT_COMMIT
4220 -------------------------------
4222 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4223 :Architectures: powerpc
4225 :Parameters: struct kvm_ppc_resize_hpt (in)
4226 :Returns: 0 on successful completion,
4227 -EFAULT if struct kvm_reinject_control cannot be read,
4228 -EINVAL if the supplied shift or flags are invalid,
4229 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4230 have the requested size,
4231 -EBUSY if the pending HPT is not fully prepared,
4232 -ENOSPC if there was a hash collision when moving existing
4233 HPT entries to the new HPT,
4234 -EIO on other error conditions
4236 Used to implement the PAPR extension for runtime resizing of a guest's
4237 Hashed Page Table (HPT). Specifically this requests that the guest be
4238 transferred to working with the new HPT, essentially implementing the
4239 H_RESIZE_HPT_COMMIT hypercall.
4243 struct kvm_ppc_resize_hpt {
4249 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4250 returned 0 with the same parameters. In other cases
4251 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4252 -EBUSY, though others may be possible if the preparation was started,
4255 This will have undefined effects on the guest if it has not already
4256 placed itself in a quiescent state where no vcpu will make MMU enabled
4259 On succsful completion, the pending HPT will become the guest's active
4260 HPT and the previous HPT will be discarded.
4262 On failure, the guest will still be operating on its previous HPT.
4264 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4265 -----------------------------------
4267 :Capability: KVM_CAP_MCE
4270 :Parameters: u64 mce_cap (out)
4271 :Returns: 0 on success, -1 on error
4273 Returns supported MCE capabilities. The u64 mce_cap parameter
4274 has the same format as the MSR_IA32_MCG_CAP register. Supported
4275 capabilities will have the corresponding bits set.
4277 4.105 KVM_X86_SETUP_MCE
4278 -----------------------
4280 :Capability: KVM_CAP_MCE
4283 :Parameters: u64 mcg_cap (in)
4284 :Returns: 0 on success,
4285 -EFAULT if u64 mcg_cap cannot be read,
4286 -EINVAL if the requested number of banks is invalid,
4287 -EINVAL if requested MCE capability is not supported.
4289 Initializes MCE support for use. The u64 mcg_cap parameter
4290 has the same format as the MSR_IA32_MCG_CAP register and
4291 specifies which capabilities should be enabled. The maximum
4292 supported number of error-reporting banks can be retrieved when
4293 checking for KVM_CAP_MCE. The supported capabilities can be
4294 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4296 4.106 KVM_X86_SET_MCE
4297 ---------------------
4299 :Capability: KVM_CAP_MCE
4302 :Parameters: struct kvm_x86_mce (in)
4303 :Returns: 0 on success,
4304 -EFAULT if struct kvm_x86_mce cannot be read,
4305 -EINVAL if the bank number is invalid,
4306 -EINVAL if VAL bit is not set in status field.
4308 Inject a machine check error (MCE) into the guest. The input
4311 struct kvm_x86_mce {
4321 If the MCE being reported is an uncorrected error, KVM will
4322 inject it as an MCE exception into the guest. If the guest
4323 MCG_STATUS register reports that an MCE is in progress, KVM
4324 causes an KVM_EXIT_SHUTDOWN vmexit.
4326 Otherwise, if the MCE is a corrected error, KVM will just
4327 store it in the corresponding bank (provided this bank is
4328 not holding a previously reported uncorrected error).
4330 4.107 KVM_S390_GET_CMMA_BITS
4331 ----------------------------
4333 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4334 :Architectures: s390
4336 :Parameters: struct kvm_s390_cmma_log (in, out)
4337 :Returns: 0 on success, a negative value on error
4339 This ioctl is used to get the values of the CMMA bits on the s390
4340 architecture. It is meant to be used in two scenarios:
4342 - During live migration to save the CMMA values. Live migration needs
4343 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4344 - To non-destructively peek at the CMMA values, with the flag
4345 KVM_S390_CMMA_PEEK set.
4347 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4348 values are written to a buffer whose location is indicated via the "values"
4349 member in the kvm_s390_cmma_log struct. The values in the input struct are
4350 also updated as needed.
4352 Each CMMA value takes up one byte.
4356 struct kvm_s390_cmma_log {
4367 start_gfn is the number of the first guest frame whose CMMA values are
4370 count is the length of the buffer in bytes,
4372 values points to the buffer where the result will be written to.
4374 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4375 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4378 The result is written in the buffer pointed to by the field values, and
4379 the values of the input parameter are updated as follows.
4381 Depending on the flags, different actions are performed. The only
4382 supported flag so far is KVM_S390_CMMA_PEEK.
4384 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4385 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4386 It is not necessarily the same as the one passed as input, as clean pages
4389 count will indicate the number of bytes actually written in the buffer.
4390 It can (and very often will) be smaller than the input value, since the
4391 buffer is only filled until 16 bytes of clean values are found (which
4392 are then not copied in the buffer). Since a CMMA migration block needs
4393 the base address and the length, for a total of 16 bytes, we will send
4394 back some clean data if there is some dirty data afterwards, as long as
4395 the size of the clean data does not exceed the size of the header. This
4396 allows to minimize the amount of data to be saved or transferred over
4397 the network at the expense of more roundtrips to userspace. The next
4398 invocation of the ioctl will skip over all the clean values, saving
4399 potentially more than just the 16 bytes we found.
4401 If KVM_S390_CMMA_PEEK is set:
4402 the existing storage attributes are read even when not in migration
4403 mode, and no other action is performed;
4405 the output start_gfn will be equal to the input start_gfn,
4407 the output count will be equal to the input count, except if the end of
4408 memory has been reached.
4411 the field "remaining" will indicate the total number of dirty CMMA values
4412 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4417 values points to the userspace buffer where the result will be stored.
4419 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4420 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4421 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4422 -EFAULT if the userspace address is invalid or if no page table is
4423 present for the addresses (e.g. when using hugepages).
4425 4.108 KVM_S390_SET_CMMA_BITS
4426 ----------------------------
4428 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4429 :Architectures: s390
4431 :Parameters: struct kvm_s390_cmma_log (in)
4432 :Returns: 0 on success, a negative value on error
4434 This ioctl is used to set the values of the CMMA bits on the s390
4435 architecture. It is meant to be used during live migration to restore
4436 the CMMA values, but there are no restrictions on its use.
4437 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4438 Each CMMA value takes up one byte.
4442 struct kvm_s390_cmma_log {
4453 start_gfn indicates the starting guest frame number,
4455 count indicates how many values are to be considered in the buffer,
4457 flags is not used and must be 0.
4459 mask indicates which PGSTE bits are to be considered.
4461 remaining is not used.
4463 values points to the buffer in userspace where to store the values.
4465 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4466 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4467 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4468 if the flags field was not 0, with -EFAULT if the userspace address is
4469 invalid, if invalid pages are written to (e.g. after the end of memory)
4470 or if no page table is present for the addresses (e.g. when using
4473 4.109 KVM_PPC_GET_CPU_CHAR
4474 --------------------------
4476 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4477 :Architectures: powerpc
4479 :Parameters: struct kvm_ppc_cpu_char (out)
4480 :Returns: 0 on successful completion,
4481 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4483 This ioctl gives userspace information about certain characteristics
4484 of the CPU relating to speculative execution of instructions and
4485 possible information leakage resulting from speculative execution (see
4486 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4487 returned in struct kvm_ppc_cpu_char, which looks like this::
4489 struct kvm_ppc_cpu_char {
4490 __u64 character; /* characteristics of the CPU */
4491 __u64 behaviour; /* recommended software behaviour */
4492 __u64 character_mask; /* valid bits in character */
4493 __u64 behaviour_mask; /* valid bits in behaviour */
4496 For extensibility, the character_mask and behaviour_mask fields
4497 indicate which bits of character and behaviour have been filled in by
4498 the kernel. If the set of defined bits is extended in future then
4499 userspace will be able to tell whether it is running on a kernel that
4500 knows about the new bits.
4502 The character field describes attributes of the CPU which can help
4503 with preventing inadvertent information disclosure - specifically,
4504 whether there is an instruction to flash-invalidate the L1 data cache
4505 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4506 to a mode where entries can only be used by the thread that created
4507 them, whether the bcctr[l] instruction prevents speculation, and
4508 whether a speculation barrier instruction (ori 31,31,0) is provided.
4510 The behaviour field describes actions that software should take to
4511 prevent inadvertent information disclosure, and thus describes which
4512 vulnerabilities the hardware is subject to; specifically whether the
4513 L1 data cache should be flushed when returning to user mode from the
4514 kernel, and whether a speculation barrier should be placed between an
4515 array bounds check and the array access.
4517 These fields use the same bit definitions as the new
4518 H_GET_CPU_CHARACTERISTICS hypercall.
4520 4.110 KVM_MEMORY_ENCRYPT_OP
4521 ---------------------------
4526 :Parameters: an opaque platform specific structure (in/out)
4527 :Returns: 0 on success; -1 on error
4529 If the platform supports creating encrypted VMs then this ioctl can be used
4530 for issuing platform-specific memory encryption commands to manage those
4533 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4534 (SEV) commands on AMD Processors. The SEV commands are defined in
4535 Documentation/virt/kvm/amd-memory-encryption.rst.
4537 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4538 -----------------------------------
4543 :Parameters: struct kvm_enc_region (in)
4544 :Returns: 0 on success; -1 on error
4546 This ioctl can be used to register a guest memory region which may
4547 contain encrypted data (e.g. guest RAM, SMRAM etc).
4549 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4550 memory region may contain encrypted data. The SEV memory encryption
4551 engine uses a tweak such that two identical plaintext pages, each at
4552 different locations will have differing ciphertexts. So swapping or
4553 moving ciphertext of those pages will not result in plaintext being
4554 swapped. So relocating (or migrating) physical backing pages for the SEV
4555 guest will require some additional steps.
4557 Note: The current SEV key management spec does not provide commands to
4558 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4559 memory region registered with the ioctl.
4561 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4562 -------------------------------------
4567 :Parameters: struct kvm_enc_region (in)
4568 :Returns: 0 on success; -1 on error
4570 This ioctl can be used to unregister the guest memory region registered
4571 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4573 4.113 KVM_HYPERV_EVENTFD
4574 ------------------------
4576 :Capability: KVM_CAP_HYPERV_EVENTFD
4579 :Parameters: struct kvm_hyperv_eventfd (in)
4581 This ioctl (un)registers an eventfd to receive notifications from the guest on
4582 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4583 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4584 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4588 struct kvm_hyperv_eventfd {
4595 The conn_id field should fit within 24 bits::
4597 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4599 The acceptable values for the flags field are::
4601 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4603 :Returns: 0 on success,
4604 -EINVAL if conn_id or flags is outside the allowed range,
4605 -ENOENT on deassign if the conn_id isn't registered,
4606 -EEXIST on assign if the conn_id is already registered
4608 4.114 KVM_GET_NESTED_STATE
4609 --------------------------
4611 :Capability: KVM_CAP_NESTED_STATE
4614 :Parameters: struct kvm_nested_state (in/out)
4615 :Returns: 0 on success, -1 on error
4619 ===== =============================================================
4620 E2BIG the total state size exceeds the value of 'size' specified by
4621 the user; the size required will be written into size.
4622 ===== =============================================================
4626 struct kvm_nested_state {
4632 struct kvm_vmx_nested_state_hdr vmx;
4633 struct kvm_svm_nested_state_hdr svm;
4635 /* Pad the header to 128 bytes. */
4640 struct kvm_vmx_nested_state_data vmx[0];
4641 struct kvm_svm_nested_state_data svm[0];
4645 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4646 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4647 #define KVM_STATE_NESTED_EVMCS 0x00000004
4649 #define KVM_STATE_NESTED_FORMAT_VMX 0
4650 #define KVM_STATE_NESTED_FORMAT_SVM 1
4652 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4654 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4655 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4657 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4659 struct kvm_vmx_nested_state_hdr {
4668 __u64 preemption_timer_deadline;
4671 struct kvm_vmx_nested_state_data {
4672 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4673 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4676 This ioctl copies the vcpu's nested virtualization state from the kernel to
4679 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4680 to the KVM_CHECK_EXTENSION ioctl().
4682 4.115 KVM_SET_NESTED_STATE
4683 --------------------------
4685 :Capability: KVM_CAP_NESTED_STATE
4688 :Parameters: struct kvm_nested_state (in)
4689 :Returns: 0 on success, -1 on error
4691 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4692 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4694 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4695 -------------------------------------
4697 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4698 KVM_CAP_COALESCED_PIO (for coalesced pio)
4701 :Parameters: struct kvm_coalesced_mmio_zone
4702 :Returns: 0 on success, < 0 on error
4704 Coalesced I/O is a performance optimization that defers hardware
4705 register write emulation so that userspace exits are avoided. It is
4706 typically used to reduce the overhead of emulating frequently accessed
4709 When a hardware register is configured for coalesced I/O, write accesses
4710 do not exit to userspace and their value is recorded in a ring buffer
4711 that is shared between kernel and userspace.
4713 Coalesced I/O is used if one or more write accesses to a hardware
4714 register can be deferred until a read or a write to another hardware
4715 register on the same device. This last access will cause a vmexit and
4716 userspace will process accesses from the ring buffer before emulating
4717 it. That will avoid exiting to userspace on repeated writes.
4719 Coalesced pio is based on coalesced mmio. There is little difference
4720 between coalesced mmio and pio except that coalesced pio records accesses
4723 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4724 ------------------------------------
4726 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4727 :Architectures: x86, arm, arm64, mips
4729 :Parameters: struct kvm_clear_dirty_log (in)
4730 :Returns: 0 on success, -1 on error
4734 /* for KVM_CLEAR_DIRTY_LOG */
4735 struct kvm_clear_dirty_log {
4740 void __user *dirty_bitmap; /* one bit per page */
4745 The ioctl clears the dirty status of pages in a memory slot, according to
4746 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4747 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4748 memory slot, and num_pages is the size in bits of the input bitmap.
4749 first_page must be a multiple of 64; num_pages must also be a multiple of
4750 64 unless first_page + num_pages is the size of the memory slot. For each
4751 bit that is set in the input bitmap, the corresponding page is marked "clean"
4752 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4753 (for example via write-protection, or by clearing the dirty bit in
4754 a page table entry).
4756 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4757 the address space for which you want to clear the dirty status. See
4758 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4760 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4761 is enabled; for more information, see the description of the capability.
4762 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4763 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4765 4.118 KVM_GET_SUPPORTED_HV_CPUID
4766 --------------------------------
4768 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4770 :Type: system ioctl, vcpu ioctl
4771 :Parameters: struct kvm_cpuid2 (in/out)
4772 :Returns: 0 on success, -1 on error
4779 struct kvm_cpuid_entry2 entries[0];
4782 struct kvm_cpuid_entry2 {
4793 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4794 KVM. Userspace can use the information returned by this ioctl to construct
4795 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4796 Windows or Hyper-V guests).
4798 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4799 Functional Specification (TLFS). These leaves can't be obtained with
4800 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4801 leaves (0x40000000, 0x40000001).
4803 Currently, the following list of CPUID leaves are returned:
4805 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4806 - HYPERV_CPUID_INTERFACE
4807 - HYPERV_CPUID_VERSION
4808 - HYPERV_CPUID_FEATURES
4809 - HYPERV_CPUID_ENLIGHTMENT_INFO
4810 - HYPERV_CPUID_IMPLEMENT_LIMITS
4811 - HYPERV_CPUID_NESTED_FEATURES
4812 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4813 - HYPERV_CPUID_SYNDBG_INTERFACE
4814 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4816 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4817 with the 'nent' field indicating the number of entries in the variable-size
4818 array 'entries'. If the number of entries is too low to describe all Hyper-V
4819 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4820 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4821 number of valid entries in the 'entries' array, which is then filled.
4823 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4824 userspace should not expect to get any particular value there.
4826 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4827 system ioctl which exposes all supported feature bits unconditionally, vcpu
4828 version has the following quirks:
4830 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4831 feature bit are only exposed when Enlightened VMCS was previously enabled
4832 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4833 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4834 (presumes KVM_CREATE_IRQCHIP has already been called).
4836 4.119 KVM_ARM_VCPU_FINALIZE
4837 ---------------------------
4839 :Architectures: arm, arm64
4841 :Parameters: int feature (in)
4842 :Returns: 0 on success, -1 on error
4846 ====== ==============================================================
4847 EPERM feature not enabled, needs configuration, or already finalized
4848 EINVAL feature unknown or not present
4849 ====== ==============================================================
4851 Recognised values for feature:
4853 ===== ===========================================
4854 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4855 ===== ===========================================
4857 Finalizes the configuration of the specified vcpu feature.
4859 The vcpu must already have been initialised, enabling the affected feature, by
4860 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4863 For affected vcpu features, this is a mandatory step that must be performed
4864 before the vcpu is fully usable.
4866 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4867 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4868 that should be performaned and how to do it are feature-dependent.
4870 Other calls that depend on a particular feature being finalized, such as
4871 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4872 -EPERM unless the feature has already been finalized by means of a
4873 KVM_ARM_VCPU_FINALIZE call.
4875 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4878 4.120 KVM_SET_PMU_EVENT_FILTER
4879 ------------------------------
4881 :Capability: KVM_CAP_PMU_EVENT_FILTER
4884 :Parameters: struct kvm_pmu_event_filter (in)
4885 :Returns: 0 on success, -1 on error
4889 struct kvm_pmu_event_filter {
4892 __u32 fixed_counter_bitmap;
4898 This ioctl restricts the set of PMU events that the guest can program.
4899 The argument holds a list of events which will be allowed or denied.
4900 The eventsel+umask of each event the guest attempts to program is compared
4901 against the events field to determine whether the guest should have access.
4902 The events field only controls general purpose counters; fixed purpose
4903 counters are controlled by the fixed_counter_bitmap.
4905 No flags are defined yet, the field must be zero.
4907 Valid values for 'action'::
4909 #define KVM_PMU_EVENT_ALLOW 0
4910 #define KVM_PMU_EVENT_DENY 1
4912 4.121 KVM_PPC_SVM_OFF
4913 ---------------------
4916 :Architectures: powerpc
4919 :Returns: 0 on successful completion,
4923 ====== ================================================================
4924 EINVAL if ultravisor failed to terminate the secure guest
4925 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4926 ====== ================================================================
4928 This ioctl is used to turn off the secure mode of the guest or transition
4929 the guest from secure mode to normal mode. This is invoked when the guest
4930 is reset. This has no effect if called for a normal guest.
4932 This ioctl issues an ultravisor call to terminate the secure guest,
4933 unpins the VPA pages and releases all the device pages that are used to
4934 track the secure pages by hypervisor.
4936 4.122 KVM_S390_NORMAL_RESET
4937 ---------------------------
4939 :Capability: KVM_CAP_S390_VCPU_RESETS
4940 :Architectures: s390
4945 This ioctl resets VCPU registers and control structures according to
4946 the cpu reset definition in the POP (Principles Of Operation).
4948 4.123 KVM_S390_INITIAL_RESET
4949 ----------------------------
4952 :Architectures: s390
4957 This ioctl resets VCPU registers and control structures according to
4958 the initial cpu reset definition in the POP. However, the cpu is not
4959 put into ESA mode. This reset is a superset of the normal reset.
4961 4.124 KVM_S390_CLEAR_RESET
4962 --------------------------
4964 :Capability: KVM_CAP_S390_VCPU_RESETS
4965 :Architectures: s390
4970 This ioctl resets VCPU registers and control structures according to
4971 the clear cpu reset definition in the POP. However, the cpu is not put
4972 into ESA mode. This reset is a superset of the initial reset.
4975 4.125 KVM_S390_PV_COMMAND
4976 -------------------------
4978 :Capability: KVM_CAP_S390_PROTECTED
4979 :Architectures: s390
4981 :Parameters: struct kvm_pv_cmd
4982 :Returns: 0 on success, < 0 on error
4987 __u32 cmd; /* Command to be executed */
4988 __u16 rc; /* Ultravisor return code */
4989 __u16 rrc; /* Ultravisor return reason code */
4990 __u64 data; /* Data or address */
4991 __u32 flags; /* flags for future extensions. Must be 0 for now */
4998 Allocate memory and register the VM with the Ultravisor, thereby
4999 donating memory to the Ultravisor that will become inaccessible to
5000 KVM. All existing CPUs are converted to protected ones. After this
5001 command has succeeded, any CPU added via hotplug will become
5002 protected during its creation as well.
5006 ===== =============================
5007 EINTR an unmasked signal is pending
5008 ===== =============================
5012 Deregister the VM from the Ultravisor and reclaim the memory that
5013 had been donated to the Ultravisor, making it usable by the kernel
5014 again. All registered VCPUs are converted back to non-protected
5017 KVM_PV_VM_SET_SEC_PARMS
5018 Pass the image header from VM memory to the Ultravisor in
5019 preparation of image unpacking and verification.
5022 Unpack (protect and decrypt) a page of the encrypted boot image.
5025 Verify the integrity of the unpacked image. Only if this succeeds,
5026 KVM is allowed to start protected VCPUs.
5028 4.126 KVM_X86_SET_MSR_FILTER
5029 ----------------------------
5031 :Capability: KVM_CAP_X86_MSR_FILTER
5034 :Parameters: struct kvm_msr_filter
5035 :Returns: 0 on success, < 0 on error
5039 struct kvm_msr_filter_range {
5040 #define KVM_MSR_FILTER_READ (1 << 0)
5041 #define KVM_MSR_FILTER_WRITE (1 << 1)
5043 __u32 nmsrs; /* number of msrs in bitmap */
5044 __u32 base; /* MSR index the bitmap starts at */
5045 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
5048 #define KVM_MSR_FILTER_MAX_RANGES 16
5049 struct kvm_msr_filter {
5050 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
5051 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
5053 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
5056 flags values for ``struct kvm_msr_filter_range``:
5058 ``KVM_MSR_FILTER_READ``
5060 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
5061 indicates that a read should immediately fail, while a 1 indicates that
5062 a read for a particular MSR should be handled regardless of the default
5065 ``KVM_MSR_FILTER_WRITE``
5067 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
5068 indicates that a write should immediately fail, while a 1 indicates that
5069 a write for a particular MSR should be handled regardless of the default
5072 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
5074 Filter both read and write accesses to MSRs using the given bitmap. A 0
5075 in the bitmap indicates that both reads and writes should immediately fail,
5076 while a 1 indicates that reads and writes for a particular MSR are not
5077 filtered by this range.
5079 flags values for ``struct kvm_msr_filter``:
5081 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
5083 If no filter range matches an MSR index that is getting accessed, KVM will
5084 fall back to allowing access to the MSR.
5086 ``KVM_MSR_FILTER_DEFAULT_DENY``
5088 If no filter range matches an MSR index that is getting accessed, KVM will
5089 fall back to rejecting access to the MSR. In this mode, all MSRs that should
5090 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
5092 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
5093 specify whether a certain MSR access should be explicitly filtered for or not.
5095 If this ioctl has never been invoked, MSR accesses are not guarded and the
5096 default KVM in-kernel emulation behavior is fully preserved.
5098 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
5099 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
5102 As soon as the filtering is in place, every MSR access is processed through
5103 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
5104 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
5105 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
5108 If a bit is within one of the defined ranges, read and write accesses are
5109 guarded by the bitmap's value for the MSR index if the kind of access
5110 is included in the ``struct kvm_msr_filter_range`` flags. If no range
5111 cover this particular access, the behavior is determined by the flags
5112 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
5113 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
5115 Each bitmap range specifies a range of MSRs to potentially allow access on.
5116 The range goes from MSR index [base .. base+nmsrs]. The flags field
5117 indicates whether reads, writes or both reads and writes are filtered
5118 by setting a 1 bit in the bitmap for the corresponding MSR index.
5120 If an MSR access is not permitted through the filtering, it generates a
5121 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
5122 allows user space to deflect and potentially handle various MSR accesses
5125 Note, invoking this ioctl with a vCPU is running is inherently racy. However,
5126 KVM does guarantee that vCPUs will see either the previous filter or the new
5127 filter, e.g. MSRs with identical settings in both the old and new filter will
5128 have deterministic behavior.
5130 4.127 KVM_XEN_HVM_SET_ATTR
5131 --------------------------
5133 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5136 :Parameters: struct kvm_xen_hvm_attr
5137 :Returns: 0 on success, < 0 on error
5141 struct kvm_xen_hvm_attr {
5156 KVM_XEN_ATTR_TYPE_LONG_MODE
5157 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
5158 determines the layout of the shared info pages exposed to the VM.
5160 KVM_XEN_ATTR_TYPE_SHARED_INFO
5161 Sets the guest physical frame number at which the Xen "shared info"
5162 page resides. Note that although Xen places vcpu_info for the first
5163 32 vCPUs in the shared_info page, KVM does not automatically do so
5164 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
5165 explicitly even when the vcpu_info for a given vCPU resides at the
5166 "default" location in the shared_info page. This is because KVM is
5167 not aware of the Xen CPU id which is used as the index into the
5168 vcpu_info[] array, so cannot know the correct default location.
5170 Note that the shared info page may be constantly written to by KVM;
5171 it contains the event channel bitmap used to deliver interrupts to
5172 a Xen guest, amongst other things. It is exempt from dirty tracking
5173 mechanisms — KVM will not explicitly mark the page as dirty each
5174 time an event channel interrupt is delivered to the guest! Thus,
5175 userspace should always assume that the designated GFN is dirty if
5176 any vCPU has been running or any event channel interrupts can be
5177 routed to the guest.
5179 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
5180 Sets the exception vector used to deliver Xen event channel upcalls.
5182 4.127 KVM_XEN_HVM_GET_ATTR
5183 --------------------------
5185 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5188 :Parameters: struct kvm_xen_hvm_attr
5189 :Returns: 0 on success, < 0 on error
5191 Allows Xen VM attributes to be read. For the structure and types,
5192 see KVM_XEN_HVM_SET_ATTR above.
5194 4.128 KVM_XEN_VCPU_SET_ATTR
5195 ---------------------------
5197 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5200 :Parameters: struct kvm_xen_vcpu_attr
5201 :Returns: 0 on success, < 0 on error
5205 struct kvm_xen_vcpu_attr {
5213 __u64 state_entry_time;
5215 __u64 time_runnable;
5224 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
5225 Sets the guest physical address of the vcpu_info for a given vCPU.
5227 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
5228 Sets the guest physical address of an additional pvclock structure
5229 for a given vCPU. This is typically used for guest vsyscall support.
5231 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5232 Sets the guest physical address of the vcpu_runstate_info for a given
5233 vCPU. This is how a Xen guest tracks CPU state such as steal time.
5235 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5236 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5237 the given vCPU from the .u.runstate.state member of the structure.
5238 KVM automatically accounts running and runnable time but blocked
5239 and offline states are only entered explicitly.
5241 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5242 Sets all fields of the vCPU runstate data from the .u.runstate member
5243 of the structure, including the current runstate. The state_entry_time
5244 must equal the sum of the other four times.
5246 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5247 This *adds* the contents of the .u.runstate members of the structure
5248 to the corresponding members of the given vCPU's runstate data, thus
5249 permitting atomic adjustments to the runstate times. The adjustment
5250 to the state_entry_time must equal the sum of the adjustments to the
5251 other four times. The state field must be set to -1, or to a valid
5252 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5253 or RUNSTATE_offline) to set the current accounted state as of the
5254 adjusted state_entry_time.
5256 4.129 KVM_XEN_VCPU_GET_ATTR
5257 ---------------------------
5259 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5262 :Parameters: struct kvm_xen_vcpu_attr
5263 :Returns: 0 on success, < 0 on error
5265 Allows Xen vCPU attributes to be read. For the structure and types,
5266 see KVM_XEN_VCPU_SET_ATTR above.
5268 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5269 with the KVM_XEN_VCPU_GET_ATTR ioctl.
5271 4.130 KVM_ARM_MTE_COPY_TAGS
5272 ---------------------------
5274 :Capability: KVM_CAP_ARM_MTE
5275 :Architectures: arm64
5277 :Parameters: struct kvm_arm_copy_mte_tags
5278 :Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5279 arguments, -EFAULT if memory cannot be accessed).
5283 struct kvm_arm_copy_mte_tags {
5291 Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5292 ``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr``
5293 field must point to a buffer which the tags will be copied to or from.
5295 ``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5296 ``KVM_ARM_TAGS_FROM_GUEST``.
5298 The size of the buffer to store the tags is ``(length / 16)`` bytes
5299 (granules in MTE are 16 bytes long). Each byte contains a single tag
5300 value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5301 ``PTRACE_POKEMTETAGS``.
5303 If an error occurs before any data is copied then a negative error code is
5304 returned. If some tags have been copied before an error occurs then the number
5305 of bytes successfully copied is returned. If the call completes successfully
5306 then ``length`` is returned.
5308 4.131 KVM_GET_SREGS2
5309 --------------------
5311 :Capability: KVM_CAP_SREGS2
5314 :Parameters: struct kvm_sregs2 (out)
5315 :Returns: 0 on success, -1 on error
5317 Reads special registers from the vcpu.
5318 This ioctl (when supported) replaces the KVM_GET_SREGS.
5323 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5324 struct kvm_segment cs, ds, es, fs, gs, ss;
5325 struct kvm_segment tr, ldt;
5326 struct kvm_dtable gdt, idt;
5327 __u64 cr0, cr2, cr3, cr4, cr8;
5334 flags values for ``kvm_sregs2``:
5336 ``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5338 Indicates thats the struct contain valid PDPTR values.
5341 4.132 KVM_SET_SREGS2
5342 --------------------
5344 :Capability: KVM_CAP_SREGS2
5347 :Parameters: struct kvm_sregs2 (in)
5348 :Returns: 0 on success, -1 on error
5350 Writes special registers into the vcpu.
5351 See KVM_GET_SREGS2 for the data structures.
5352 This ioctl (when supported) replaces the KVM_SET_SREGS.
5354 4.133 KVM_GET_STATS_FD
5355 ----------------------
5357 :Capability: KVM_CAP_STATS_BINARY_FD
5359 :Type: vm ioctl, vcpu ioctl
5361 :Returns: statistics file descriptor on success, < 0 on error
5365 ====== ======================================================
5366 ENOMEM if the fd could not be created due to lack of memory
5367 EMFILE if the number of opened files exceeds the limit
5368 ====== ======================================================
5370 The returned file descriptor can be used to read VM/vCPU statistics data in
5371 binary format. The data in the file descriptor consists of four blocks
5372 organized as follows:
5384 Apart from the header starting at offset 0, please be aware that it is
5385 not guaranteed that the four blocks are adjacent or in the above order;
5386 the offsets of the id, descriptors and data blocks are found in the
5387 header. However, all four blocks are aligned to 64 bit offsets in the
5388 file and they do not overlap.
5390 All blocks except the data block are immutable. Userspace can read them
5391 only one time after retrieving the file descriptor, and then use ``pread`` or
5392 ``lseek`` to read the statistics repeatedly.
5394 All data is in system endianness.
5396 The format of the header is as follows::
5398 struct kvm_stats_header {
5407 The ``flags`` field is not used at the moment. It is always read as 0.
5409 The ``name_size`` field is the size (in byte) of the statistics name string
5410 (including trailing '\0') which is contained in the "id string" block and
5411 appended at the end of every descriptor.
5413 The ``num_desc`` field is the number of descriptors that are included in the
5414 descriptor block. (The actual number of values in the data block may be
5415 larger, since each descriptor may comprise more than one value).
5417 The ``id_offset`` field is the offset of the id string from the start of the
5418 file indicated by the file descriptor. It is a multiple of 8.
5420 The ``desc_offset`` field is the offset of the Descriptors block from the start
5421 of the file indicated by the file descriptor. It is a multiple of 8.
5423 The ``data_offset`` field is the offset of the Stats Data block from the start
5424 of the file indicated by the file descriptor. It is a multiple of 8.
5426 The id string block contains a string which identifies the file descriptor on
5427 which KVM_GET_STATS_FD was invoked. The size of the block, including the
5428 trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5430 The descriptors block is only needed to be read once for the lifetime of the
5431 file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5432 by a string of size ``name_size``.
5435 #define KVM_STATS_TYPE_SHIFT 0
5436 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT)
5437 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT)
5438 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT)
5439 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT)
5440 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT)
5441 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT)
5442 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST
5444 #define KVM_STATS_UNIT_SHIFT 4
5445 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT)
5446 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT)
5447 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT)
5448 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT)
5449 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT)
5450 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_CYCLES
5452 #define KVM_STATS_BASE_SHIFT 8
5453 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT)
5454 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT)
5455 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT)
5456 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2
5458 struct kvm_stats_desc {
5467 The ``flags`` field contains the type and unit of the statistics data described
5468 by this descriptor. Its endianness is CPU native.
5469 The following flags are supported:
5471 Bits 0-3 of ``flags`` encode the type:
5473 * ``KVM_STATS_TYPE_CUMULATIVE``
5474 The statistics reports a cumulative count. The value of data can only be increased.
5475 Most of the counters used in KVM are of this type.
5476 The corresponding ``size`` field for this type is always 1.
5477 All cumulative statistics data are read/write.
5478 * ``KVM_STATS_TYPE_INSTANT``
5479 The statistics reports an instantaneous value. Its value can be increased or
5480 decreased. This type is usually used as a measurement of some resources,
5481 like the number of dirty pages, the number of large pages, etc.
5482 All instant statistics are read only.
5483 The corresponding ``size`` field for this type is always 1.
5484 * ``KVM_STATS_TYPE_PEAK``
5485 The statistics data reports a peak value, for example the maximum number
5486 of items in a hash table bucket, the longest time waited and so on.
5487 The value of data can only be increased.
5488 The corresponding ``size`` field for this type is always 1.
5489 * ``KVM_STATS_TYPE_LINEAR_HIST``
5490 The statistic is reported as a linear histogram. The number of
5491 buckets is specified by the ``size`` field. The size of buckets is specified
5492 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``)
5493 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last
5494 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity
5495 value.) The bucket value indicates how many samples fell in the bucket's range.
5496 * ``KVM_STATS_TYPE_LOG_HIST``
5497 The statistic is reported as a logarithmic histogram. The number of
5498 buckets is specified by the ``size`` field. The range of the first bucket is
5499 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF).
5500 Otherwise, The Nth bucket (1 < N < ``size``) covers
5501 [pow(2, N-2), pow(2, N-1)). The bucket value indicates how many samples fell
5502 in the bucket's range.
5504 Bits 4-7 of ``flags`` encode the unit:
5506 * ``KVM_STATS_UNIT_NONE``
5507 There is no unit for the value of statistics data. This usually means that
5508 the value is a simple counter of an event.
5509 * ``KVM_STATS_UNIT_BYTES``
5510 It indicates that the statistics data is used to measure memory size, in the
5511 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
5512 determined by the ``exponent`` field in the descriptor.
5513 * ``KVM_STATS_UNIT_SECONDS``
5514 It indicates that the statistics data is used to measure time or latency.
5515 * ``KVM_STATS_UNIT_CYCLES``
5516 It indicates that the statistics data is used to measure CPU clock cycles.
5518 Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
5521 * ``KVM_STATS_BASE_POW10``
5522 The scale is based on power of 10. It is used for measurement of time and
5523 CPU clock cycles. For example, an exponent of -9 can be used with
5524 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
5525 * ``KVM_STATS_BASE_POW2``
5526 The scale is based on power of 2. It is used for measurement of memory size.
5527 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
5528 express that the unit is MiB.
5530 The ``size`` field is the number of values of this statistics data. Its
5531 value is usually 1 for most of simple statistics. 1 means it contains an
5532 unsigned 64bit data.
5534 The ``offset`` field is the offset from the start of Data Block to the start of
5535 the corresponding statistics data.
5537 The ``bucket_size`` field is used as a parameter for histogram statistics data.
5538 It is only used by linear histogram statistics data, specifying the size of a
5541 The ``name`` field is the name string of the statistics data. The name string
5542 starts at the end of ``struct kvm_stats_desc``. The maximum length including
5543 the trailing ``'\0'``, is indicated by ``name_size`` in the header.
5545 The Stats Data block contains an array of 64-bit values in the same order
5546 as the descriptors in Descriptors block.
5548 4.134 KVM_GET_XSAVE2
5549 --------------------
5551 :Capability: KVM_CAP_XSAVE2
5554 :Parameters: struct kvm_xsave (out)
5555 :Returns: 0 on success, -1 on error
5565 This ioctl would copy current vcpu's xsave struct to the userspace. It
5566 copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2)
5567 when invoked on the vm file descriptor. The size value returned by
5568 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
5569 Currently, it is only greater than 4096 if a dynamic feature has been
5570 enabled with ``arch_prctl()``, but this may change in the future.
5572 The offsets of the state save areas in struct kvm_xsave follow the contents
5573 of CPUID leaf 0xD on the host.
5576 5. The kvm_run structure
5577 ========================
5579 Application code obtains a pointer to the kvm_run structure by
5580 mmap()ing a vcpu fd. From that point, application code can control
5581 execution by changing fields in kvm_run prior to calling the KVM_RUN
5582 ioctl, and obtain information about the reason KVM_RUN returned by
5583 looking up structure members.
5589 __u8 request_interrupt_window;
5591 Request that KVM_RUN return when it becomes possible to inject external
5592 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
5596 __u8 immediate_exit;
5598 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
5599 exits immediately, returning -EINTR. In the common scenario where a
5600 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
5601 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
5602 Rather than blocking the signal outside KVM_RUN, userspace can set up
5603 a signal handler that sets run->immediate_exit to a non-zero value.
5605 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
5614 When KVM_RUN has returned successfully (return value 0), this informs
5615 application code why KVM_RUN has returned. Allowable values for this
5616 field are detailed below.
5620 __u8 ready_for_interrupt_injection;
5622 If request_interrupt_window has been specified, this field indicates
5623 an interrupt can be injected now with KVM_INTERRUPT.
5629 The value of the current interrupt flag. Only valid if in-kernel
5630 local APIC is not used.
5636 More architecture-specific flags detailing state of the VCPU that may
5637 affect the device's behavior. Current defined flags::
5639 /* x86, set if the VCPU is in system management mode */
5640 #define KVM_RUN_X86_SMM (1 << 0)
5641 /* x86, set if bus lock detected in VM */
5642 #define KVM_RUN_BUS_LOCK (1 << 1)
5646 /* in (pre_kvm_run), out (post_kvm_run) */
5649 The value of the cr8 register. Only valid if in-kernel local APIC is
5650 not used. Both input and output.
5656 The value of the APIC BASE msr. Only valid if in-kernel local
5657 APIC is not used. Both input and output.
5662 /* KVM_EXIT_UNKNOWN */
5664 __u64 hardware_exit_reason;
5667 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
5668 reasons. Further architecture-specific information is available in
5669 hardware_exit_reason.
5673 /* KVM_EXIT_FAIL_ENTRY */
5675 __u64 hardware_entry_failure_reason;
5676 __u32 cpu; /* if KVM_LAST_CPU */
5679 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
5680 to unknown reasons. Further architecture-specific information is
5681 available in hardware_entry_failure_reason.
5685 /* KVM_EXIT_EXCEPTION */
5697 #define KVM_EXIT_IO_IN 0
5698 #define KVM_EXIT_IO_OUT 1
5700 __u8 size; /* bytes */
5703 __u64 data_offset; /* relative to kvm_run start */
5706 If exit_reason is KVM_EXIT_IO, then the vcpu has
5707 executed a port I/O instruction which could not be satisfied by kvm.
5708 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
5709 where kvm expects application code to place the data for the next
5710 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
5714 /* KVM_EXIT_DEBUG */
5716 struct kvm_debug_exit_arch arch;
5719 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
5720 for which architecture specific information is returned.
5732 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
5733 executed a memory-mapped I/O instruction which could not be satisfied
5734 by kvm. The 'data' member contains the written data if 'is_write' is
5735 true, and should be filled by application code otherwise.
5737 The 'data' member contains, in its first 'len' bytes, the value as it would
5738 appear if the VCPU performed a load or store of the appropriate width directly
5743 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
5744 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5745 operations are complete (and guest state is consistent) only after userspace
5746 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5747 incomplete operations and then check for pending signals.
5749 The pending state of the operation is not preserved in state which is
5750 visible to userspace, thus userspace should ensure that the operation is
5751 completed before performing a live migration. Userspace can re-enter the
5752 guest with an unmasked signal pending or with the immediate_exit field set
5753 to complete pending operations without allowing any further instructions
5758 /* KVM_EXIT_HYPERCALL */
5767 Unused. This was once used for 'hypercall to userspace'. To implement
5768 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5770 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5774 /* KVM_EXIT_TPR_ACCESS */
5781 To be documented (KVM_TPR_ACCESS_REPORTING).
5785 /* KVM_EXIT_S390_SIEIC */
5788 __u64 mask; /* psw upper half */
5789 __u64 addr; /* psw lower half */
5798 /* KVM_EXIT_S390_RESET */
5799 #define KVM_S390_RESET_POR 1
5800 #define KVM_S390_RESET_CLEAR 2
5801 #define KVM_S390_RESET_SUBSYSTEM 4
5802 #define KVM_S390_RESET_CPU_INIT 8
5803 #define KVM_S390_RESET_IPL 16
5804 __u64 s390_reset_flags;
5810 /* KVM_EXIT_S390_UCONTROL */
5812 __u64 trans_exc_code;
5816 s390 specific. A page fault has occurred for a user controlled virtual
5817 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5818 resolved by the kernel.
5819 The program code and the translation exception code that were placed
5820 in the cpu's lowcore are presented here as defined by the z Architecture
5821 Principles of Operation Book in the Chapter for Dynamic Address Translation
5833 Deprecated - was used for 440 KVM.
5842 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5843 hypercalls and exit with this exit struct that contains all the guest gprs.
5845 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5846 Userspace can now handle the hypercall and when it's done modify the gprs as
5847 necessary. Upon guest entry all guest GPRs will then be replaced by the values
5852 /* KVM_EXIT_PAPR_HCALL */
5859 This is used on 64-bit PowerPC when emulating a pSeries partition,
5860 e.g. with the 'pseries' machine type in qemu. It occurs when the
5861 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
5862 contains the hypercall number (from the guest R3), and 'args' contains
5863 the arguments (from the guest R4 - R12). Userspace should put the
5864 return code in 'ret' and any extra returned values in args[].
5865 The possible hypercalls are defined in the Power Architecture Platform
5866 Requirements (PAPR) document available from www.power.org (free
5867 developer registration required to access it).
5871 /* KVM_EXIT_S390_TSCH */
5873 __u16 subchannel_id;
5874 __u16 subchannel_nr;
5881 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5882 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5883 interrupt for the target subchannel has been dequeued and subchannel_id,
5884 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5885 interrupt. ipb is needed for instruction parameter decoding.
5894 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5895 interrupt acknowledge path to the core. When the core successfully
5896 delivers an interrupt, it automatically populates the EPR register with
5897 the interrupt vector number and acknowledges the interrupt inside
5898 the interrupt controller.
5900 In case the interrupt controller lives in user space, we need to do
5901 the interrupt acknowledge cycle through it to fetch the next to be
5902 delivered interrupt vector using this exit.
5904 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5905 external interrupt has just been delivered into the guest. User space
5906 should put the acknowledged interrupt vector into the 'epr' field.
5910 /* KVM_EXIT_SYSTEM_EVENT */
5912 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5913 #define KVM_SYSTEM_EVENT_RESET 2
5914 #define KVM_SYSTEM_EVENT_CRASH 3
5919 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5920 a system-level event using some architecture specific mechanism (hypercall
5921 or some special instruction). In case of ARM/ARM64, this is triggered using
5922 HVC instruction based PSCI call from the vcpu. The 'type' field describes
5923 the system-level event type. The 'flags' field describes architecture
5924 specific flags for the system-level event.
5926 Valid values for 'type' are:
5928 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5929 VM. Userspace is not obliged to honour this, and if it does honour
5930 this does not need to destroy the VM synchronously (ie it may call
5931 KVM_RUN again before shutdown finally occurs).
5932 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5933 As with SHUTDOWN, userspace can choose to ignore the request, or
5934 to schedule the reset to occur in the future and may call KVM_RUN again.
5935 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5936 has requested a crash condition maintenance. Userspace can choose
5937 to ignore the request, or to gather VM memory core dump and/or
5938 reset/shutdown of the VM.
5942 /* KVM_EXIT_IOAPIC_EOI */
5947 Indicates that the VCPU's in-kernel local APIC received an EOI for a
5948 level-triggered IOAPIC interrupt. This exit only triggers when the
5949 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5950 the userspace IOAPIC should process the EOI and retrigger the interrupt if
5951 it is still asserted. Vector is the LAPIC interrupt vector for which the
5956 struct kvm_hyperv_exit {
5957 #define KVM_EXIT_HYPERV_SYNIC 1
5958 #define KVM_EXIT_HYPERV_HCALL 2
5959 #define KVM_EXIT_HYPERV_SYNDBG 3
5986 /* KVM_EXIT_HYPERV */
5987 struct kvm_hyperv_exit hyperv;
5989 Indicates that the VCPU exits into userspace to process some tasks
5990 related to Hyper-V emulation.
5992 Valid values for 'type' are:
5994 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5996 Hyper-V SynIC state change. Notification is used to remap SynIC
5997 event/message pages and to enable/disable SynIC messages/events processing
6000 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
6002 Hyper-V Synthetic debugger state change. Notification is used to either update
6003 the pending_page location or to send a control command (send the buffer located
6004 in send_page or recv a buffer to recv_page).
6008 /* KVM_EXIT_ARM_NISV */
6014 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
6015 KVM will typically return to userspace and ask it to do MMIO emulation on its
6016 behalf. However, for certain classes of instructions, no instruction decode
6017 (direction, length of memory access) is provided, and fetching and decoding
6018 the instruction from the VM is overly complicated to live in the kernel.
6020 Historically, when this situation occurred, KVM would print a warning and kill
6021 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
6022 trying to do I/O, which just couldn't be emulated, and the warning message was
6023 phrased accordingly. However, what happened more often was that a guest bug
6024 caused access outside the guest memory areas which should lead to a more
6025 meaningful warning message and an external abort in the guest, if the access
6026 did not fall within an I/O window.
6028 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
6029 this capability at VM creation. Once this is done, these types of errors will
6030 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
6031 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
6032 in the fault_ipa field. Userspace can either fix up the access if it's
6033 actually an I/O access by decoding the instruction from guest memory (if it's
6034 very brave) and continue executing the guest, or it can decide to suspend,
6035 dump, or restart the guest.
6037 Note that KVM does not skip the faulting instruction as it does for
6038 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
6039 if it decides to decode and emulate the instruction.
6043 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
6045 __u8 error; /* user -> kernel */
6047 __u32 reason; /* kernel -> user */
6048 __u32 index; /* kernel -> user */
6049 __u64 data; /* kernel <-> user */
6052 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
6053 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
6054 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
6057 The "reason" field specifies why the MSR trap occurred. User space will only
6058 receive MSR exit traps when a particular reason was requested during through
6059 ENABLE_CAP. Currently valid exit reasons are:
6061 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
6062 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
6063 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
6065 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
6066 wants to read. To respond to this request with a successful read, user space
6067 writes the respective data into the "data" field and must continue guest
6068 execution to ensure the read data is transferred into guest register state.
6070 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
6071 the "error" field. This will inject a #GP into the guest when the VCPU is
6074 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
6075 wants to write. Once finished processing the event, user space must continue
6076 vCPU execution. If the MSR write was unsuccessful, user space also sets the
6077 "error" field to "1".
6082 struct kvm_xen_exit {
6083 #define KVM_EXIT_XEN_HCALL 1
6096 struct kvm_hyperv_exit xen;
6098 Indicates that the VCPU exits into userspace to process some tasks
6099 related to Xen emulation.
6101 Valid values for 'type' are:
6103 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
6104 Userspace is expected to place the hypercall result into the appropriate
6105 field before invoking KVM_RUN again.
6109 /* KVM_EXIT_RISCV_SBI */
6111 unsigned long extension_id;
6112 unsigned long function_id;
6113 unsigned long args[6];
6114 unsigned long ret[2];
6116 If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has
6117 done a SBI call which is not handled by KVM RISC-V kernel module. The details
6118 of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The
6119 'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the
6120 'function_id' field represents function ID of given SBI extension. The 'args'
6121 array field of 'riscv_sbi' represents parameters for the SBI call and 'ret'
6122 array field represents return values. The userspace should update the return
6123 values of SBI call before resuming the VCPU. For more details on RISC-V SBI
6124 spec refer, https://github.com/riscv/riscv-sbi-doc.
6128 /* Fix the size of the union. */
6133 * shared registers between kvm and userspace.
6134 * kvm_valid_regs specifies the register classes set by the host
6135 * kvm_dirty_regs specified the register classes dirtied by userspace
6136 * struct kvm_sync_regs is architecture specific, as well as the
6137 * bits for kvm_valid_regs and kvm_dirty_regs
6139 __u64 kvm_valid_regs;
6140 __u64 kvm_dirty_regs;
6142 struct kvm_sync_regs regs;
6143 char padding[SYNC_REGS_SIZE_BYTES];
6146 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
6147 certain guest registers without having to call SET/GET_*REGS. Thus we can
6148 avoid some system call overhead if userspace has to handle the exit.
6149 Userspace can query the validity of the structure by checking
6150 kvm_valid_regs for specific bits. These bits are architecture specific
6151 and usually define the validity of a groups of registers. (e.g. one bit
6152 for general purpose registers)
6154 Please note that the kernel is allowed to use the kvm_run structure as the
6155 primary storage for certain register types. Therefore, the kernel may use the
6156 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
6164 6. Capabilities that can be enabled on vCPUs
6165 ============================================
6167 There are certain capabilities that change the behavior of the virtual CPU or
6168 the virtual machine when enabled. To enable them, please see section 4.37.
6169 Below you can find a list of capabilities and what their effect on the vCPU or
6170 the virtual machine is when enabling them.
6172 The following information is provided along with the description:
6175 which instruction set architectures provide this ioctl.
6176 x86 includes both i386 and x86_64.
6179 whether this is a per-vcpu or per-vm capability.
6182 what parameters are accepted by the capability.
6185 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6186 are not detailed, but errors with specific meanings are.
6195 :Returns: 0 on success; -1 on error
6197 This capability enables interception of OSI hypercalls that otherwise would
6198 be treated as normal system calls to be injected into the guest. OSI hypercalls
6199 were invented by Mac-on-Linux to have a standardized communication mechanism
6200 between the guest and the host.
6202 When this capability is enabled, KVM_EXIT_OSI can occur.
6205 6.2 KVM_CAP_PPC_PAPR
6206 --------------------
6211 :Returns: 0 on success; -1 on error
6213 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
6214 done using the hypercall instruction "sc 1".
6216 It also sets the guest privilege level to "supervisor" mode. Usually the guest
6217 runs in "hypervisor" privilege mode with a few missing features.
6219 In addition to the above, it changes the semantics of SDR1. In this mode, the
6220 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
6221 HTAB invisible to the guest.
6223 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
6231 :Parameters: args[0] is the address of a struct kvm_config_tlb
6232 :Returns: 0 on success; -1 on error
6236 struct kvm_config_tlb {
6243 Configures the virtual CPU's TLB array, establishing a shared memory area
6244 between userspace and KVM. The "params" and "array" fields are userspace
6245 addresses of mmu-type-specific data structures. The "array_len" field is an
6246 safety mechanism, and should be set to the size in bytes of the memory that
6247 userspace has reserved for the array. It must be at least the size dictated
6248 by "mmu_type" and "params".
6250 While KVM_RUN is active, the shared region is under control of KVM. Its
6251 contents are undefined, and any modification by userspace results in
6252 boundedly undefined behavior.
6254 On return from KVM_RUN, the shared region will reflect the current state of
6255 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
6256 to tell KVM which entries have been changed, prior to calling KVM_RUN again
6259 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
6261 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
6262 - The "array" field points to an array of type "struct
6263 kvm_book3e_206_tlb_entry".
6264 - The array consists of all entries in the first TLB, followed by all
6265 entries in the second TLB.
6266 - Within a TLB, entries are ordered first by increasing set number. Within a
6267 set, entries are ordered by way (increasing ESEL).
6268 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
6269 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
6270 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
6271 hardware ignores this value for TLB0.
6273 6.4 KVM_CAP_S390_CSS_SUPPORT
6274 ----------------------------
6276 :Architectures: s390
6279 :Returns: 0 on success; -1 on error
6281 This capability enables support for handling of channel I/O instructions.
6283 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
6284 handled in-kernel, while the other I/O instructions are passed to userspace.
6286 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
6287 SUBCHANNEL intercepts.
6289 Note that even though this capability is enabled per-vcpu, the complete
6290 virtual machine is affected.
6297 :Parameters: args[0] defines whether the proxy facility is active
6298 :Returns: 0 on success; -1 on error
6300 This capability enables or disables the delivery of interrupts through the
6301 external proxy facility.
6303 When enabled (args[0] != 0), every time the guest gets an external interrupt
6304 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
6305 to receive the topmost interrupt vector.
6307 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
6309 When this capability is enabled, KVM_EXIT_EPR can occur.
6311 6.6 KVM_CAP_IRQ_MPIC
6312 --------------------
6315 :Parameters: args[0] is the MPIC device fd;
6316 args[1] is the MPIC CPU number for this vcpu
6318 This capability connects the vcpu to an in-kernel MPIC device.
6320 6.7 KVM_CAP_IRQ_XICS
6321 --------------------
6325 :Parameters: args[0] is the XICS device fd;
6326 args[1] is the XICS CPU number (server ID) for this vcpu
6328 This capability connects the vcpu to an in-kernel XICS device.
6330 6.8 KVM_CAP_S390_IRQCHIP
6331 ------------------------
6333 :Architectures: s390
6337 This capability enables the in-kernel irqchip for s390. Please refer to
6338 "4.24 KVM_CREATE_IRQCHIP" for details.
6340 6.9 KVM_CAP_MIPS_FPU
6341 --------------------
6343 :Architectures: mips
6345 :Parameters: args[0] is reserved for future use (should be 0).
6347 This capability allows the use of the host Floating Point Unit by the guest. It
6348 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
6349 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
6350 accessed (depending on the current guest FPU register mode), and the Status.FR,
6351 Config5.FRE bits are accessible via the KVM API and also from the guest,
6352 depending on them being supported by the FPU.
6354 6.10 KVM_CAP_MIPS_MSA
6355 ---------------------
6357 :Architectures: mips
6359 :Parameters: args[0] is reserved for future use (should be 0).
6361 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
6362 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
6363 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
6364 registers can be accessed, and the Config5.MSAEn bit is accessible via the
6365 KVM API and also from the guest.
6367 6.74 KVM_CAP_SYNC_REGS
6368 ----------------------
6370 :Architectures: s390, x86
6371 :Target: s390: always enabled, x86: vcpu
6373 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
6375 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
6377 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
6378 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
6379 without having to call SET/GET_*REGS". This reduces overhead by eliminating
6380 repeated ioctl calls for setting and/or getting register values. This is
6381 particularly important when userspace is making synchronous guest state
6382 modifications, e.g. when emulating and/or intercepting instructions in
6385 For s390 specifics, please refer to the source code.
6389 - the register sets to be copied out to kvm_run are selectable
6390 by userspace (rather that all sets being copied out for every exit).
6391 - vcpu_events are available in addition to regs and sregs.
6393 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
6394 function as an input bit-array field set by userspace to indicate the
6395 specific register sets to be copied out on the next exit.
6397 To indicate when userspace has modified values that should be copied into
6398 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
6399 This is done using the same bitflags as for the 'kvm_valid_regs' field.
6400 If the dirty bit is not set, then the register set values will not be copied
6401 into the vCPU even if they've been modified.
6403 Unused bitfields in the bitarrays must be set to zero.
6407 struct kvm_sync_regs {
6408 struct kvm_regs regs;
6409 struct kvm_sregs sregs;
6410 struct kvm_vcpu_events events;
6413 6.75 KVM_CAP_PPC_IRQ_XIVE
6414 -------------------------
6418 :Parameters: args[0] is the XIVE device fd;
6419 args[1] is the XIVE CPU number (server ID) for this vcpu
6421 This capability connects the vcpu to an in-kernel XIVE device.
6423 7. Capabilities that can be enabled on VMs
6424 ==========================================
6426 There are certain capabilities that change the behavior of the virtual
6427 machine when enabled. To enable them, please see section 4.37. Below
6428 you can find a list of capabilities and what their effect on the VM
6429 is when enabling them.
6431 The following information is provided along with the description:
6434 which instruction set architectures provide this ioctl.
6435 x86 includes both i386 and x86_64.
6438 what parameters are accepted by the capability.
6441 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6442 are not detailed, but errors with specific meanings are.
6445 7.1 KVM_CAP_PPC_ENABLE_HCALL
6446 ----------------------------
6449 :Parameters: args[0] is the sPAPR hcall number;
6450 args[1] is 0 to disable, 1 to enable in-kernel handling
6452 This capability controls whether individual sPAPR hypercalls (hcalls)
6453 get handled by the kernel or not. Enabling or disabling in-kernel
6454 handling of an hcall is effective across the VM. On creation, an
6455 initial set of hcalls are enabled for in-kernel handling, which
6456 consists of those hcalls for which in-kernel handlers were implemented
6457 before this capability was implemented. If disabled, the kernel will
6458 not to attempt to handle the hcall, but will always exit to userspace
6459 to handle it. Note that it may not make sense to enable some and
6460 disable others of a group of related hcalls, but KVM does not prevent
6461 userspace from doing that.
6463 If the hcall number specified is not one that has an in-kernel
6464 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
6467 7.2 KVM_CAP_S390_USER_SIGP
6468 --------------------------
6470 :Architectures: s390
6473 This capability controls which SIGP orders will be handled completely in user
6474 space. With this capability enabled, all fast orders will be handled completely
6481 - CONDITIONAL EMERGENCY SIGNAL
6483 All other orders will be handled completely in user space.
6485 Only privileged operation exceptions will be checked for in the kernel (or even
6486 in the hardware prior to interception). If this capability is not enabled, the
6487 old way of handling SIGP orders is used (partially in kernel and user space).
6489 7.3 KVM_CAP_S390_VECTOR_REGISTERS
6490 ---------------------------------
6492 :Architectures: s390
6494 :Returns: 0 on success, negative value on error
6496 Allows use of the vector registers introduced with z13 processor, and
6497 provides for the synchronization between host and user space. Will
6498 return -EINVAL if the machine does not support vectors.
6500 7.4 KVM_CAP_S390_USER_STSI
6501 --------------------------
6503 :Architectures: s390
6506 This capability allows post-handlers for the STSI instruction. After
6507 initial handling in the kernel, KVM exits to user space with
6508 KVM_EXIT_S390_STSI to allow user space to insert further data.
6510 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
6522 @addr - guest address of STSI SYSIB
6526 @ar - access register number
6528 KVM handlers should exit to userspace with rc = -EREMOTE.
6530 7.5 KVM_CAP_SPLIT_IRQCHIP
6531 -------------------------
6534 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
6535 :Returns: 0 on success, -1 on error
6537 Create a local apic for each processor in the kernel. This can be used
6538 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
6539 IOAPIC and PIC (and also the PIT, even though this has to be enabled
6542 This capability also enables in kernel routing of interrupt requests;
6543 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
6544 used in the IRQ routing table. The first args[0] MSI routes are reserved
6545 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
6546 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
6548 Fails if VCPU has already been created, or if the irqchip is already in the
6549 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
6554 :Architectures: s390
6557 Allows use of runtime-instrumentation introduced with zEC12 processor.
6558 Will return -EINVAL if the machine does not support runtime-instrumentation.
6559 Will return -EBUSY if a VCPU has already been created.
6561 7.7 KVM_CAP_X2APIC_API
6562 ----------------------
6565 :Parameters: args[0] - features that should be enabled
6566 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
6568 Valid feature flags in args[0] are::
6570 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
6571 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
6573 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
6574 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
6575 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
6576 respective sections.
6578 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
6579 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
6580 as a broadcast even in x2APIC mode in order to support physical x2APIC
6581 without interrupt remapping. This is undesirable in logical mode,
6582 where 0xff represents CPUs 0-7 in cluster 0.
6584 7.8 KVM_CAP_S390_USER_INSTR0
6585 ----------------------------
6587 :Architectures: s390
6590 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
6591 be intercepted and forwarded to user space. User space can use this
6592 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
6593 not inject an operating exception for these instructions, user space has
6594 to take care of that.
6596 This capability can be enabled dynamically even if VCPUs were already
6597 created and are running.
6602 :Architectures: s390
6604 :Returns: 0 on success; -EINVAL if the machine does not support
6605 guarded storage; -EBUSY if a VCPU has already been created.
6607 Allows use of guarded storage for the KVM guest.
6609 7.10 KVM_CAP_S390_AIS
6610 ---------------------
6612 :Architectures: s390
6615 Allow use of adapter-interruption suppression.
6616 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
6618 7.11 KVM_CAP_PPC_SMT
6619 --------------------
6622 :Parameters: vsmt_mode, flags
6624 Enabling this capability on a VM provides userspace with a way to set
6625 the desired virtual SMT mode (i.e. the number of virtual CPUs per
6626 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
6627 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
6628 the number of threads per subcore for the host. Currently flags must
6629 be 0. A successful call to enable this capability will result in
6630 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
6631 subsequently queried for the VM. This capability is only supported by
6632 HV KVM, and can only be set before any VCPUs have been created.
6633 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
6634 modes are available.
6636 7.12 KVM_CAP_PPC_FWNMI
6637 ----------------------
6642 With this capability a machine check exception in the guest address
6643 space will cause KVM to exit the guest with NMI exit reason. This
6644 enables QEMU to build error log and branch to guest kernel registered
6645 machine check handling routine. Without this capability KVM will
6646 branch to guests' 0x200 interrupt vector.
6648 7.13 KVM_CAP_X86_DISABLE_EXITS
6649 ------------------------------
6652 :Parameters: args[0] defines which exits are disabled
6653 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
6655 Valid bits in args[0] are::
6657 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
6658 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
6659 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
6660 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
6662 Enabling this capability on a VM provides userspace with a way to no
6663 longer intercept some instructions for improved latency in some
6664 workloads, and is suggested when vCPUs are associated to dedicated
6665 physical CPUs. More bits can be added in the future; userspace can
6666 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
6669 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
6671 7.14 KVM_CAP_S390_HPAGE_1M
6672 --------------------------
6674 :Architectures: s390
6676 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
6677 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
6680 With this capability the KVM support for memory backing with 1m pages
6681 through hugetlbfs can be enabled for a VM. After the capability is
6682 enabled, cmma can't be enabled anymore and pfmfi and the storage key
6683 interpretation are disabled. If cmma has already been enabled or the
6684 hpage module parameter is not set to 1, -EINVAL is returned.
6686 While it is generally possible to create a huge page backed VM without
6687 this capability, the VM will not be able to run.
6689 7.15 KVM_CAP_MSR_PLATFORM_INFO
6690 ------------------------------
6693 :Parameters: args[0] whether feature should be enabled or not
6695 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
6696 a #GP would be raised when the guest tries to access. Currently, this
6697 capability does not enable write permissions of this MSR for the guest.
6699 7.16 KVM_CAP_PPC_NESTED_HV
6700 --------------------------
6704 :Returns: 0 on success, -EINVAL when the implementation doesn't support
6705 nested-HV virtualization.
6707 HV-KVM on POWER9 and later systems allows for "nested-HV"
6708 virtualization, which provides a way for a guest VM to run guests that
6709 can run using the CPU's supervisor mode (privileged non-hypervisor
6710 state). Enabling this capability on a VM depends on the CPU having
6711 the necessary functionality and on the facility being enabled with a
6712 kvm-hv module parameter.
6714 7.17 KVM_CAP_EXCEPTION_PAYLOAD
6715 ------------------------------
6718 :Parameters: args[0] whether feature should be enabled or not
6720 With this capability enabled, CR2 will not be modified prior to the
6721 emulated VM-exit when L1 intercepts a #PF exception that occurs in
6722 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
6723 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
6724 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
6725 #DB) exception for L2, exception.has_payload will be set and the
6726 faulting address (or the new DR6 bits*) will be reported in the
6727 exception_payload field. Similarly, when userspace injects a #PF (or
6728 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
6729 exception.has_payload and to put the faulting address - or the new DR6
6730 bits\ [#]_ - in the exception_payload field.
6732 This capability also enables exception.pending in struct
6733 kvm_vcpu_events, which allows userspace to distinguish between pending
6734 and injected exceptions.
6737 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
6740 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
6742 :Architectures: x86, arm, arm64, mips
6743 :Parameters: args[0] whether feature should be enabled or not
6747 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
6748 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
6750 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
6751 automatically clear and write-protect all pages that are returned as dirty.
6752 Rather, userspace will have to do this operation separately using
6753 KVM_CLEAR_DIRTY_LOG.
6755 At the cost of a slightly more complicated operation, this provides better
6756 scalability and responsiveness for two reasons. First,
6757 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
6758 than requiring to sync a full memslot; this ensures that KVM does not
6759 take spinlocks for an extended period of time. Second, in some cases a
6760 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
6761 userspace actually using the data in the page. Pages can be modified
6762 during this time, which is inefficient for both the guest and userspace:
6763 the guest will incur a higher penalty due to write protection faults,
6764 while userspace can see false reports of dirty pages. Manual reprotection
6765 helps reducing this time, improving guest performance and reducing the
6766 number of dirty log false positives.
6768 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
6769 will be initialized to 1 when created. This also improves performance because
6770 dirty logging can be enabled gradually in small chunks on the first call
6771 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
6772 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
6773 x86 and arm64 for now).
6775 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
6776 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
6777 it hard or impossible to use it correctly. The availability of
6778 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
6779 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
6781 7.19 KVM_CAP_PPC_SECURE_GUEST
6782 ------------------------------
6786 This capability indicates that KVM is running on a host that has
6787 ultravisor firmware and thus can support a secure guest. On such a
6788 system, a guest can ask the ultravisor to make it a secure guest,
6789 one whose memory is inaccessible to the host except for pages which
6790 are explicitly requested to be shared with the host. The ultravisor
6791 notifies KVM when a guest requests to become a secure guest, and KVM
6792 has the opportunity to veto the transition.
6794 If present, this capability can be enabled for a VM, meaning that KVM
6795 will allow the transition to secure guest mode. Otherwise KVM will
6796 veto the transition.
6798 7.20 KVM_CAP_HALT_POLL
6799 ----------------------
6803 :Parameters: args[0] is the maximum poll time in nanoseconds
6804 :Returns: 0 on success; -1 on error
6806 This capability overrides the kvm module parameter halt_poll_ns for the
6809 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6810 scheduling during guest halts. The maximum time a VCPU can spend polling is
6811 controlled by the kvm module parameter halt_poll_ns. This capability allows
6812 the maximum halt time to specified on a per-VM basis, effectively overriding
6813 the module parameter for the target VM.
6815 7.21 KVM_CAP_X86_USER_SPACE_MSR
6816 -------------------------------
6820 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6821 :Returns: 0 on success; -1 on error
6823 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6826 When a guest requests to read or write an MSR, KVM may not implement all MSRs
6827 that are relevant to a respective system. It also does not differentiate by
6830 To allow more fine grained control over MSR handling, user space may enable
6831 this capability. With it enabled, MSR accesses that match the mask specified in
6832 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6833 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6834 can then handle to implement model specific MSR handling and/or user notifications
6835 to inform a user that an MSR was not handled.
6837 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
6838 -------------------------------
6842 :Parameters: args[0] defines the policy used when bus locks detected in guest
6843 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
6845 Valid bits in args[0] are::
6847 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
6848 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
6850 Enabling this capability on a VM provides userspace with a way to select
6851 a policy to handle the bus locks detected in guest. Userspace can obtain
6852 the supported modes from the result of KVM_CHECK_EXTENSION and define it
6853 through the KVM_ENABLE_CAP.
6855 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
6856 currently and mutually exclusive with each other. More bits can be added in
6859 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
6860 so that no additional actions are needed. This is the default mode.
6862 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
6863 in VM. KVM just exits to userspace when handling them. Userspace can enforce
6864 its own throttling or other policy based mitigations.
6866 This capability is aimed to address the thread that VM can exploit bus locks to
6867 degree the performance of the whole system. Once the userspace enable this
6868 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
6869 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
6870 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
6871 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
6872 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
6874 7.23 KVM_CAP_PPC_DAWR1
6875 ----------------------
6879 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
6881 This capability can be used to check / enable 2nd DAWR feature provided
6882 by POWER10 processor.
6885 7.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
6886 -------------------------------------
6888 Architectures: x86 SEV enabled
6890 Parameters: args[0] is the fd of the source vm
6891 Returns: 0 on success; ENOTTY on error
6893 This capability enables userspace to copy encryption context from the vm
6894 indicated by the fd to the vm this is called on.
6896 This is intended to support in-guest workloads scheduled by the host. This
6897 allows the in-guest workload to maintain its own NPTs and keeps the two vms
6898 from accidentally clobbering each other with interrupts and the like (separate
6901 7.25 KVM_CAP_SGX_ATTRIBUTE
6902 --------------------------
6906 :Parameters: args[0] is a file handle of a SGX attribute file in securityfs
6907 :Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
6908 attribute is not supported by KVM.
6910 KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
6911 more priveleged enclave attributes. args[0] must hold a file handle to a valid
6912 SGX attribute file corresponding to an attribute that is supported/restricted
6913 by KVM (currently only PROVISIONKEY).
6915 The SGX subsystem restricts access to a subset of enclave attributes to provide
6916 additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
6917 is restricted to deter malware from using the PROVISIONKEY to obtain a stable
6918 system fingerprint. To prevent userspace from circumventing such restrictions
6919 by running an enclave in a VM, KVM prevents access to privileged attributes by
6922 See Documentation/x86/sgx.rst for more details.
6924 7.26 KVM_CAP_PPC_RPT_INVALIDATE
6925 -------------------------------
6927 :Capability: KVM_CAP_PPC_RPT_INVALIDATE
6931 This capability indicates that the kernel is capable of handling
6932 H_RPT_INVALIDATE hcall.
6934 In order to enable the use of H_RPT_INVALIDATE in the guest,
6935 user space might have to advertise it for the guest. For example,
6936 IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
6937 present in the "ibm,hypertas-functions" device-tree property.
6939 This capability is enabled for hypervisors on platforms like POWER9
6940 that support radix MMU.
6942 7.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
6943 --------------------------------------
6946 :Parameters: args[0] whether the feature should be enabled or not
6948 When this capability is enabled, an emulation failure will result in an exit
6949 to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
6950 to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
6951 to 15 instruction bytes for any exit to userspace resulting from an emulation
6952 failure. When these exits to userspace occur use the emulation_failure struct
6953 instead of the internal struct. They both have the same layout, but the
6954 emulation_failure struct matches the content better. It also explicitly
6955 defines the 'flags' field which is used to describe the fields in the struct
6956 that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
6957 set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
6960 7.28 KVM_CAP_ARM_MTE
6961 --------------------
6963 :Architectures: arm64
6966 This capability indicates that KVM (and the hardware) supports exposing the
6967 Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
6968 VMM before creating any VCPUs to allow the guest access. Note that MTE is only
6969 available to a guest running in AArch64 mode and enabling this capability will
6970 cause attempts to create AArch32 VCPUs to fail.
6972 When enabled the guest is able to access tags associated with any memory given
6973 to the guest. KVM will ensure that the tags are maintained during swap or
6974 hibernation of the host; however the VMM needs to manually save/restore the
6975 tags as appropriate if the VM is migrated.
6977 When this capability is enabled all memory in memslots must be mapped as
6978 not-shareable (no MAP_SHARED), attempts to create a memslot with a
6979 MAP_SHARED mmap will result in an -EINVAL return.
6981 When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
6982 perform a bulk copy of tags to/from the guest.
6984 7.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM
6985 -------------------------------------
6987 Architectures: x86 SEV enabled
6989 Parameters: args[0] is the fd of the source vm
6990 Returns: 0 on success
6992 This capability enables userspace to migrate the encryption context from the VM
6993 indicated by the fd to the VM this is called on.
6995 This is intended to support intra-host migration of VMs between userspace VMMs,
6996 upgrading the VMM process without interrupting the guest.
6998 8. Other capabilities.
6999 ======================
7001 This section lists capabilities that give information about other
7002 features of the KVM implementation.
7004 8.1 KVM_CAP_PPC_HWRNG
7005 ---------------------
7009 This capability, if KVM_CHECK_EXTENSION indicates that it is
7010 available, means that the kernel has an implementation of the
7011 H_RANDOM hypercall backed by a hardware random-number generator.
7012 If present, the kernel H_RANDOM handler can be enabled for guest use
7013 with the KVM_CAP_PPC_ENABLE_HCALL capability.
7015 8.2 KVM_CAP_HYPERV_SYNIC
7016 ------------------------
7020 This capability, if KVM_CHECK_EXTENSION indicates that it is
7021 available, means that the kernel has an implementation of the
7022 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
7023 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
7025 In order to use SynIC, it has to be activated by setting this
7026 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
7027 will disable the use of APIC hardware virtualization even if supported
7028 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
7030 8.3 KVM_CAP_PPC_RADIX_MMU
7031 -------------------------
7035 This capability, if KVM_CHECK_EXTENSION indicates that it is
7036 available, means that the kernel can support guests using the
7037 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
7040 8.4 KVM_CAP_PPC_HASH_MMU_V3
7041 ---------------------------
7045 This capability, if KVM_CHECK_EXTENSION indicates that it is
7046 available, means that the kernel can support guests using the
7047 hashed page table MMU defined in Power ISA V3.00 (as implemented in
7048 the POWER9 processor), including in-memory segment tables.
7053 :Architectures: mips
7055 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7056 it is available, means that full hardware assisted virtualization capabilities
7057 of the hardware are available for use through KVM. An appropriate
7058 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
7061 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7062 available, it means that the VM is using full hardware assisted virtualization
7063 capabilities of the hardware. This is useful to check after creating a VM with
7064 KVM_VM_MIPS_DEFAULT.
7066 The value returned by KVM_CHECK_EXTENSION should be compared against known
7067 values (see below). All other values are reserved. This is to allow for the
7068 possibility of other hardware assisted virtualization implementations which
7069 may be incompatible with the MIPS VZ ASE.
7071 == ==========================================================================
7072 0 The trap & emulate implementation is in use to run guest code in user
7073 mode. Guest virtual memory segments are rearranged to fit the guest in the
7074 user mode address space.
7076 1 The MIPS VZ ASE is in use, providing full hardware assisted
7077 virtualization, including standard guest virtual memory segments.
7078 == ==========================================================================
7083 :Architectures: mips
7085 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7086 it is available, means that the trap & emulate implementation is available to
7087 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
7088 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
7089 to KVM_CREATE_VM to create a VM which utilises it.
7091 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7092 available, it means that the VM is using trap & emulate.
7094 8.7 KVM_CAP_MIPS_64BIT
7095 ----------------------
7097 :Architectures: mips
7099 This capability indicates the supported architecture type of the guest, i.e. the
7100 supported register and address width.
7102 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
7103 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
7104 be checked specifically against known values (see below). All other values are
7107 == ========================================================================
7108 0 MIPS32 or microMIPS32.
7109 Both registers and addresses are 32-bits wide.
7110 It will only be possible to run 32-bit guest code.
7112 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
7113 Registers are 64-bits wide, but addresses are 32-bits wide.
7114 64-bit guest code may run but cannot access MIPS64 memory segments.
7115 It will also be possible to run 32-bit guest code.
7117 2 MIPS64 or microMIPS64 with access to all address segments.
7118 Both registers and addresses are 64-bits wide.
7119 It will be possible to run 64-bit or 32-bit guest code.
7120 == ========================================================================
7122 8.9 KVM_CAP_ARM_USER_IRQ
7123 ------------------------
7125 :Architectures: arm, arm64
7127 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
7128 that if userspace creates a VM without an in-kernel interrupt controller, it
7129 will be notified of changes to the output level of in-kernel emulated devices,
7130 which can generate virtual interrupts, presented to the VM.
7131 For such VMs, on every return to userspace, the kernel
7132 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
7133 output level of the device.
7135 Whenever kvm detects a change in the device output level, kvm guarantees at
7136 least one return to userspace before running the VM. This exit could either
7137 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
7138 userspace can always sample the device output level and re-compute the state of
7139 the userspace interrupt controller. Userspace should always check the state
7140 of run->s.regs.device_irq_level on every kvm exit.
7141 The value in run->s.regs.device_irq_level can represent both level and edge
7142 triggered interrupt signals, depending on the device. Edge triggered interrupt
7143 signals will exit to userspace with the bit in run->s.regs.device_irq_level
7144 set exactly once per edge signal.
7146 The field run->s.regs.device_irq_level is available independent of
7147 run->kvm_valid_regs or run->kvm_dirty_regs bits.
7149 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
7150 number larger than 0 indicating the version of this capability is implemented
7151 and thereby which bits in run->s.regs.device_irq_level can signal values.
7153 Currently the following bits are defined for the device_irq_level bitmap::
7155 KVM_CAP_ARM_USER_IRQ >= 1:
7157 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
7158 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
7159 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
7161 Future versions of kvm may implement additional events. These will get
7162 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
7165 8.10 KVM_CAP_PPC_SMT_POSSIBLE
7166 -----------------------------
7170 Querying this capability returns a bitmap indicating the possible
7171 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
7172 (counting from the right) is set, then a virtual SMT mode of 2^N is
7175 8.11 KVM_CAP_HYPERV_SYNIC2
7176 --------------------------
7180 This capability enables a newer version of Hyper-V Synthetic interrupt
7181 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
7182 doesn't clear SynIC message and event flags pages when they are enabled by
7183 writing to the respective MSRs.
7185 8.12 KVM_CAP_HYPERV_VP_INDEX
7186 ----------------------------
7190 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
7191 value is used to denote the target vcpu for a SynIC interrupt. For
7192 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
7193 capability is absent, userspace can still query this msr's value.
7195 8.13 KVM_CAP_S390_AIS_MIGRATION
7196 -------------------------------
7198 :Architectures: s390
7201 This capability indicates if the flic device will be able to get/set the
7202 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
7203 to discover this without having to create a flic device.
7205 8.14 KVM_CAP_S390_PSW
7206 ---------------------
7208 :Architectures: s390
7210 This capability indicates that the PSW is exposed via the kvm_run structure.
7212 8.15 KVM_CAP_S390_GMAP
7213 ----------------------
7215 :Architectures: s390
7217 This capability indicates that the user space memory used as guest mapping can
7218 be anywhere in the user memory address space, as long as the memory slots are
7219 aligned and sized to a segment (1MB) boundary.
7221 8.16 KVM_CAP_S390_COW
7222 ---------------------
7224 :Architectures: s390
7226 This capability indicates that the user space memory used as guest mapping can
7227 use copy-on-write semantics as well as dirty pages tracking via read-only page
7230 8.17 KVM_CAP_S390_BPB
7231 ---------------------
7233 :Architectures: s390
7235 This capability indicates that kvm will implement the interfaces to handle
7236 reset, migration and nested KVM for branch prediction blocking. The stfle
7237 facility 82 should not be provided to the guest without this capability.
7239 8.18 KVM_CAP_HYPERV_TLBFLUSH
7240 ----------------------------
7244 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
7246 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
7247 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
7249 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
7250 ----------------------------------
7252 :Architectures: arm, arm64
7254 This capability indicates that userspace can specify (via the
7255 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
7256 takes a virtual SError interrupt exception.
7257 If KVM advertises this capability, userspace can only specify the ISS field for
7258 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
7259 CPU when the exception is taken. If this virtual SError is taken to EL1 using
7260 AArch64, this value will be reported in the ISS field of ESR_ELx.
7262 See KVM_CAP_VCPU_EVENTS for more details.
7264 8.20 KVM_CAP_HYPERV_SEND_IPI
7265 ----------------------------
7269 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
7271 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
7273 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
7274 -----------------------------------
7278 This capability indicates that KVM running on top of Hyper-V hypervisor
7279 enables Direct TLB flush for its guests meaning that TLB flush
7280 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
7281 Due to the different ABI for hypercall parameters between Hyper-V and
7282 KVM, enabling this capability effectively disables all hypercall
7283 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
7284 flush hypercalls by Hyper-V) so userspace should disable KVM identification
7285 in CPUID and only exposes Hyper-V identification. In this case, guest
7286 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
7288 8.22 KVM_CAP_S390_VCPU_RESETS
7289 -----------------------------
7291 :Architectures: s390
7293 This capability indicates that the KVM_S390_NORMAL_RESET and
7294 KVM_S390_CLEAR_RESET ioctls are available.
7296 8.23 KVM_CAP_S390_PROTECTED
7297 ---------------------------
7299 :Architectures: s390
7301 This capability indicates that the Ultravisor has been initialized and
7302 KVM can therefore start protected VMs.
7303 This capability governs the KVM_S390_PV_COMMAND ioctl and the
7304 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
7305 guests when the state change is invalid.
7307 8.24 KVM_CAP_STEAL_TIME
7308 -----------------------
7310 :Architectures: arm64, x86
7312 This capability indicates that KVM supports steal time accounting.
7313 When steal time accounting is supported it may be enabled with
7314 architecture-specific interfaces. This capability and the architecture-
7315 specific interfaces must be consistent, i.e. if one says the feature
7316 is supported, than the other should as well and vice versa. For arm64
7317 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
7318 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
7320 8.25 KVM_CAP_S390_DIAG318
7321 -------------------------
7323 :Architectures: s390
7325 This capability enables a guest to set information about its control program
7326 (i.e. guest kernel type and version). The information is helpful during
7327 system/firmware service events, providing additional data about the guest
7328 environments running on the machine.
7330 The information is associated with the DIAGNOSE 0x318 instruction, which sets
7331 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
7332 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
7333 environment the control program is running in (e.g. Linux, z/VM...), and the
7334 CPVC is used for information specific to OS (e.g. Linux version, Linux
7337 If this capability is available, then the CPNC and CPVC can be synchronized
7338 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
7340 8.26 KVM_CAP_X86_USER_SPACE_MSR
7341 -------------------------------
7345 This capability indicates that KVM supports deflection of MSR reads and
7346 writes to user space. It can be enabled on a VM level. If enabled, MSR
7347 accesses that would usually trigger a #GP by KVM into the guest will
7348 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
7349 KVM_EXIT_X86_WRMSR exit notifications.
7351 8.27 KVM_CAP_X86_MSR_FILTER
7352 ---------------------------
7356 This capability indicates that KVM supports that accesses to user defined MSRs
7357 may be rejected. With this capability exposed, KVM exports new VM ioctl
7358 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
7359 ranges that KVM should reject access to.
7361 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
7362 trap and emulate MSRs that are outside of the scope of KVM as well as
7363 limit the attack surface on KVM's MSR emulation code.
7365 8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
7366 -------------------------------------
7370 When enabled, KVM will disable paravirtual features provided to the
7371 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
7372 (0x40000001). Otherwise, a guest may use the paravirtual features
7373 regardless of what has actually been exposed through the CPUID leaf.
7375 8.29 KVM_CAP_DIRTY_LOG_RING
7376 ---------------------------
7379 :Parameters: args[0] - size of the dirty log ring
7381 KVM is capable of tracking dirty memory using ring buffers that are
7382 mmaped into userspace; there is one dirty ring per vcpu.
7384 The dirty ring is available to userspace as an array of
7385 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
7387 struct kvm_dirty_gfn {
7389 __u32 slot; /* as_id | slot_id */
7393 The following values are defined for the flags field to define the
7394 current state of the entry::
7396 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
7397 #define KVM_DIRTY_GFN_F_RESET BIT(1)
7398 #define KVM_DIRTY_GFN_F_MASK 0x3
7400 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
7401 ioctl to enable this capability for the new guest and set the size of
7402 the rings. Enabling the capability is only allowed before creating any
7403 vCPU, and the size of the ring must be a power of two. The larger the
7404 ring buffer, the less likely the ring is full and the VM is forced to
7405 exit to userspace. The optimal size depends on the workload, but it is
7406 recommended that it be at least 64 KiB (4096 entries).
7408 Just like for dirty page bitmaps, the buffer tracks writes to
7409 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
7410 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
7411 with the flag set, userspace can start harvesting dirty pages from the
7414 An entry in the ring buffer can be unused (flag bits ``00``),
7415 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
7416 state machine for the entry is as follows::
7418 dirtied harvested reset
7419 00 -----------> 01 -------------> 1X -------+
7422 +------------------------------------------+
7424 To harvest the dirty pages, userspace accesses the mmaped ring buffer
7425 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
7426 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
7427 The userspace should harvest this GFN and mark the flags from state
7428 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
7429 to show that this GFN is harvested and waiting for a reset), and move
7430 on to the next GFN. The userspace should continue to do this until the
7431 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
7432 all the dirty GFNs that were available.
7434 It's not necessary for userspace to harvest the all dirty GFNs at once.
7435 However it must collect the dirty GFNs in sequence, i.e., the userspace
7436 program cannot skip one dirty GFN to collect the one next to it.
7438 After processing one or more entries in the ring buffer, userspace
7439 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
7440 it, so that the kernel will reprotect those collected GFNs.
7441 Therefore, the ioctl must be called *before* reading the content of
7444 The dirty ring can get full. When it happens, the KVM_RUN of the
7445 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
7447 The dirty ring interface has a major difference comparing to the
7448 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
7449 userspace, it's still possible that the kernel has not yet flushed the
7450 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
7451 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
7452 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
7453 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
7455 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
7456 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
7457 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
7458 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
7459 machine will switch to ring-buffer dirty page tracking and further
7460 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
7462 8.30 KVM_CAP_XEN_HVM
7463 --------------------
7467 This capability indicates the features that Xen supports for hosting Xen
7468 PVHVM guests. Valid flags are::
7470 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
7471 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
7472 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
7473 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2)
7474 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 3)
7476 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
7477 ioctl is available, for the guest to set its hypercall page.
7479 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
7480 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
7481 contents, to request that KVM generate hypercall page content automatically
7482 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
7484 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
7485 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
7486 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
7487 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
7490 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
7491 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
7492 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
7494 The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries
7495 of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority
7496 field set to indicate 2 level event channel delivery.
7498 8.31 KVM_CAP_PPC_MULTITCE
7499 -------------------------
7501 :Capability: KVM_CAP_PPC_MULTITCE
7505 This capability means the kernel is capable of handling hypercalls
7506 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
7507 space. This significantly accelerates DMA operations for PPC KVM guests.
7508 User space should expect that its handlers for these hypercalls
7509 are not going to be called if user space previously registered LIOBN
7510 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
7512 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
7513 user space might have to advertise it for the guest. For example,
7514 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
7515 present in the "ibm,hypertas-functions" device-tree property.
7517 The hypercalls mentioned above may or may not be processed successfully
7518 in the kernel based fast path. If they can not be handled by the kernel,
7519 they will get passed on to user space. So user space still has to have
7520 an implementation for these despite the in kernel acceleration.
7522 This capability is always enabled.
7524 8.32 KVM_CAP_PTP_KVM
7525 --------------------
7527 :Architectures: arm64
7529 This capability indicates that the KVM virtual PTP service is
7530 supported in the host. A VMM can check whether the service is
7531 available to the guest on migration.
7533 8.33 KVM_CAP_HYPERV_ENFORCE_CPUID
7534 ---------------------------------
7538 When enabled, KVM will disable emulated Hyper-V features provided to the
7539 guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
7540 currently implmented Hyper-V features are provided unconditionally when
7541 Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
7544 8.34 KVM_CAP_EXIT_HYPERCALL
7545 ---------------------------
7547 :Capability: KVM_CAP_EXIT_HYPERCALL
7551 This capability, if enabled, will cause KVM to exit to userspace
7552 with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
7554 Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
7555 of hypercalls that can be configured to exit to userspace.
7556 Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
7558 The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
7559 of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace
7560 the hypercalls whose corresponding bit is in the argument, and return
7561 ENOSYS for the others.