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 4.9 KVM_SET_MEMORY_ALIAS
375 ------------------------
380 :Parameters: struct kvm_memory_alias (in)
381 :Returns: 0 (success), -1 (error)
383 This ioctl is obsolete and has been removed.
393 :Returns: 0 on success, -1 on error
397 ======= ==============================================================
398 EINTR an unmasked signal is pending
399 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
400 instructions from device memory (arm64)
401 ENOSYS data abort outside memslots with no syndrome info and
402 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
403 EPERM SVE feature set but not finalized (arm64)
404 ======= ==============================================================
406 This ioctl is used to run a guest virtual cpu. While there are no
407 explicit parameters, there is an implicit parameter block that can be
408 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
409 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
410 kvm_run' (see below).
417 :Architectures: all except ARM, arm64
419 :Parameters: struct kvm_regs (out)
420 :Returns: 0 on success, -1 on error
422 Reads the general purpose registers from the vcpu.
428 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
429 __u64 rax, rbx, rcx, rdx;
430 __u64 rsi, rdi, rsp, rbp;
431 __u64 r8, r9, r10, r11;
432 __u64 r12, r13, r14, r15;
438 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
450 :Architectures: all except ARM, arm64
452 :Parameters: struct kvm_regs (in)
453 :Returns: 0 on success, -1 on error
455 Writes the general purpose registers into the vcpu.
457 See KVM_GET_REGS for the data structure.
464 :Architectures: x86, ppc
466 :Parameters: struct kvm_sregs (out)
467 :Returns: 0 on success, -1 on error
469 Reads special registers from the vcpu.
475 struct kvm_segment cs, ds, es, fs, gs, ss;
476 struct kvm_segment tr, ldt;
477 struct kvm_dtable gdt, idt;
478 __u64 cr0, cr2, cr3, cr4, cr8;
481 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
484 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
486 interrupt_bitmap is a bitmap of pending external interrupts. At most
487 one bit may be set. This interrupt has been acknowledged by the APIC
488 but not yet injected into the cpu core.
495 :Architectures: x86, ppc
497 :Parameters: struct kvm_sregs (in)
498 :Returns: 0 on success, -1 on error
500 Writes special registers into the vcpu. See KVM_GET_SREGS for the
510 :Parameters: struct kvm_translation (in/out)
511 :Returns: 0 on success, -1 on error
513 Translates a virtual address according to the vcpu's current address
518 struct kvm_translation {
520 __u64 linear_address;
523 __u64 physical_address;
535 :Architectures: x86, ppc, mips
537 :Parameters: struct kvm_interrupt (in)
538 :Returns: 0 on success, negative on failure.
540 Queues a hardware interrupt vector to be injected.
544 /* for KVM_INTERRUPT */
545 struct kvm_interrupt {
555 ========= ===================================
557 -EEXIST if an interrupt is already enqueued
558 -EINVAL the irq number is invalid
559 -ENXIO if the PIC is in the kernel
560 -EFAULT if the pointer is invalid
561 ========= ===================================
563 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
564 ioctl is useful if the in-kernel PIC is not used.
569 Queues an external interrupt to be injected. This ioctl is overleaded
570 with 3 different irq values:
574 This injects an edge type external interrupt into the guest once it's ready
575 to receive interrupts. When injected, the interrupt is done.
577 b) KVM_INTERRUPT_UNSET
579 This unsets any pending interrupt.
581 Only available with KVM_CAP_PPC_UNSET_IRQ.
583 c) KVM_INTERRUPT_SET_LEVEL
585 This injects a level type external interrupt into the guest context. The
586 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
589 Only available with KVM_CAP_PPC_IRQ_LEVEL.
591 Note that any value for 'irq' other than the ones stated above is invalid
592 and incurs unexpected behavior.
594 This is an asynchronous vcpu ioctl and can be invoked from any thread.
599 Queues an external interrupt to be injected into the virtual CPU. A negative
600 interrupt number dequeues the interrupt.
602 This is an asynchronous vcpu ioctl and can be invoked from any thread.
612 :Returns: -1 on error
614 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
620 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
622 :Type: system ioctl, vcpu ioctl
623 :Parameters: struct kvm_msrs (in/out)
624 :Returns: number of msrs successfully returned;
627 When used as a system ioctl:
628 Reads the values of MSR-based features that are available for the VM. This
629 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
630 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
633 When used as a vcpu ioctl:
634 Reads model-specific registers from the vcpu. Supported msr indices can
635 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
640 __u32 nmsrs; /* number of msrs in entries */
643 struct kvm_msr_entry entries[0];
646 struct kvm_msr_entry {
652 Application code should set the 'nmsrs' member (which indicates the
653 size of the entries array) and the 'index' member of each array entry.
654 kvm will fill in the 'data' member.
663 :Parameters: struct kvm_msrs (in)
664 :Returns: number of msrs successfully set (see below), -1 on error
666 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
669 Application code should set the 'nmsrs' member (which indicates the
670 size of the entries array), and the 'index' and 'data' members of each
673 It tries to set the MSRs in array entries[] one by one. If setting an MSR
674 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
675 by KVM, etc..., it stops processing the MSR list and returns the number of
676 MSRs that have been set successfully.
685 :Parameters: struct kvm_cpuid (in)
686 :Returns: 0 on success, -1 on error
688 Defines the vcpu responses to the cpuid instruction. Applications
689 should use the KVM_SET_CPUID2 ioctl if available.
692 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID
693 configuration (if there is) is not corrupted. Userspace can get a copy
694 of the resulting CPUID configuration through KVM_GET_CPUID2 in case.
695 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model
696 after running the guest, may cause guest instability.
697 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc...
698 may cause guest instability.
702 struct kvm_cpuid_entry {
711 /* for KVM_SET_CPUID */
715 struct kvm_cpuid_entry entries[0];
719 4.21 KVM_SET_SIGNAL_MASK
720 ------------------------
725 :Parameters: struct kvm_signal_mask (in)
726 :Returns: 0 on success, -1 on error
728 Defines which signals are blocked during execution of KVM_RUN. This
729 signal mask temporarily overrides the threads signal mask. Any
730 unblocked signal received (except SIGKILL and SIGSTOP, which retain
731 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
733 Note the signal will only be delivered if not blocked by the original
738 /* for KVM_SET_SIGNAL_MASK */
739 struct kvm_signal_mask {
751 :Parameters: struct kvm_fpu (out)
752 :Returns: 0 on success, -1 on error
754 Reads the floating point state from the vcpu.
758 /* for KVM_GET_FPU and KVM_SET_FPU */
763 __u8 ftwx; /* in fxsave format */
780 :Parameters: struct kvm_fpu (in)
781 :Returns: 0 on success, -1 on error
783 Writes the floating point state to the vcpu.
787 /* for KVM_GET_FPU and KVM_SET_FPU */
792 __u8 ftwx; /* in fxsave format */
803 4.24 KVM_CREATE_IRQCHIP
804 -----------------------
806 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
807 :Architectures: x86, ARM, arm64, s390
810 :Returns: 0 on success, -1 on error
812 Creates an interrupt controller model in the kernel.
813 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
814 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
815 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
816 On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
817 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
818 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
819 On s390, a dummy irq routing table is created.
821 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
822 before KVM_CREATE_IRQCHIP can be used.
828 :Capability: KVM_CAP_IRQCHIP
829 :Architectures: x86, arm, arm64
831 :Parameters: struct kvm_irq_level
832 :Returns: 0 on success, -1 on error
834 Sets the level of a GSI input to the interrupt controller model in the kernel.
835 On some architectures it is required that an interrupt controller model has
836 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
837 interrupts require the level to be set to 1 and then back to 0.
839 On real hardware, interrupt pins can be active-low or active-high. This
840 does not matter for the level field of struct kvm_irq_level: 1 always
841 means active (asserted), 0 means inactive (deasserted).
843 x86 allows the operating system to program the interrupt polarity
844 (active-low/active-high) for level-triggered interrupts, and KVM used
845 to consider the polarity. However, due to bitrot in the handling of
846 active-low interrupts, the above convention is now valid on x86 too.
847 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
848 should not present interrupts to the guest as active-low unless this
849 capability is present (or unless it is not using the in-kernel irqchip,
853 ARM/arm64 can signal an interrupt either at the CPU level, or at the
854 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
855 use PPIs designated for specific cpus. The irq field is interpreted
858 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
859 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
861 The irq_type field has the following values:
864 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
866 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
867 (the vcpu_index field is ignored)
869 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
871 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
873 In both cases, level is used to assert/deassert the line.
875 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
876 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
879 Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
880 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
881 be used for a userspace interrupt controller.
885 struct kvm_irq_level {
888 __s32 status; /* not used for KVM_IRQ_LEVEL */
890 __u32 level; /* 0 or 1 */
897 :Capability: KVM_CAP_IRQCHIP
900 :Parameters: struct kvm_irqchip (in/out)
901 :Returns: 0 on success, -1 on error
903 Reads the state of a kernel interrupt controller created with
904 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
909 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
912 char dummy[512]; /* reserving space */
913 struct kvm_pic_state pic;
914 struct kvm_ioapic_state ioapic;
922 :Capability: KVM_CAP_IRQCHIP
925 :Parameters: struct kvm_irqchip (in)
926 :Returns: 0 on success, -1 on error
928 Sets the state of a kernel interrupt controller created with
929 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
934 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
937 char dummy[512]; /* reserving space */
938 struct kvm_pic_state pic;
939 struct kvm_ioapic_state ioapic;
944 4.28 KVM_XEN_HVM_CONFIG
945 -----------------------
947 :Capability: KVM_CAP_XEN_HVM
950 :Parameters: struct kvm_xen_hvm_config (in)
951 :Returns: 0 on success, -1 on error
953 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
954 page, and provides the starting address and size of the hypercall
955 blobs in userspace. When the guest writes the MSR, kvm copies one
956 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
961 struct kvm_xen_hvm_config {
971 If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the
972 KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl.
973 This requests KVM to generate the contents of the hypercall page
974 automatically; hypercalls will be intercepted and passed to userspace
975 through KVM_EXIT_XEN. In this case, all of the blob size and address
978 No other flags are currently valid in the struct kvm_xen_hvm_config.
983 :Capability: KVM_CAP_ADJUST_CLOCK
986 :Parameters: struct kvm_clock_data (out)
987 :Returns: 0 on success, -1 on error
989 Gets the current timestamp of kvmclock as seen by the current guest. In
990 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
993 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
994 set of bits that KVM can return in struct kvm_clock_data's flag member.
996 The only flag defined now is KVM_CLOCK_TSC_STABLE. If set, the returned
997 value is the exact kvmclock value seen by all VCPUs at the instant
998 when KVM_GET_CLOCK was called. If clear, the returned value is simply
999 CLOCK_MONOTONIC plus a constant offset; the offset can be modified
1000 with KVM_SET_CLOCK. KVM will try to make all VCPUs follow this clock,
1001 but the exact value read by each VCPU could differ, because the host
1006 struct kvm_clock_data {
1007 __u64 clock; /* kvmclock current value */
1016 :Capability: KVM_CAP_ADJUST_CLOCK
1019 :Parameters: struct kvm_clock_data (in)
1020 :Returns: 0 on success, -1 on error
1022 Sets the current timestamp of kvmclock to the value specified in its parameter.
1023 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1028 struct kvm_clock_data {
1029 __u64 clock; /* kvmclock current value */
1035 4.31 KVM_GET_VCPU_EVENTS
1036 ------------------------
1038 :Capability: KVM_CAP_VCPU_EVENTS
1039 :Extended by: KVM_CAP_INTR_SHADOW
1040 :Architectures: x86, arm, arm64
1042 :Parameters: struct kvm_vcpu_event (out)
1043 :Returns: 0 on success, -1 on error
1048 Gets currently pending exceptions, interrupts, and NMIs as well as related
1053 struct kvm_vcpu_events {
1057 __u8 has_error_code;
1078 __u8 smm_inside_nmi;
1082 __u8 exception_has_payload;
1083 __u64 exception_payload;
1086 The following bits are defined in the flags field:
1088 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1089 interrupt.shadow contains a valid state.
1091 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1094 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1095 exception_has_payload, exception_payload, and exception.pending
1096 fields contain a valid state. This bit will be set whenever
1097 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1102 If the guest accesses a device that is being emulated by the host kernel in
1103 such a way that a real device would generate a physical SError, KVM may make
1104 a virtual SError pending for that VCPU. This system error interrupt remains
1105 pending until the guest takes the exception by unmasking PSTATE.A.
1107 Running the VCPU may cause it to take a pending SError, or make an access that
1108 causes an SError to become pending. The event's description is only valid while
1109 the VPCU is not running.
1111 This API provides a way to read and write the pending 'event' state that is not
1112 visible to the guest. To save, restore or migrate a VCPU the struct representing
1113 the state can be read then written using this GET/SET API, along with the other
1114 guest-visible registers. It is not possible to 'cancel' an SError that has been
1117 A device being emulated in user-space may also wish to generate an SError. To do
1118 this the events structure can be populated by user-space. The current state
1119 should be read first, to ensure no existing SError is pending. If an existing
1120 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1121 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1122 Serviceability (RAS) Specification").
1124 SError exceptions always have an ESR value. Some CPUs have the ability to
1125 specify what the virtual SError's ESR value should be. These systems will
1126 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1127 always have a non-zero value when read, and the agent making an SError pending
1128 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1129 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1130 with exception.has_esr as zero, KVM will choose an ESR.
1132 Specifying exception.has_esr on a system that does not support it will return
1133 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1134 will return -EINVAL.
1136 It is not possible to read back a pending external abort (injected via
1137 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1138 directly to the virtual CPU).
1142 struct kvm_vcpu_events {
1144 __u8 serror_pending;
1145 __u8 serror_has_esr;
1146 __u8 ext_dabt_pending;
1147 /* Align it to 8 bytes */
1154 4.32 KVM_SET_VCPU_EVENTS
1155 ------------------------
1157 :Capability: KVM_CAP_VCPU_EVENTS
1158 :Extended by: KVM_CAP_INTR_SHADOW
1159 :Architectures: x86, arm, arm64
1161 :Parameters: struct kvm_vcpu_event (in)
1162 :Returns: 0 on success, -1 on error
1167 Set pending exceptions, interrupts, and NMIs as well as related states of the
1170 See KVM_GET_VCPU_EVENTS for the data structure.
1172 Fields that may be modified asynchronously by running VCPUs can be excluded
1173 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1174 smi.pending. Keep the corresponding bits in the flags field cleared to
1175 suppress overwriting the current in-kernel state. The bits are:
1177 =============================== ==================================
1178 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1179 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1180 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1181 =============================== ==================================
1183 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1184 the flags field to signal that interrupt.shadow contains a valid state and
1185 shall be written into the VCPU.
1187 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1189 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1190 can be set in the flags field to signal that the
1191 exception_has_payload, exception_payload, and exception.pending fields
1192 contain a valid state and shall be written into the VCPU.
1197 User space may need to inject several types of events to the guest.
1199 Set the pending SError exception state for this VCPU. It is not possible to
1200 'cancel' an Serror that has been made pending.
1202 If the guest performed an access to I/O memory which could not be handled by
1203 userspace, for example because of missing instruction syndrome decode
1204 information or because there is no device mapped at the accessed IPA, then
1205 userspace can ask the kernel to inject an external abort using the address
1206 from the exiting fault on the VCPU. It is a programming error to set
1207 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1208 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1209 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1210 how userspace reports accesses for the above cases to guests, across different
1211 userspace implementations. Nevertheless, userspace can still emulate all Arm
1212 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1214 See KVM_GET_VCPU_EVENTS for the data structure.
1217 4.33 KVM_GET_DEBUGREGS
1218 ----------------------
1220 :Capability: KVM_CAP_DEBUGREGS
1223 :Parameters: struct kvm_debugregs (out)
1224 :Returns: 0 on success, -1 on error
1226 Reads debug registers from the vcpu.
1230 struct kvm_debugregs {
1239 4.34 KVM_SET_DEBUGREGS
1240 ----------------------
1242 :Capability: KVM_CAP_DEBUGREGS
1245 :Parameters: struct kvm_debugregs (in)
1246 :Returns: 0 on success, -1 on error
1248 Writes debug registers into the vcpu.
1250 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1251 yet and must be cleared on entry.
1254 4.35 KVM_SET_USER_MEMORY_REGION
1255 -------------------------------
1257 :Capability: KVM_CAP_USER_MEMORY
1260 :Parameters: struct kvm_userspace_memory_region (in)
1261 :Returns: 0 on success, -1 on error
1265 struct kvm_userspace_memory_region {
1268 __u64 guest_phys_addr;
1269 __u64 memory_size; /* bytes */
1270 __u64 userspace_addr; /* start of the userspace allocated memory */
1273 /* for kvm_memory_region::flags */
1274 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1275 #define KVM_MEM_READONLY (1UL << 1)
1277 This ioctl allows the user to create, modify or delete a guest physical
1278 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1279 should be less than the maximum number of user memory slots supported per
1280 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1281 Slots may not overlap in guest physical address space.
1283 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1284 specifies the address space which is being modified. They must be
1285 less than the value that KVM_CHECK_EXTENSION returns for the
1286 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1287 are unrelated; the restriction on overlapping slots only applies within
1290 Deleting a slot is done by passing zero for memory_size. When changing
1291 an existing slot, it may be moved in the guest physical memory space,
1292 or its flags may be modified, but it may not be resized.
1294 Memory for the region is taken starting at the address denoted by the
1295 field userspace_addr, which must point at user addressable memory for
1296 the entire memory slot size. Any object may back this memory, including
1297 anonymous memory, ordinary files, and hugetlbfs.
1299 On architectures that support a form of address tagging, userspace_addr must
1300 be an untagged address.
1302 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1303 be identical. This allows large pages in the guest to be backed by large
1306 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1307 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1308 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1309 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1310 to make a new slot read-only. In this case, writes to this memory will be
1311 posted to userspace as KVM_EXIT_MMIO exits.
1313 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1314 the memory region are automatically reflected into the guest. For example, an
1315 mmap() that affects the region will be made visible immediately. Another
1316 example is madvise(MADV_DROP).
1318 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1319 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1320 allocation and is deprecated.
1323 4.36 KVM_SET_TSS_ADDR
1324 ---------------------
1326 :Capability: KVM_CAP_SET_TSS_ADDR
1329 :Parameters: unsigned long tss_address (in)
1330 :Returns: 0 on success, -1 on error
1332 This ioctl defines the physical address of a three-page region in the guest
1333 physical address space. The region must be within the first 4GB of the
1334 guest physical address space and must not conflict with any memory slot
1335 or any mmio address. The guest may malfunction if it accesses this memory
1338 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1339 because of a quirk in the virtualization implementation (see the internals
1340 documentation when it pops into existence).
1346 :Capability: KVM_CAP_ENABLE_CAP
1347 :Architectures: mips, ppc, s390
1349 :Parameters: struct kvm_enable_cap (in)
1350 :Returns: 0 on success; -1 on error
1352 :Capability: KVM_CAP_ENABLE_CAP_VM
1355 :Parameters: struct kvm_enable_cap (in)
1356 :Returns: 0 on success; -1 on error
1360 Not all extensions are enabled by default. Using this ioctl the application
1361 can enable an extension, making it available to the guest.
1363 On systems that do not support this ioctl, it always fails. On systems that
1364 do support it, it only works for extensions that are supported for enablement.
1366 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1371 struct kvm_enable_cap {
1375 The capability that is supposed to get enabled.
1381 A bitfield indicating future enhancements. Has to be 0 for now.
1387 Arguments for enabling a feature. If a feature needs initial values to
1388 function properly, this is the place to put them.
1395 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1396 for vm-wide capabilities.
1398 4.38 KVM_GET_MP_STATE
1399 ---------------------
1401 :Capability: KVM_CAP_MP_STATE
1402 :Architectures: x86, s390, arm, arm64
1404 :Parameters: struct kvm_mp_state (out)
1405 :Returns: 0 on success; -1 on error
1409 struct kvm_mp_state {
1413 Returns the vcpu's current "multiprocessing state" (though also valid on
1414 uniprocessor guests).
1416 Possible values are:
1418 ========================== ===============================================
1419 KVM_MP_STATE_RUNNABLE the vcpu is currently running [x86,arm/arm64]
1420 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1421 which has not yet received an INIT signal [x86]
1422 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1423 now ready for a SIPI [x86]
1424 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1425 is waiting for an interrupt [x86]
1426 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1427 accessible via KVM_GET_VCPU_EVENTS) [x86]
1428 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm/arm64]
1429 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1430 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1432 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1434 ========================== ===============================================
1436 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1437 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1438 these architectures.
1443 The only states that are valid are KVM_MP_STATE_STOPPED and
1444 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1446 4.39 KVM_SET_MP_STATE
1447 ---------------------
1449 :Capability: KVM_CAP_MP_STATE
1450 :Architectures: x86, s390, arm, arm64
1452 :Parameters: struct kvm_mp_state (in)
1453 :Returns: 0 on success; -1 on error
1455 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1458 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1459 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1460 these architectures.
1465 The only states that are valid are KVM_MP_STATE_STOPPED and
1466 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1468 4.40 KVM_SET_IDENTITY_MAP_ADDR
1469 ------------------------------
1471 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1474 :Parameters: unsigned long identity (in)
1475 :Returns: 0 on success, -1 on error
1477 This ioctl defines the physical address of a one-page region in the guest
1478 physical address space. The region must be within the first 4GB of the
1479 guest physical address space and must not conflict with any memory slot
1480 or any mmio address. The guest may malfunction if it accesses this memory
1483 Setting the address to 0 will result in resetting the address to its default
1486 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1487 because of a quirk in the virtualization implementation (see the internals
1488 documentation when it pops into existence).
1490 Fails if any VCPU has already been created.
1492 4.41 KVM_SET_BOOT_CPU_ID
1493 ------------------------
1495 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1498 :Parameters: unsigned long vcpu_id
1499 :Returns: 0 on success, -1 on error
1501 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1502 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1503 is vcpu 0. This ioctl has to be called before vcpu creation,
1504 otherwise it will return EBUSY error.
1510 :Capability: KVM_CAP_XSAVE
1513 :Parameters: struct kvm_xsave (out)
1514 :Returns: 0 on success, -1 on error
1523 This ioctl would copy current vcpu's xsave struct to the userspace.
1529 :Capability: KVM_CAP_XSAVE
1532 :Parameters: struct kvm_xsave (in)
1533 :Returns: 0 on success, -1 on error
1542 This ioctl would copy userspace's xsave struct to the kernel.
1548 :Capability: KVM_CAP_XCRS
1551 :Parameters: struct kvm_xcrs (out)
1552 :Returns: 0 on success, -1 on error
1565 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1569 This ioctl would copy current vcpu's xcrs to the userspace.
1575 :Capability: KVM_CAP_XCRS
1578 :Parameters: struct kvm_xcrs (in)
1579 :Returns: 0 on success, -1 on error
1592 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1596 This ioctl would set vcpu's xcr to the value userspace specified.
1599 4.46 KVM_GET_SUPPORTED_CPUID
1600 ----------------------------
1602 :Capability: KVM_CAP_EXT_CPUID
1605 :Parameters: struct kvm_cpuid2 (in/out)
1606 :Returns: 0 on success, -1 on error
1613 struct kvm_cpuid_entry2 entries[0];
1616 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1617 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1618 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1620 struct kvm_cpuid_entry2 {
1631 This ioctl returns x86 cpuid features which are supported by both the
1632 hardware and kvm in its default configuration. Userspace can use the
1633 information returned by this ioctl to construct cpuid information (for
1634 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1635 userspace capabilities, and with user requirements (for example, the
1636 user may wish to constrain cpuid to emulate older hardware, or for
1637 feature consistency across a cluster).
1639 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1640 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1641 its default configuration. If userspace enables such capabilities, it
1642 is responsible for modifying the results of this ioctl appropriately.
1644 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1645 with the 'nent' field indicating the number of entries in the variable-size
1646 array 'entries'. If the number of entries is too low to describe the cpu
1647 capabilities, an error (E2BIG) is returned. If the number is too high,
1648 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1649 number is just right, the 'nent' field is adjusted to the number of valid
1650 entries in the 'entries' array, which is then filled.
1652 The entries returned are the host cpuid as returned by the cpuid instruction,
1653 with unknown or unsupported features masked out. Some features (for example,
1654 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1655 emulate them efficiently. The fields in each entry are defined as follows:
1658 the eax value used to obtain the entry
1661 the ecx value used to obtain the entry (for entries that are
1665 an OR of zero or more of the following:
1667 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1668 if the index field is valid
1671 the values returned by the cpuid instruction for
1672 this function/index combination
1674 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1675 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1676 support. Instead it is reported via::
1678 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1680 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1681 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1684 4.47 KVM_PPC_GET_PVINFO
1685 -----------------------
1687 :Capability: KVM_CAP_PPC_GET_PVINFO
1690 :Parameters: struct kvm_ppc_pvinfo (out)
1691 :Returns: 0 on success, !0 on error
1695 struct kvm_ppc_pvinfo {
1701 This ioctl fetches PV specific information that need to be passed to the guest
1702 using the device tree or other means from vm context.
1704 The hcall array defines 4 instructions that make up a hypercall.
1706 If any additional field gets added to this structure later on, a bit for that
1707 additional piece of information will be set in the flags bitmap.
1709 The flags bitmap is defined as::
1711 /* the host supports the ePAPR idle hcall
1712 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1714 4.52 KVM_SET_GSI_ROUTING
1715 ------------------------
1717 :Capability: KVM_CAP_IRQ_ROUTING
1718 :Architectures: x86 s390 arm arm64
1720 :Parameters: struct kvm_irq_routing (in)
1721 :Returns: 0 on success, -1 on error
1723 Sets the GSI routing table entries, overwriting any previously set entries.
1725 On arm/arm64, GSI routing has the following limitation:
1727 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1731 struct kvm_irq_routing {
1734 struct kvm_irq_routing_entry entries[0];
1737 No flags are specified so far, the corresponding field must be set to zero.
1741 struct kvm_irq_routing_entry {
1747 struct kvm_irq_routing_irqchip irqchip;
1748 struct kvm_irq_routing_msi msi;
1749 struct kvm_irq_routing_s390_adapter adapter;
1750 struct kvm_irq_routing_hv_sint hv_sint;
1755 /* gsi routing entry types */
1756 #define KVM_IRQ_ROUTING_IRQCHIP 1
1757 #define KVM_IRQ_ROUTING_MSI 2
1758 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1759 #define KVM_IRQ_ROUTING_HV_SINT 4
1763 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1764 type, specifies that the devid field contains a valid value. The per-VM
1765 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1766 the device ID. If this capability is not available, userspace should
1767 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1772 struct kvm_irq_routing_irqchip {
1777 struct kvm_irq_routing_msi {
1787 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1788 for the device that wrote the MSI message. For PCI, this is usually a
1789 BFD identifier in the lower 16 bits.
1791 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1792 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1793 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1794 address_hi must be zero.
1798 struct kvm_irq_routing_s390_adapter {
1802 __u32 summary_offset;
1806 struct kvm_irq_routing_hv_sint {
1812 4.55 KVM_SET_TSC_KHZ
1813 --------------------
1815 :Capability: KVM_CAP_TSC_CONTROL
1818 :Parameters: virtual tsc_khz
1819 :Returns: 0 on success, -1 on error
1821 Specifies the tsc frequency for the virtual machine. The unit of the
1825 4.56 KVM_GET_TSC_KHZ
1826 --------------------
1828 :Capability: KVM_CAP_GET_TSC_KHZ
1832 :Returns: virtual tsc-khz on success, negative value on error
1834 Returns the tsc frequency of the guest. The unit of the return value is
1835 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1842 :Capability: KVM_CAP_IRQCHIP
1845 :Parameters: struct kvm_lapic_state (out)
1846 :Returns: 0 on success, -1 on error
1850 #define KVM_APIC_REG_SIZE 0x400
1851 struct kvm_lapic_state {
1852 char regs[KVM_APIC_REG_SIZE];
1855 Reads the Local APIC registers and copies them into the input argument. The
1856 data format and layout are the same as documented in the architecture manual.
1858 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1859 enabled, then the format of APIC_ID register depends on the APIC mode
1860 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1861 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1862 which is stored in bits 31-24 of the APIC register, or equivalently in
1863 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1864 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1866 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1867 always uses xAPIC format.
1873 :Capability: KVM_CAP_IRQCHIP
1876 :Parameters: struct kvm_lapic_state (in)
1877 :Returns: 0 on success, -1 on error
1881 #define KVM_APIC_REG_SIZE 0x400
1882 struct kvm_lapic_state {
1883 char regs[KVM_APIC_REG_SIZE];
1886 Copies the input argument into the Local APIC registers. The data format
1887 and layout are the same as documented in the architecture manual.
1889 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1890 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1891 See the note in KVM_GET_LAPIC.
1897 :Capability: KVM_CAP_IOEVENTFD
1900 :Parameters: struct kvm_ioeventfd (in)
1901 :Returns: 0 on success, !0 on error
1903 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1904 within the guest. A guest write in the registered address will signal the
1905 provided event instead of triggering an exit.
1909 struct kvm_ioeventfd {
1911 __u64 addr; /* legal pio/mmio address */
1912 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1918 For the special case of virtio-ccw devices on s390, the ioevent is matched
1919 to a subchannel/virtqueue tuple instead.
1921 The following flags are defined::
1923 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1924 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
1925 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
1926 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1927 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1929 If datamatch flag is set, the event will be signaled only if the written value
1930 to the registered address is equal to datamatch in struct kvm_ioeventfd.
1932 For virtio-ccw devices, addr contains the subchannel id and datamatch the
1935 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1936 the kernel will ignore the length of guest write and may get a faster vmexit.
1937 The speedup may only apply to specific architectures, but the ioeventfd will
1943 :Capability: KVM_CAP_SW_TLB
1946 :Parameters: struct kvm_dirty_tlb (in)
1947 :Returns: 0 on success, -1 on error
1951 struct kvm_dirty_tlb {
1956 This must be called whenever userspace has changed an entry in the shared
1957 TLB, prior to calling KVM_RUN on the associated vcpu.
1959 The "bitmap" field is the userspace address of an array. This array
1960 consists of a number of bits, equal to the total number of TLB entries as
1961 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1962 nearest multiple of 64.
1964 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1967 The array is little-endian: the bit 0 is the least significant bit of the
1968 first byte, bit 8 is the least significant bit of the second byte, etc.
1969 This avoids any complications with differing word sizes.
1971 The "num_dirty" field is a performance hint for KVM to determine whether it
1972 should skip processing the bitmap and just invalidate everything. It must
1973 be set to the number of set bits in the bitmap.
1976 4.62 KVM_CREATE_SPAPR_TCE
1977 -------------------------
1979 :Capability: KVM_CAP_SPAPR_TCE
1980 :Architectures: powerpc
1982 :Parameters: struct kvm_create_spapr_tce (in)
1983 :Returns: file descriptor for manipulating the created TCE table
1985 This creates a virtual TCE (translation control entry) table, which
1986 is an IOMMU for PAPR-style virtual I/O. It is used to translate
1987 logical addresses used in virtual I/O into guest physical addresses,
1988 and provides a scatter/gather capability for PAPR virtual I/O.
1992 /* for KVM_CAP_SPAPR_TCE */
1993 struct kvm_create_spapr_tce {
1998 The liobn field gives the logical IO bus number for which to create a
1999 TCE table. The window_size field specifies the size of the DMA window
2000 which this TCE table will translate - the table will contain one 64
2001 bit TCE entry for every 4kiB of the DMA window.
2003 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2004 table has been created using this ioctl(), the kernel will handle it
2005 in real mode, updating the TCE table. H_PUT_TCE calls for other
2006 liobns will cause a vm exit and must be handled by userspace.
2008 The return value is a file descriptor which can be passed to mmap(2)
2009 to map the created TCE table into userspace. This lets userspace read
2010 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2011 userspace update the TCE table directly which is useful in some
2015 4.63 KVM_ALLOCATE_RMA
2016 ---------------------
2018 :Capability: KVM_CAP_PPC_RMA
2019 :Architectures: powerpc
2021 :Parameters: struct kvm_allocate_rma (out)
2022 :Returns: file descriptor for mapping the allocated RMA
2024 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2025 time by the kernel. An RMA is a physically-contiguous, aligned region
2026 of memory used on older POWER processors to provide the memory which
2027 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2028 POWER processors support a set of sizes for the RMA that usually
2029 includes 64MB, 128MB, 256MB and some larger powers of two.
2033 /* for KVM_ALLOCATE_RMA */
2034 struct kvm_allocate_rma {
2038 The return value is a file descriptor which can be passed to mmap(2)
2039 to map the allocated RMA into userspace. The mapped area can then be
2040 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2041 RMA for a virtual machine. The size of the RMA in bytes (which is
2042 fixed at host kernel boot time) is returned in the rma_size field of
2043 the argument structure.
2045 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2046 is supported; 2 if the processor requires all virtual machines to have
2047 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2048 because it supports the Virtual RMA (VRMA) facility.
2054 :Capability: KVM_CAP_USER_NMI
2058 :Returns: 0 on success, -1 on error
2060 Queues an NMI on the thread's vcpu. Note this is well defined only
2061 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2062 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2063 has been called, this interface is completely emulated within the kernel.
2065 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2066 following algorithm:
2069 - read the local APIC's state (KVM_GET_LAPIC)
2070 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2071 - if so, issue KVM_NMI
2074 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2078 4.65 KVM_S390_UCAS_MAP
2079 ----------------------
2081 :Capability: KVM_CAP_S390_UCONTROL
2082 :Architectures: s390
2084 :Parameters: struct kvm_s390_ucas_mapping (in)
2085 :Returns: 0 in case of success
2087 The parameter is defined like this::
2089 struct kvm_s390_ucas_mapping {
2095 This ioctl maps the memory at "user_addr" with the length "length" to
2096 the vcpu's address space starting at "vcpu_addr". All parameters need to
2097 be aligned by 1 megabyte.
2100 4.66 KVM_S390_UCAS_UNMAP
2101 ------------------------
2103 :Capability: KVM_CAP_S390_UCONTROL
2104 :Architectures: s390
2106 :Parameters: struct kvm_s390_ucas_mapping (in)
2107 :Returns: 0 in case of success
2109 The parameter is defined like this::
2111 struct kvm_s390_ucas_mapping {
2117 This ioctl unmaps the memory in the vcpu's address space starting at
2118 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2119 All parameters need to be aligned by 1 megabyte.
2122 4.67 KVM_S390_VCPU_FAULT
2123 ------------------------
2125 :Capability: KVM_CAP_S390_UCONTROL
2126 :Architectures: s390
2128 :Parameters: vcpu absolute address (in)
2129 :Returns: 0 in case of success
2131 This call creates a page table entry on the virtual cpu's address space
2132 (for user controlled virtual machines) or the virtual machine's address
2133 space (for regular virtual machines). This only works for minor faults,
2134 thus it's recommended to access subject memory page via the user page
2135 table upfront. This is useful to handle validity intercepts for user
2136 controlled virtual machines to fault in the virtual cpu's lowcore pages
2137 prior to calling the KVM_RUN ioctl.
2140 4.68 KVM_SET_ONE_REG
2141 --------------------
2143 :Capability: KVM_CAP_ONE_REG
2146 :Parameters: struct kvm_one_reg (in)
2147 :Returns: 0 on success, negative value on failure
2151 ====== ============================================================
2152 ENOENT no such register
2153 EINVAL invalid register ID, or no such register or used with VMs in
2154 protected virtualization mode on s390
2155 EPERM (arm64) register access not allowed before vcpu finalization
2156 ====== ============================================================
2158 (These error codes are indicative only: do not rely on a specific error
2159 code being returned in a specific situation.)
2163 struct kvm_one_reg {
2168 Using this ioctl, a single vcpu register can be set to a specific value
2169 defined by user space with the passed in struct kvm_one_reg, where id
2170 refers to the register identifier as described below and addr is a pointer
2171 to a variable with the respective size. There can be architecture agnostic
2172 and architecture specific registers. Each have their own range of operation
2173 and their own constants and width. To keep track of the implemented
2174 registers, find a list below:
2176 ======= =============================== ============
2177 Arch Register Width (bits)
2178 ======= =============================== ============
2179 PPC KVM_REG_PPC_HIOR 64
2180 PPC KVM_REG_PPC_IAC1 64
2181 PPC KVM_REG_PPC_IAC2 64
2182 PPC KVM_REG_PPC_IAC3 64
2183 PPC KVM_REG_PPC_IAC4 64
2184 PPC KVM_REG_PPC_DAC1 64
2185 PPC KVM_REG_PPC_DAC2 64
2186 PPC KVM_REG_PPC_DABR 64
2187 PPC KVM_REG_PPC_DSCR 64
2188 PPC KVM_REG_PPC_PURR 64
2189 PPC KVM_REG_PPC_SPURR 64
2190 PPC KVM_REG_PPC_DAR 64
2191 PPC KVM_REG_PPC_DSISR 32
2192 PPC KVM_REG_PPC_AMR 64
2193 PPC KVM_REG_PPC_UAMOR 64
2194 PPC KVM_REG_PPC_MMCR0 64
2195 PPC KVM_REG_PPC_MMCR1 64
2196 PPC KVM_REG_PPC_MMCRA 64
2197 PPC KVM_REG_PPC_MMCR2 64
2198 PPC KVM_REG_PPC_MMCRS 64
2199 PPC KVM_REG_PPC_MMCR3 64
2200 PPC KVM_REG_PPC_SIAR 64
2201 PPC KVM_REG_PPC_SDAR 64
2202 PPC KVM_REG_PPC_SIER 64
2203 PPC KVM_REG_PPC_SIER2 64
2204 PPC KVM_REG_PPC_SIER3 64
2205 PPC KVM_REG_PPC_PMC1 32
2206 PPC KVM_REG_PPC_PMC2 32
2207 PPC KVM_REG_PPC_PMC3 32
2208 PPC KVM_REG_PPC_PMC4 32
2209 PPC KVM_REG_PPC_PMC5 32
2210 PPC KVM_REG_PPC_PMC6 32
2211 PPC KVM_REG_PPC_PMC7 32
2212 PPC KVM_REG_PPC_PMC8 32
2213 PPC KVM_REG_PPC_FPR0 64
2215 PPC KVM_REG_PPC_FPR31 64
2216 PPC KVM_REG_PPC_VR0 128
2218 PPC KVM_REG_PPC_VR31 128
2219 PPC KVM_REG_PPC_VSR0 128
2221 PPC KVM_REG_PPC_VSR31 128
2222 PPC KVM_REG_PPC_FPSCR 64
2223 PPC KVM_REG_PPC_VSCR 32
2224 PPC KVM_REG_PPC_VPA_ADDR 64
2225 PPC KVM_REG_PPC_VPA_SLB 128
2226 PPC KVM_REG_PPC_VPA_DTL 128
2227 PPC KVM_REG_PPC_EPCR 32
2228 PPC KVM_REG_PPC_EPR 32
2229 PPC KVM_REG_PPC_TCR 32
2230 PPC KVM_REG_PPC_TSR 32
2231 PPC KVM_REG_PPC_OR_TSR 32
2232 PPC KVM_REG_PPC_CLEAR_TSR 32
2233 PPC KVM_REG_PPC_MAS0 32
2234 PPC KVM_REG_PPC_MAS1 32
2235 PPC KVM_REG_PPC_MAS2 64
2236 PPC KVM_REG_PPC_MAS7_3 64
2237 PPC KVM_REG_PPC_MAS4 32
2238 PPC KVM_REG_PPC_MAS6 32
2239 PPC KVM_REG_PPC_MMUCFG 32
2240 PPC KVM_REG_PPC_TLB0CFG 32
2241 PPC KVM_REG_PPC_TLB1CFG 32
2242 PPC KVM_REG_PPC_TLB2CFG 32
2243 PPC KVM_REG_PPC_TLB3CFG 32
2244 PPC KVM_REG_PPC_TLB0PS 32
2245 PPC KVM_REG_PPC_TLB1PS 32
2246 PPC KVM_REG_PPC_TLB2PS 32
2247 PPC KVM_REG_PPC_TLB3PS 32
2248 PPC KVM_REG_PPC_EPTCFG 32
2249 PPC KVM_REG_PPC_ICP_STATE 64
2250 PPC KVM_REG_PPC_VP_STATE 128
2251 PPC KVM_REG_PPC_TB_OFFSET 64
2252 PPC KVM_REG_PPC_SPMC1 32
2253 PPC KVM_REG_PPC_SPMC2 32
2254 PPC KVM_REG_PPC_IAMR 64
2255 PPC KVM_REG_PPC_TFHAR 64
2256 PPC KVM_REG_PPC_TFIAR 64
2257 PPC KVM_REG_PPC_TEXASR 64
2258 PPC KVM_REG_PPC_FSCR 64
2259 PPC KVM_REG_PPC_PSPB 32
2260 PPC KVM_REG_PPC_EBBHR 64
2261 PPC KVM_REG_PPC_EBBRR 64
2262 PPC KVM_REG_PPC_BESCR 64
2263 PPC KVM_REG_PPC_TAR 64
2264 PPC KVM_REG_PPC_DPDES 64
2265 PPC KVM_REG_PPC_DAWR 64
2266 PPC KVM_REG_PPC_DAWRX 64
2267 PPC KVM_REG_PPC_CIABR 64
2268 PPC KVM_REG_PPC_IC 64
2269 PPC KVM_REG_PPC_VTB 64
2270 PPC KVM_REG_PPC_CSIGR 64
2271 PPC KVM_REG_PPC_TACR 64
2272 PPC KVM_REG_PPC_TCSCR 64
2273 PPC KVM_REG_PPC_PID 64
2274 PPC KVM_REG_PPC_ACOP 64
2275 PPC KVM_REG_PPC_VRSAVE 32
2276 PPC KVM_REG_PPC_LPCR 32
2277 PPC KVM_REG_PPC_LPCR_64 64
2278 PPC KVM_REG_PPC_PPR 64
2279 PPC KVM_REG_PPC_ARCH_COMPAT 32
2280 PPC KVM_REG_PPC_DABRX 32
2281 PPC KVM_REG_PPC_WORT 64
2282 PPC KVM_REG_PPC_SPRG9 64
2283 PPC KVM_REG_PPC_DBSR 32
2284 PPC KVM_REG_PPC_TIDR 64
2285 PPC KVM_REG_PPC_PSSCR 64
2286 PPC KVM_REG_PPC_DEC_EXPIRY 64
2287 PPC KVM_REG_PPC_PTCR 64
2288 PPC KVM_REG_PPC_DAWR1 64
2289 PPC KVM_REG_PPC_DAWRX1 64
2290 PPC KVM_REG_PPC_TM_GPR0 64
2292 PPC KVM_REG_PPC_TM_GPR31 64
2293 PPC KVM_REG_PPC_TM_VSR0 128
2295 PPC KVM_REG_PPC_TM_VSR63 128
2296 PPC KVM_REG_PPC_TM_CR 64
2297 PPC KVM_REG_PPC_TM_LR 64
2298 PPC KVM_REG_PPC_TM_CTR 64
2299 PPC KVM_REG_PPC_TM_FPSCR 64
2300 PPC KVM_REG_PPC_TM_AMR 64
2301 PPC KVM_REG_PPC_TM_PPR 64
2302 PPC KVM_REG_PPC_TM_VRSAVE 64
2303 PPC KVM_REG_PPC_TM_VSCR 32
2304 PPC KVM_REG_PPC_TM_DSCR 64
2305 PPC KVM_REG_PPC_TM_TAR 64
2306 PPC KVM_REG_PPC_TM_XER 64
2308 MIPS KVM_REG_MIPS_R0 64
2310 MIPS KVM_REG_MIPS_R31 64
2311 MIPS KVM_REG_MIPS_HI 64
2312 MIPS KVM_REG_MIPS_LO 64
2313 MIPS KVM_REG_MIPS_PC 64
2314 MIPS KVM_REG_MIPS_CP0_INDEX 32
2315 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2316 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2317 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2318 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2319 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2320 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2321 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2322 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2323 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2324 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2325 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2326 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2327 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2328 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2329 MIPS KVM_REG_MIPS_CP0_WIRED 32
2330 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2331 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2332 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2333 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2334 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2335 MIPS KVM_REG_MIPS_CP0_COUNT 32
2336 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2337 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2338 MIPS KVM_REG_MIPS_CP0_STATUS 32
2339 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2340 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2341 MIPS KVM_REG_MIPS_CP0_EPC 64
2342 MIPS KVM_REG_MIPS_CP0_PRID 32
2343 MIPS KVM_REG_MIPS_CP0_EBASE 64
2344 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2345 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2346 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2347 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2348 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2349 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2350 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2351 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2352 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2353 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2354 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2355 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2356 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2357 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2358 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2359 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2360 MIPS KVM_REG_MIPS_COUNT_CTL 64
2361 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2362 MIPS KVM_REG_MIPS_COUNT_HZ 64
2363 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2364 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2365 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2366 MIPS KVM_REG_MIPS_FCR_IR 32
2367 MIPS KVM_REG_MIPS_FCR_CSR 32
2368 MIPS KVM_REG_MIPS_MSA_IR 32
2369 MIPS KVM_REG_MIPS_MSA_CSR 32
2370 ======= =============================== ============
2372 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2373 is the register group type, or coprocessor number:
2375 ARM core registers have the following id bit patterns::
2377 0x4020 0000 0010 <index into the kvm_regs struct:16>
2379 ARM 32-bit CP15 registers have the following id bit patterns::
2381 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2383 ARM 64-bit CP15 registers have the following id bit patterns::
2385 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2387 ARM CCSIDR registers are demultiplexed by CSSELR value::
2389 0x4020 0000 0011 00 <csselr:8>
2391 ARM 32-bit VFP control registers have the following id bit patterns::
2393 0x4020 0000 0012 1 <regno:12>
2395 ARM 64-bit FP registers have the following id bit patterns::
2397 0x4030 0000 0012 0 <regno:12>
2399 ARM firmware pseudo-registers have the following bit pattern::
2401 0x4030 0000 0014 <regno:16>
2404 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2405 that is the register group type, or coprocessor number:
2407 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2408 that the size of the access is variable, as the kvm_regs structure
2409 contains elements ranging from 32 to 128 bits. The index is a 32bit
2410 value in the kvm_regs structure seen as a 32bit array::
2412 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2416 ======================= ========= ===== =======================================
2417 Encoding Register Bits kvm_regs member
2418 ======================= ========= ===== =======================================
2419 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2420 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2422 0x6030 0000 0010 003c X30 64 regs.regs[30]
2423 0x6030 0000 0010 003e SP 64 regs.sp
2424 0x6030 0000 0010 0040 PC 64 regs.pc
2425 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2426 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2427 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2428 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2429 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2430 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2431 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2432 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2433 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2434 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2436 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2437 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2438 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2439 ======================= ========= ===== =======================================
2441 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2444 The equivalent register content can be accessed via bits [127:0] of
2445 the corresponding SVE Zn registers instead for vcpus that have SVE
2446 enabled (see below).
2448 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2450 0x6020 0000 0011 00 <csselr:8>
2452 arm64 system registers have the following id bit patterns::
2454 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2458 Two system register IDs do not follow the specified pattern. These
2459 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2460 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2461 two had their values accidentally swapped, which means TIMER_CVAL is
2462 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2463 derived from the register encoding for CNTV_CVAL_EL0. As this is
2464 API, it must remain this way.
2466 arm64 firmware pseudo-registers have the following bit pattern::
2468 0x6030 0000 0014 <regno:16>
2470 arm64 SVE registers have the following bit patterns::
2472 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2473 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2474 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2475 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2477 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2478 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2479 quadwords: see [2]_ below.
2481 These registers are only accessible on vcpus for which SVE is enabled.
2482 See KVM_ARM_VCPU_INIT for details.
2484 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2485 accessible until the vcpu's SVE configuration has been finalized
2486 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2487 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2489 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2490 lengths supported by the vcpu to be discovered and configured by
2491 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2492 or KVM_SET_ONE_REG, the value of this register is of type
2493 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2496 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2498 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2499 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2500 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2501 /* Vector length vq * 16 bytes supported */
2503 /* Vector length vq * 16 bytes not supported */
2505 .. [2] The maximum value vq for which the above condition is true is
2506 max_vq. This is the maximum vector length available to the guest on
2507 this vcpu, and determines which register slices are visible through
2508 this ioctl interface.
2510 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2513 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2514 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2517 Userspace may subsequently modify it if desired until the vcpu's SVE
2518 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2520 Apart from simply removing all vector lengths from the host set that
2521 exceed some value, support for arbitrarily chosen sets of vector lengths
2522 is hardware-dependent and may not be available. Attempting to configure
2523 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2526 After the vcpu's SVE configuration is finalized, further attempts to
2527 write this register will fail with EPERM.
2530 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2531 the register group type:
2533 MIPS core registers (see above) have the following id bit patterns::
2535 0x7030 0000 0000 <reg:16>
2537 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2538 patterns depending on whether they're 32-bit or 64-bit registers::
2540 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2541 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2543 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2544 versions of the EntryLo registers regardless of the word size of the host
2545 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2546 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2547 the PFNX field starting at bit 30.
2549 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2552 0x7030 0000 0001 01 <reg:8>
2554 MIPS KVM control registers (see above) have the following id bit patterns::
2556 0x7030 0000 0002 <reg:16>
2558 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2559 id bit patterns depending on the size of the register being accessed. They are
2560 always accessed according to the current guest FPU mode (Status.FR and
2561 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2562 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2563 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2564 overlap the FPU registers::
2566 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2567 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2568 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2570 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2571 following id bit patterns::
2573 0x7020 0000 0003 01 <0:3> <reg:5>
2575 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2576 following id bit patterns::
2578 0x7020 0000 0003 02 <0:3> <reg:5>
2581 4.69 KVM_GET_ONE_REG
2582 --------------------
2584 :Capability: KVM_CAP_ONE_REG
2587 :Parameters: struct kvm_one_reg (in and out)
2588 :Returns: 0 on success, negative value on failure
2592 ======== ============================================================
2593 ENOENT no such register
2594 EINVAL invalid register ID, or no such register or used with VMs in
2595 protected virtualization mode on s390
2596 EPERM (arm64) register access not allowed before vcpu finalization
2597 ======== ============================================================
2599 (These error codes are indicative only: do not rely on a specific error
2600 code being returned in a specific situation.)
2602 This ioctl allows to receive the value of a single register implemented
2603 in a vcpu. The register to read is indicated by the "id" field of the
2604 kvm_one_reg struct passed in. On success, the register value can be found
2605 at the memory location pointed to by "addr".
2607 The list of registers accessible using this interface is identical to the
2611 4.70 KVM_KVMCLOCK_CTRL
2612 ----------------------
2614 :Capability: KVM_CAP_KVMCLOCK_CTRL
2615 :Architectures: Any that implement pvclocks (currently x86 only)
2618 :Returns: 0 on success, -1 on error
2620 This ioctl sets a flag accessible to the guest indicating that the specified
2621 vCPU has been paused by the host userspace.
2623 The host will set a flag in the pvclock structure that is checked from the
2624 soft lockup watchdog. The flag is part of the pvclock structure that is
2625 shared between guest and host, specifically the second bit of the flags
2626 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2627 the host and read/cleared exclusively by the guest. The guest operation of
2628 checking and clearing the flag must be an atomic operation so
2629 load-link/store-conditional, or equivalent must be used. There are two cases
2630 where the guest will clear the flag: when the soft lockup watchdog timer resets
2631 itself or when a soft lockup is detected. This ioctl can be called any time
2632 after pausing the vcpu, but before it is resumed.
2638 :Capability: KVM_CAP_SIGNAL_MSI
2639 :Architectures: x86 arm arm64
2641 :Parameters: struct kvm_msi (in)
2642 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2644 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2659 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2660 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2661 the device ID. If this capability is not available, userspace
2662 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2664 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2665 for the device that wrote the MSI message. For PCI, this is usually a
2666 BFD identifier in the lower 16 bits.
2668 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2669 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2670 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2671 address_hi must be zero.
2674 4.71 KVM_CREATE_PIT2
2675 --------------------
2677 :Capability: KVM_CAP_PIT2
2680 :Parameters: struct kvm_pit_config (in)
2681 :Returns: 0 on success, -1 on error
2683 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2684 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2685 parameters have to be passed::
2687 struct kvm_pit_config {
2694 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2696 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2697 exists, this thread will have a name of the following pattern::
2699 kvm-pit/<owner-process-pid>
2701 When running a guest with elevated priorities, the scheduling parameters of
2702 this thread may have to be adjusted accordingly.
2704 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2710 :Capability: KVM_CAP_PIT_STATE2
2713 :Parameters: struct kvm_pit_state2 (out)
2714 :Returns: 0 on success, -1 on error
2716 Retrieves the state of the in-kernel PIT model. Only valid after
2717 KVM_CREATE_PIT2. The state is returned in the following structure::
2719 struct kvm_pit_state2 {
2720 struct kvm_pit_channel_state channels[3];
2727 /* disable PIT in HPET legacy mode */
2728 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2730 This IOCTL replaces the obsolete KVM_GET_PIT.
2736 :Capability: KVM_CAP_PIT_STATE2
2739 :Parameters: struct kvm_pit_state2 (in)
2740 :Returns: 0 on success, -1 on error
2742 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2743 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2745 This IOCTL replaces the obsolete KVM_SET_PIT.
2748 4.74 KVM_PPC_GET_SMMU_INFO
2749 --------------------------
2751 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2752 :Architectures: powerpc
2755 :Returns: 0 on success, -1 on error
2757 This populates and returns a structure describing the features of
2758 the "Server" class MMU emulation supported by KVM.
2759 This can in turn be used by userspace to generate the appropriate
2760 device-tree properties for the guest operating system.
2762 The structure contains some global information, followed by an
2763 array of supported segment page sizes::
2765 struct kvm_ppc_smmu_info {
2769 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2772 The supported flags are:
2774 - KVM_PPC_PAGE_SIZES_REAL:
2775 When that flag is set, guest page sizes must "fit" the backing
2776 store page sizes. When not set, any page size in the list can
2777 be used regardless of how they are backed by userspace.
2779 - KVM_PPC_1T_SEGMENTS
2780 The emulated MMU supports 1T segments in addition to the
2784 This flag indicates that HPT guests are not supported by KVM,
2785 thus all guests must use radix MMU mode.
2787 The "slb_size" field indicates how many SLB entries are supported
2789 The "sps" array contains 8 entries indicating the supported base
2790 page sizes for a segment in increasing order. Each entry is defined
2793 struct kvm_ppc_one_seg_page_size {
2794 __u32 page_shift; /* Base page shift of segment (or 0) */
2795 __u32 slb_enc; /* SLB encoding for BookS */
2796 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2799 An entry with a "page_shift" of 0 is unused. Because the array is
2800 organized in increasing order, a lookup can stop when encoutering
2803 The "slb_enc" field provides the encoding to use in the SLB for the
2804 page size. The bits are in positions such as the value can directly
2805 be OR'ed into the "vsid" argument of the slbmte instruction.
2807 The "enc" array is a list which for each of those segment base page
2808 size provides the list of supported actual page sizes (which can be
2809 only larger or equal to the base page size), along with the
2810 corresponding encoding in the hash PTE. Similarly, the array is
2811 8 entries sorted by increasing sizes and an entry with a "0" shift
2812 is an empty entry and a terminator::
2814 struct kvm_ppc_one_page_size {
2815 __u32 page_shift; /* Page shift (or 0) */
2816 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
2819 The "pte_enc" field provides a value that can OR'ed into the hash
2820 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2821 into the hash PTE second double word).
2826 :Capability: KVM_CAP_IRQFD
2827 :Architectures: x86 s390 arm arm64
2829 :Parameters: struct kvm_irqfd (in)
2830 :Returns: 0 on success, -1 on error
2832 Allows setting an eventfd to directly trigger a guest interrupt.
2833 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2834 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
2835 an event is triggered on the eventfd, an interrupt is injected into
2836 the guest using the specified gsi pin. The irqfd is removed using
2837 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2840 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2841 mechanism allowing emulation of level-triggered, irqfd-based
2842 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2843 additional eventfd in the kvm_irqfd.resamplefd field. When operating
2844 in resample mode, posting of an interrupt through kvm_irq.fd asserts
2845 the specified gsi in the irqchip. When the irqchip is resampled, such
2846 as from an EOI, the gsi is de-asserted and the user is notified via
2847 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
2848 the interrupt if the device making use of it still requires service.
2849 Note that closing the resamplefd is not sufficient to disable the
2850 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2851 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2853 On arm/arm64, gsi routing being supported, the following can happen:
2855 - in case no routing entry is associated to this gsi, injection fails
2856 - in case the gsi is associated to an irqchip routing entry,
2857 irqchip.pin + 32 corresponds to the injected SPI ID.
2858 - in case the gsi is associated to an MSI routing entry, the MSI
2859 message and device ID are translated into an LPI (support restricted
2860 to GICv3 ITS in-kernel emulation).
2862 4.76 KVM_PPC_ALLOCATE_HTAB
2863 --------------------------
2865 :Capability: KVM_CAP_PPC_ALLOC_HTAB
2866 :Architectures: powerpc
2868 :Parameters: Pointer to u32 containing hash table order (in/out)
2869 :Returns: 0 on success, -1 on error
2871 This requests the host kernel to allocate an MMU hash table for a
2872 guest using the PAPR paravirtualization interface. This only does
2873 anything if the kernel is configured to use the Book 3S HV style of
2874 virtualization. Otherwise the capability doesn't exist and the ioctl
2875 returns an ENOTTY error. The rest of this description assumes Book 3S
2878 There must be no vcpus running when this ioctl is called; if there
2879 are, it will do nothing and return an EBUSY error.
2881 The parameter is a pointer to a 32-bit unsigned integer variable
2882 containing the order (log base 2) of the desired size of the hash
2883 table, which must be between 18 and 46. On successful return from the
2884 ioctl, the value will not be changed by the kernel.
2886 If no hash table has been allocated when any vcpu is asked to run
2887 (with the KVM_RUN ioctl), the host kernel will allocate a
2888 default-sized hash table (16 MB).
2890 If this ioctl is called when a hash table has already been allocated,
2891 with a different order from the existing hash table, the existing hash
2892 table will be freed and a new one allocated. If this is ioctl is
2893 called when a hash table has already been allocated of the same order
2894 as specified, the kernel will clear out the existing hash table (zero
2895 all HPTEs). In either case, if the guest is using the virtualized
2896 real-mode area (VRMA) facility, the kernel will re-create the VMRA
2897 HPTEs on the next KVM_RUN of any vcpu.
2899 4.77 KVM_S390_INTERRUPT
2900 -----------------------
2903 :Architectures: s390
2904 :Type: vm ioctl, vcpu ioctl
2905 :Parameters: struct kvm_s390_interrupt (in)
2906 :Returns: 0 on success, -1 on error
2908 Allows to inject an interrupt to the guest. Interrupts can be floating
2909 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2911 Interrupt parameters are passed via kvm_s390_interrupt::
2913 struct kvm_s390_interrupt {
2919 type can be one of the following:
2921 KVM_S390_SIGP_STOP (vcpu)
2922 - sigp stop; optional flags in parm
2923 KVM_S390_PROGRAM_INT (vcpu)
2924 - program check; code in parm
2925 KVM_S390_SIGP_SET_PREFIX (vcpu)
2926 - sigp set prefix; prefix address in parm
2927 KVM_S390_RESTART (vcpu)
2929 KVM_S390_INT_CLOCK_COMP (vcpu)
2930 - clock comparator interrupt
2931 KVM_S390_INT_CPU_TIMER (vcpu)
2932 - CPU timer interrupt
2933 KVM_S390_INT_VIRTIO (vm)
2934 - virtio external interrupt; external interrupt
2935 parameters in parm and parm64
2936 KVM_S390_INT_SERVICE (vm)
2937 - sclp external interrupt; sclp parameter in parm
2938 KVM_S390_INT_EMERGENCY (vcpu)
2939 - sigp emergency; source cpu in parm
2940 KVM_S390_INT_EXTERNAL_CALL (vcpu)
2941 - sigp external call; source cpu in parm
2942 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
2943 - compound value to indicate an
2944 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2945 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2946 interruption subclass)
2947 KVM_S390_MCHK (vm, vcpu)
2948 - machine check interrupt; cr 14 bits in parm, machine check interrupt
2949 code in parm64 (note that machine checks needing further payload are not
2950 supported by this ioctl)
2952 This is an asynchronous vcpu ioctl and can be invoked from any thread.
2954 4.78 KVM_PPC_GET_HTAB_FD
2955 ------------------------
2957 :Capability: KVM_CAP_PPC_HTAB_FD
2958 :Architectures: powerpc
2960 :Parameters: Pointer to struct kvm_get_htab_fd (in)
2961 :Returns: file descriptor number (>= 0) on success, -1 on error
2963 This returns a file descriptor that can be used either to read out the
2964 entries in the guest's hashed page table (HPT), or to write entries to
2965 initialize the HPT. The returned fd can only be written to if the
2966 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2967 can only be read if that bit is clear. The argument struct looks like
2970 /* For KVM_PPC_GET_HTAB_FD */
2971 struct kvm_get_htab_fd {
2977 /* Values for kvm_get_htab_fd.flags */
2978 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
2979 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
2981 The 'start_index' field gives the index in the HPT of the entry at
2982 which to start reading. It is ignored when writing.
2984 Reads on the fd will initially supply information about all
2985 "interesting" HPT entries. Interesting entries are those with the
2986 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2987 all entries. When the end of the HPT is reached, the read() will
2988 return. If read() is called again on the fd, it will start again from
2989 the beginning of the HPT, but will only return HPT entries that have
2990 changed since they were last read.
2992 Data read or written is structured as a header (8 bytes) followed by a
2993 series of valid HPT entries (16 bytes) each. The header indicates how
2994 many valid HPT entries there are and how many invalid entries follow
2995 the valid entries. The invalid entries are not represented explicitly
2996 in the stream. The header format is::
2998 struct kvm_get_htab_header {
3004 Writes to the fd create HPT entries starting at the index given in the
3005 header; first 'n_valid' valid entries with contents from the data
3006 written, then 'n_invalid' invalid entries, invalidating any previously
3007 valid entries found.
3009 4.79 KVM_CREATE_DEVICE
3010 ----------------------
3012 :Capability: KVM_CAP_DEVICE_CTRL
3014 :Parameters: struct kvm_create_device (in/out)
3015 :Returns: 0 on success, -1 on error
3019 ====== =======================================================
3020 ENODEV The device type is unknown or unsupported
3021 EEXIST Device already created, and this type of device may not
3022 be instantiated multiple times
3023 ====== =======================================================
3025 Other error conditions may be defined by individual device types or
3026 have their standard meanings.
3028 Creates an emulated device in the kernel. The file descriptor returned
3029 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3031 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3032 device type is supported (not necessarily whether it can be created
3035 Individual devices should not define flags. Attributes should be used
3036 for specifying any behavior that is not implied by the device type
3041 struct kvm_create_device {
3042 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3043 __u32 fd; /* out: device handle */
3044 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3047 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3048 --------------------------------------------
3050 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3051 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3052 :Type: device ioctl, vm ioctl, vcpu ioctl
3053 :Parameters: struct kvm_device_attr
3054 :Returns: 0 on success, -1 on error
3058 ===== =============================================================
3059 ENXIO The group or attribute is unknown/unsupported for this device
3060 or hardware support is missing.
3061 EPERM The attribute cannot (currently) be accessed this way
3062 (e.g. read-only attribute, or attribute that only makes
3063 sense when the device is in a different state)
3064 ===== =============================================================
3066 Other error conditions may be defined by individual device types.
3068 Gets/sets a specified piece of device configuration and/or state. The
3069 semantics are device-specific. See individual device documentation in
3070 the "devices" directory. As with ONE_REG, the size of the data
3071 transferred is defined by the particular attribute.
3075 struct kvm_device_attr {
3076 __u32 flags; /* no flags currently defined */
3077 __u32 group; /* device-defined */
3078 __u64 attr; /* group-defined */
3079 __u64 addr; /* userspace address of attr data */
3082 4.81 KVM_HAS_DEVICE_ATTR
3083 ------------------------
3085 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3086 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3087 :Type: device ioctl, vm ioctl, vcpu ioctl
3088 :Parameters: struct kvm_device_attr
3089 :Returns: 0 on success, -1 on error
3093 ===== =============================================================
3094 ENXIO The group or attribute is unknown/unsupported for this device
3095 or hardware support is missing.
3096 ===== =============================================================
3098 Tests whether a device supports a particular attribute. A successful
3099 return indicates the attribute is implemented. It does not necessarily
3100 indicate that the attribute can be read or written in the device's
3101 current state. "addr" is ignored.
3103 4.82 KVM_ARM_VCPU_INIT
3104 ----------------------
3107 :Architectures: arm, arm64
3109 :Parameters: struct kvm_vcpu_init (in)
3110 :Returns: 0 on success; -1 on error
3114 ====== =================================================================
3115 EINVAL the target is unknown, or the combination of features is invalid.
3116 ENOENT a features bit specified is unknown.
3117 ====== =================================================================
3119 This tells KVM what type of CPU to present to the guest, and what
3120 optional features it should have. This will cause a reset of the cpu
3121 registers to their initial values. If this is not called, KVM_RUN will
3122 return ENOEXEC for that vcpu.
3124 The initial values are defined as:
3126 * AArch64: EL1h, D, A, I and F bits set. All other bits
3128 * AArch32: SVC, A, I and F bits set. All other bits are
3130 - General Purpose registers, including PC and SP: set to 0
3131 - FPSIMD/NEON registers: set to 0
3132 - SVE registers: set to 0
3133 - System registers: Reset to their architecturally defined
3134 values as for a warm reset to EL1 (resp. SVC)
3136 Note that because some registers reflect machine topology, all vcpus
3137 should be created before this ioctl is invoked.
3139 Userspace can call this function multiple times for a given vcpu, including
3140 after the vcpu has been run. This will reset the vcpu to its initial
3141 state. All calls to this function after the initial call must use the same
3142 target and same set of feature flags, otherwise EINVAL will be returned.
3146 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3147 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3148 and execute guest code when KVM_RUN is called.
3149 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3150 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3151 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3152 backward compatible with v0.2) for the CPU.
3153 Depends on KVM_CAP_ARM_PSCI_0_2.
3154 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3155 Depends on KVM_CAP_ARM_PMU_V3.
3157 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3159 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3160 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3161 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3162 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3165 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3167 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3168 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3169 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3170 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3173 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3174 Depends on KVM_CAP_ARM_SVE.
3175 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3177 * After KVM_ARM_VCPU_INIT:
3179 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3180 initial value of this pseudo-register indicates the best set of
3181 vector lengths possible for a vcpu on this host.
3183 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3185 - KVM_RUN and KVM_GET_REG_LIST are not available;
3187 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3188 the scalable archietctural SVE registers
3189 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3190 KVM_REG_ARM64_SVE_FFR;
3192 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3193 KVM_SET_ONE_REG, to modify the set of vector lengths available
3196 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3198 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3199 no longer be written using KVM_SET_ONE_REG.
3201 4.83 KVM_ARM_PREFERRED_TARGET
3202 -----------------------------
3205 :Architectures: arm, arm64
3207 :Parameters: struct kvm_vcpu_init (out)
3208 :Returns: 0 on success; -1 on error
3212 ====== ==========================================
3213 ENODEV no preferred target available for the host
3214 ====== ==========================================
3216 This queries KVM for preferred CPU target type which can be emulated
3217 by KVM on underlying host.
3219 The ioctl returns struct kvm_vcpu_init instance containing information
3220 about preferred CPU target type and recommended features for it. The
3221 kvm_vcpu_init->features bitmap returned will have feature bits set if
3222 the preferred target recommends setting these features, but this is
3225 The information returned by this ioctl can be used to prepare an instance
3226 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3227 VCPU matching underlying host.
3230 4.84 KVM_GET_REG_LIST
3231 ---------------------
3234 :Architectures: arm, arm64, mips
3236 :Parameters: struct kvm_reg_list (in/out)
3237 :Returns: 0 on success; -1 on error
3241 ===== ==============================================================
3242 E2BIG the reg index list is too big to fit in the array specified by
3243 the user (the number required will be written into n).
3244 ===== ==============================================================
3248 struct kvm_reg_list {
3249 __u64 n; /* number of registers in reg[] */
3253 This ioctl returns the guest registers that are supported for the
3254 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3257 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3258 -----------------------------------------
3260 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3261 :Architectures: arm, arm64
3263 :Parameters: struct kvm_arm_device_address (in)
3264 :Returns: 0 on success, -1 on error
3268 ====== ============================================
3269 ENODEV The device id is unknown
3270 ENXIO Device not supported on current system
3271 EEXIST Address already set
3272 E2BIG Address outside guest physical address space
3273 EBUSY Address overlaps with other device range
3274 ====== ============================================
3278 struct kvm_arm_device_addr {
3283 Specify a device address in the guest's physical address space where guests
3284 can access emulated or directly exposed devices, which the host kernel needs
3285 to know about. The id field is an architecture specific identifier for a
3288 ARM/arm64 divides the id field into two parts, a device id and an
3289 address type id specific to the individual device::
3291 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3292 field: | 0x00000000 | device id | addr type id |
3294 ARM/arm64 currently only require this when using the in-kernel GIC
3295 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3296 as the device id. When setting the base address for the guest's
3297 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3298 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3299 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3300 base addresses will return -EEXIST.
3302 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3303 should be used instead.
3306 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3307 ------------------------------
3309 :Capability: KVM_CAP_PPC_RTAS
3312 :Parameters: struct kvm_rtas_token_args
3313 :Returns: 0 on success, -1 on error
3315 Defines a token value for a RTAS (Run Time Abstraction Services)
3316 service in order to allow it to be handled in the kernel. The
3317 argument struct gives the name of the service, which must be the name
3318 of a service that has a kernel-side implementation. If the token
3319 value is non-zero, it will be associated with that service, and
3320 subsequent RTAS calls by the guest specifying that token will be
3321 handled by the kernel. If the token value is 0, then any token
3322 associated with the service will be forgotten, and subsequent RTAS
3323 calls by the guest for that service will be passed to userspace to be
3326 4.87 KVM_SET_GUEST_DEBUG
3327 ------------------------
3329 :Capability: KVM_CAP_SET_GUEST_DEBUG
3330 :Architectures: x86, s390, ppc, arm64
3332 :Parameters: struct kvm_guest_debug (in)
3333 :Returns: 0 on success; -1 on error
3337 struct kvm_guest_debug {
3340 struct kvm_guest_debug_arch arch;
3343 Set up the processor specific debug registers and configure vcpu for
3344 handling guest debug events. There are two parts to the structure, the
3345 first a control bitfield indicates the type of debug events to handle
3346 when running. Common control bits are:
3348 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3349 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3351 The top 16 bits of the control field are architecture specific control
3352 flags which can include the following:
3354 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3355 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390]
3356 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64]
3357 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3358 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3359 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3361 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3362 are enabled in memory so we need to ensure breakpoint exceptions are
3363 correctly trapped and the KVM run loop exits at the breakpoint and not
3364 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3365 we need to ensure the guest vCPUs architecture specific registers are
3366 updated to the correct (supplied) values.
3368 The second part of the structure is architecture specific and
3369 typically contains a set of debug registers.
3371 For arm64 the number of debug registers is implementation defined and
3372 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3373 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3374 indicating the number of supported registers.
3376 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3377 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3379 Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3380 supported KVM_GUESTDBG_* bits in the control field.
3382 When debug events exit the main run loop with the reason
3383 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3384 structure containing architecture specific debug information.
3386 4.88 KVM_GET_EMULATED_CPUID
3387 ---------------------------
3389 :Capability: KVM_CAP_EXT_EMUL_CPUID
3392 :Parameters: struct kvm_cpuid2 (in/out)
3393 :Returns: 0 on success, -1 on error
3400 struct kvm_cpuid_entry2 entries[0];
3403 The member 'flags' is used for passing flags from userspace.
3407 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3408 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3409 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3411 struct kvm_cpuid_entry2 {
3422 This ioctl returns x86 cpuid features which are emulated by
3423 kvm.Userspace can use the information returned by this ioctl to query
3424 which features are emulated by kvm instead of being present natively.
3426 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3427 structure with the 'nent' field indicating the number of entries in
3428 the variable-size array 'entries'. If the number of entries is too low
3429 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3430 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3431 is returned. If the number is just right, the 'nent' field is adjusted
3432 to the number of valid entries in the 'entries' array, which is then
3435 The entries returned are the set CPUID bits of the respective features
3436 which kvm emulates, as returned by the CPUID instruction, with unknown
3437 or unsupported feature bits cleared.
3439 Features like x2apic, for example, may not be present in the host cpu
3440 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3441 emulated efficiently and thus not included here.
3443 The fields in each entry are defined as follows:
3446 the eax value used to obtain the entry
3448 the ecx value used to obtain the entry (for entries that are
3451 an OR of zero or more of the following:
3453 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3454 if the index field is valid
3458 the values returned by the cpuid instruction for
3459 this function/index combination
3461 4.89 KVM_S390_MEM_OP
3462 --------------------
3464 :Capability: KVM_CAP_S390_MEM_OP
3465 :Architectures: s390
3467 :Parameters: struct kvm_s390_mem_op (in)
3468 :Returns: = 0 on success,
3469 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3470 > 0 if an exception occurred while walking the page tables
3472 Read or write data from/to the logical (virtual) memory of a VCPU.
3474 Parameters are specified via the following structure::
3476 struct kvm_s390_mem_op {
3477 __u64 gaddr; /* the guest address */
3478 __u64 flags; /* flags */
3479 __u32 size; /* amount of bytes */
3480 __u32 op; /* type of operation */
3481 __u64 buf; /* buffer in userspace */
3482 __u8 ar; /* the access register number */
3483 __u8 reserved[31]; /* should be set to 0 */
3486 The type of operation is specified in the "op" field. It is either
3487 KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3488 KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3489 KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3490 whether the corresponding memory access would create an access exception
3491 (without touching the data in the memory at the destination). In case an
3492 access exception occurred while walking the MMU tables of the guest, the
3493 ioctl returns a positive error number to indicate the type of exception.
3494 This exception is also raised directly at the corresponding VCPU if the
3495 flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3497 The start address of the memory region has to be specified in the "gaddr"
3498 field, and the length of the region in the "size" field (which must not
3499 be 0). The maximum value for "size" can be obtained by checking the
3500 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3501 userspace application where the read data should be written to for
3502 KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3503 stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3504 is specified, "buf" is unused and can be NULL. "ar" designates the access
3505 register number to be used; the valid range is 0..15.
3507 The "reserved" field is meant for future extensions. It is not used by
3508 KVM with the currently defined set of flags.
3510 4.90 KVM_S390_GET_SKEYS
3511 -----------------------
3513 :Capability: KVM_CAP_S390_SKEYS
3514 :Architectures: s390
3516 :Parameters: struct kvm_s390_skeys
3517 :Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3518 keys, negative value on error
3520 This ioctl is used to get guest storage key values on the s390
3521 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3523 struct kvm_s390_skeys {
3526 __u64 skeydata_addr;
3531 The start_gfn field is the number of the first guest frame whose storage keys
3534 The count field is the number of consecutive frames (starting from start_gfn)
3535 whose storage keys to get. The count field must be at least 1 and the maximum
3536 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3537 will cause the ioctl to return -EINVAL.
3539 The skeydata_addr field is the address to a buffer large enough to hold count
3540 bytes. This buffer will be filled with storage key data by the ioctl.
3542 4.91 KVM_S390_SET_SKEYS
3543 -----------------------
3545 :Capability: KVM_CAP_S390_SKEYS
3546 :Architectures: s390
3548 :Parameters: struct kvm_s390_skeys
3549 :Returns: 0 on success, negative value on error
3551 This ioctl is used to set guest storage key values on the s390
3552 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3553 See section on KVM_S390_GET_SKEYS for struct definition.
3555 The start_gfn field is the number of the first guest frame whose storage keys
3558 The count field is the number of consecutive frames (starting from start_gfn)
3559 whose storage keys to get. The count field must be at least 1 and the maximum
3560 allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3561 will cause the ioctl to return -EINVAL.
3563 The skeydata_addr field is the address to a buffer containing count bytes of
3564 storage keys. Each byte in the buffer will be set as the storage key for a
3565 single frame starting at start_gfn for count frames.
3567 Note: If any architecturally invalid key value is found in the given data then
3568 the ioctl will return -EINVAL.
3573 :Capability: KVM_CAP_S390_INJECT_IRQ
3574 :Architectures: s390
3576 :Parameters: struct kvm_s390_irq (in)
3577 :Returns: 0 on success, -1 on error
3582 ====== =================================================================
3583 EINVAL interrupt type is invalid
3584 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3585 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3586 than the maximum of VCPUs
3587 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3588 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3589 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3591 ====== =================================================================
3593 Allows to inject an interrupt to the guest.
3595 Using struct kvm_s390_irq as a parameter allows
3596 to inject additional payload which is not
3597 possible via KVM_S390_INTERRUPT.
3599 Interrupt parameters are passed via kvm_s390_irq::
3601 struct kvm_s390_irq {
3604 struct kvm_s390_io_info io;
3605 struct kvm_s390_ext_info ext;
3606 struct kvm_s390_pgm_info pgm;
3607 struct kvm_s390_emerg_info emerg;
3608 struct kvm_s390_extcall_info extcall;
3609 struct kvm_s390_prefix_info prefix;
3610 struct kvm_s390_stop_info stop;
3611 struct kvm_s390_mchk_info mchk;
3616 type can be one of the following:
3618 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3619 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3620 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3621 - KVM_S390_RESTART - restart; no parameters
3622 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3623 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3624 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3625 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3626 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3628 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3630 4.94 KVM_S390_GET_IRQ_STATE
3631 ---------------------------
3633 :Capability: KVM_CAP_S390_IRQ_STATE
3634 :Architectures: s390
3636 :Parameters: struct kvm_s390_irq_state (out)
3637 :Returns: >= number of bytes copied into buffer,
3638 -EINVAL if buffer size is 0,
3639 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3640 -EFAULT if the buffer address was invalid
3642 This ioctl allows userspace to retrieve the complete state of all currently
3643 pending interrupts in a single buffer. Use cases include migration
3644 and introspection. The parameter structure contains the address of a
3645 userspace buffer and its length::
3647 struct kvm_s390_irq_state {
3649 __u32 flags; /* will stay unused for compatibility reasons */
3651 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3654 Userspace passes in the above struct and for each pending interrupt a
3655 struct kvm_s390_irq is copied to the provided buffer.
3657 The structure contains a flags and a reserved field for future extensions. As
3658 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3659 reserved, these fields can not be used in the future without breaking
3662 If -ENOBUFS is returned the buffer provided was too small and userspace
3663 may retry with a bigger buffer.
3665 4.95 KVM_S390_SET_IRQ_STATE
3666 ---------------------------
3668 :Capability: KVM_CAP_S390_IRQ_STATE
3669 :Architectures: s390
3671 :Parameters: struct kvm_s390_irq_state (in)
3672 :Returns: 0 on success,
3673 -EFAULT if the buffer address was invalid,
3674 -EINVAL for an invalid buffer length (see below),
3675 -EBUSY if there were already interrupts pending,
3676 errors occurring when actually injecting the
3677 interrupt. See KVM_S390_IRQ.
3679 This ioctl allows userspace to set the complete state of all cpu-local
3680 interrupts currently pending for the vcpu. It is intended for restoring
3681 interrupt state after a migration. The input parameter is a userspace buffer
3682 containing a struct kvm_s390_irq_state::
3684 struct kvm_s390_irq_state {
3686 __u32 flags; /* will stay unused for compatibility reasons */
3688 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3691 The restrictions for flags and reserved apply as well.
3692 (see KVM_S390_GET_IRQ_STATE)
3694 The userspace memory referenced by buf contains a struct kvm_s390_irq
3695 for each interrupt to be injected into the guest.
3696 If one of the interrupts could not be injected for some reason the
3699 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3700 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3701 which is the maximum number of possibly pending cpu-local interrupts.
3706 :Capability: KVM_CAP_X86_SMM
3710 :Returns: 0 on success, -1 on error
3712 Queues an SMI on the thread's vcpu.
3714 4.97 KVM_X86_SET_MSR_FILTER
3715 ----------------------------
3717 :Capability: KVM_X86_SET_MSR_FILTER
3720 :Parameters: struct kvm_msr_filter
3721 :Returns: 0 on success, < 0 on error
3725 struct kvm_msr_filter_range {
3726 #define KVM_MSR_FILTER_READ (1 << 0)
3727 #define KVM_MSR_FILTER_WRITE (1 << 1)
3729 __u32 nmsrs; /* number of msrs in bitmap */
3730 __u32 base; /* MSR index the bitmap starts at */
3731 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
3734 #define KVM_MSR_FILTER_MAX_RANGES 16
3735 struct kvm_msr_filter {
3736 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
3737 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
3739 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
3742 flags values for ``struct kvm_msr_filter_range``:
3744 ``KVM_MSR_FILTER_READ``
3746 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
3747 indicates that a read should immediately fail, while a 1 indicates that
3748 a read for a particular MSR should be handled regardless of the default
3751 ``KVM_MSR_FILTER_WRITE``
3753 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
3754 indicates that a write should immediately fail, while a 1 indicates that
3755 a write for a particular MSR should be handled regardless of the default
3758 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
3760 Filter both read and write accesses to MSRs using the given bitmap. A 0
3761 in the bitmap indicates that both reads and writes should immediately fail,
3762 while a 1 indicates that reads and writes for a particular MSR are not
3763 filtered by this range.
3765 flags values for ``struct kvm_msr_filter``:
3767 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
3769 If no filter range matches an MSR index that is getting accessed, KVM will
3770 fall back to allowing access to the MSR.
3772 ``KVM_MSR_FILTER_DEFAULT_DENY``
3774 If no filter range matches an MSR index that is getting accessed, KVM will
3775 fall back to rejecting access to the MSR. In this mode, all MSRs that should
3776 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
3778 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
3779 specify whether a certain MSR access should be explicitly filtered for or not.
3781 If this ioctl has never been invoked, MSR accesses are not guarded and the
3782 default KVM in-kernel emulation behavior is fully preserved.
3784 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
3785 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
3788 As soon as the filtering is in place, every MSR access is processed through
3789 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
3790 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
3791 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
3794 If a bit is within one of the defined ranges, read and write accesses are
3795 guarded by the bitmap's value for the MSR index if the kind of access
3796 is included in the ``struct kvm_msr_filter_range`` flags. If no range
3797 cover this particular access, the behavior is determined by the flags
3798 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
3799 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
3801 Each bitmap range specifies a range of MSRs to potentially allow access on.
3802 The range goes from MSR index [base .. base+nmsrs]. The flags field
3803 indicates whether reads, writes or both reads and writes are filtered
3804 by setting a 1 bit in the bitmap for the corresponding MSR index.
3806 If an MSR access is not permitted through the filtering, it generates a
3807 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
3808 allows user space to deflect and potentially handle various MSR accesses
3811 If a vCPU is in running state while this ioctl is invoked, the vCPU may
3812 experience inconsistent filtering behavior on MSR accesses.
3814 4.98 KVM_CREATE_SPAPR_TCE_64
3815 ----------------------------
3817 :Capability: KVM_CAP_SPAPR_TCE_64
3818 :Architectures: powerpc
3820 :Parameters: struct kvm_create_spapr_tce_64 (in)
3821 :Returns: file descriptor for manipulating the created TCE table
3823 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3824 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3826 This capability uses extended struct in ioctl interface::
3828 /* for KVM_CAP_SPAPR_TCE_64 */
3829 struct kvm_create_spapr_tce_64 {
3833 __u64 offset; /* in pages */
3834 __u64 size; /* in pages */
3837 The aim of extension is to support an additional bigger DMA window with
3838 a variable page size.
3839 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3840 a bus offset of the corresponding DMA window, @size and @offset are numbers
3843 @flags are not used at the moment.
3845 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3847 4.99 KVM_REINJECT_CONTROL
3848 -------------------------
3850 :Capability: KVM_CAP_REINJECT_CONTROL
3853 :Parameters: struct kvm_reinject_control (in)
3854 :Returns: 0 on success,
3855 -EFAULT if struct kvm_reinject_control cannot be read,
3856 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3858 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
3859 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3860 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
3861 interrupt whenever there isn't a pending interrupt from i8254.
3862 !reinject mode injects an interrupt as soon as a tick arrives.
3866 struct kvm_reinject_control {
3871 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3872 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3874 4.100 KVM_PPC_CONFIGURE_V3_MMU
3875 ------------------------------
3877 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3880 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
3881 :Returns: 0 on success,
3882 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3883 -EINVAL if the configuration is invalid
3885 This ioctl controls whether the guest will use radix or HPT (hashed
3886 page table) translation, and sets the pointer to the process table for
3891 struct kvm_ppc_mmuv3_cfg {
3893 __u64 process_table;
3896 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3897 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
3898 to use radix tree translation, and if clear, to use HPT translation.
3899 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3900 to be able to use the global TLB and SLB invalidation instructions;
3901 if clear, the guest may not use these instructions.
3903 The process_table field specifies the address and size of the guest
3904 process table, which is in the guest's space. This field is formatted
3905 as the second doubleword of the partition table entry, as defined in
3906 the Power ISA V3.00, Book III section 5.7.6.1.
3908 4.101 KVM_PPC_GET_RMMU_INFO
3909 ---------------------------
3911 :Capability: KVM_CAP_PPC_RADIX_MMU
3914 :Parameters: struct kvm_ppc_rmmu_info (out)
3915 :Returns: 0 on success,
3916 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3917 -EINVAL if no useful information can be returned
3919 This ioctl returns a structure containing two things: (a) a list
3920 containing supported radix tree geometries, and (b) a list that maps
3921 page sizes to put in the "AP" (actual page size) field for the tlbie
3922 (TLB invalidate entry) instruction.
3926 struct kvm_ppc_rmmu_info {
3927 struct kvm_ppc_radix_geom {
3932 __u32 ap_encodings[8];
3935 The geometries[] field gives up to 8 supported geometries for the
3936 radix page table, in terms of the log base 2 of the smallest page
3937 size, and the number of bits indexed at each level of the tree, from
3938 the PTE level up to the PGD level in that order. Any unused entries
3939 will have 0 in the page_shift field.
3941 The ap_encodings gives the supported page sizes and their AP field
3942 encodings, encoded with the AP value in the top 3 bits and the log
3943 base 2 of the page size in the bottom 6 bits.
3945 4.102 KVM_PPC_RESIZE_HPT_PREPARE
3946 --------------------------------
3948 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
3949 :Architectures: powerpc
3951 :Parameters: struct kvm_ppc_resize_hpt (in)
3952 :Returns: 0 on successful completion,
3953 >0 if a new HPT is being prepared, the value is an estimated
3954 number of milliseconds until preparation is complete,
3955 -EFAULT if struct kvm_reinject_control cannot be read,
3956 -EINVAL if the supplied shift or flags are invalid,
3957 -ENOMEM if unable to allocate the new HPT,
3959 Used to implement the PAPR extension for runtime resizing of a guest's
3960 Hashed Page Table (HPT). Specifically this starts, stops or monitors
3961 the preparation of a new potential HPT for the guest, essentially
3962 implementing the H_RESIZE_HPT_PREPARE hypercall.
3966 struct kvm_ppc_resize_hpt {
3972 If called with shift > 0 when there is no pending HPT for the guest,
3973 this begins preparation of a new pending HPT of size 2^(shift) bytes.
3974 It then returns a positive integer with the estimated number of
3975 milliseconds until preparation is complete.
3977 If called when there is a pending HPT whose size does not match that
3978 requested in the parameters, discards the existing pending HPT and
3979 creates a new one as above.
3981 If called when there is a pending HPT of the size requested, will:
3983 * If preparation of the pending HPT is already complete, return 0
3984 * If preparation of the pending HPT has failed, return an error
3985 code, then discard the pending HPT.
3986 * If preparation of the pending HPT is still in progress, return an
3987 estimated number of milliseconds until preparation is complete.
3989 If called with shift == 0, discards any currently pending HPT and
3990 returns 0 (i.e. cancels any in-progress preparation).
3992 flags is reserved for future expansion, currently setting any bits in
3993 flags will result in an -EINVAL.
3995 Normally this will be called repeatedly with the same parameters until
3996 it returns <= 0. The first call will initiate preparation, subsequent
3997 ones will monitor preparation until it completes or fails.
3999 4.103 KVM_PPC_RESIZE_HPT_COMMIT
4000 -------------------------------
4002 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4003 :Architectures: powerpc
4005 :Parameters: struct kvm_ppc_resize_hpt (in)
4006 :Returns: 0 on successful completion,
4007 -EFAULT if struct kvm_reinject_control cannot be read,
4008 -EINVAL if the supplied shift or flags are invalid,
4009 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4010 have the requested size,
4011 -EBUSY if the pending HPT is not fully prepared,
4012 -ENOSPC if there was a hash collision when moving existing
4013 HPT entries to the new HPT,
4014 -EIO on other error conditions
4016 Used to implement the PAPR extension for runtime resizing of a guest's
4017 Hashed Page Table (HPT). Specifically this requests that the guest be
4018 transferred to working with the new HPT, essentially implementing the
4019 H_RESIZE_HPT_COMMIT hypercall.
4023 struct kvm_ppc_resize_hpt {
4029 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4030 returned 0 with the same parameters. In other cases
4031 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4032 -EBUSY, though others may be possible if the preparation was started,
4035 This will have undefined effects on the guest if it has not already
4036 placed itself in a quiescent state where no vcpu will make MMU enabled
4039 On succsful completion, the pending HPT will become the guest's active
4040 HPT and the previous HPT will be discarded.
4042 On failure, the guest will still be operating on its previous HPT.
4044 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4045 -----------------------------------
4047 :Capability: KVM_CAP_MCE
4050 :Parameters: u64 mce_cap (out)
4051 :Returns: 0 on success, -1 on error
4053 Returns supported MCE capabilities. The u64 mce_cap parameter
4054 has the same format as the MSR_IA32_MCG_CAP register. Supported
4055 capabilities will have the corresponding bits set.
4057 4.105 KVM_X86_SETUP_MCE
4058 -----------------------
4060 :Capability: KVM_CAP_MCE
4063 :Parameters: u64 mcg_cap (in)
4064 :Returns: 0 on success,
4065 -EFAULT if u64 mcg_cap cannot be read,
4066 -EINVAL if the requested number of banks is invalid,
4067 -EINVAL if requested MCE capability is not supported.
4069 Initializes MCE support for use. The u64 mcg_cap parameter
4070 has the same format as the MSR_IA32_MCG_CAP register and
4071 specifies which capabilities should be enabled. The maximum
4072 supported number of error-reporting banks can be retrieved when
4073 checking for KVM_CAP_MCE. The supported capabilities can be
4074 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4076 4.106 KVM_X86_SET_MCE
4077 ---------------------
4079 :Capability: KVM_CAP_MCE
4082 :Parameters: struct kvm_x86_mce (in)
4083 :Returns: 0 on success,
4084 -EFAULT if struct kvm_x86_mce cannot be read,
4085 -EINVAL if the bank number is invalid,
4086 -EINVAL if VAL bit is not set in status field.
4088 Inject a machine check error (MCE) into the guest. The input
4091 struct kvm_x86_mce {
4101 If the MCE being reported is an uncorrected error, KVM will
4102 inject it as an MCE exception into the guest. If the guest
4103 MCG_STATUS register reports that an MCE is in progress, KVM
4104 causes an KVM_EXIT_SHUTDOWN vmexit.
4106 Otherwise, if the MCE is a corrected error, KVM will just
4107 store it in the corresponding bank (provided this bank is
4108 not holding a previously reported uncorrected error).
4110 4.107 KVM_S390_GET_CMMA_BITS
4111 ----------------------------
4113 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4114 :Architectures: s390
4116 :Parameters: struct kvm_s390_cmma_log (in, out)
4117 :Returns: 0 on success, a negative value on error
4119 This ioctl is used to get the values of the CMMA bits on the s390
4120 architecture. It is meant to be used in two scenarios:
4122 - During live migration to save the CMMA values. Live migration needs
4123 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4124 - To non-destructively peek at the CMMA values, with the flag
4125 KVM_S390_CMMA_PEEK set.
4127 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4128 values are written to a buffer whose location is indicated via the "values"
4129 member in the kvm_s390_cmma_log struct. The values in the input struct are
4130 also updated as needed.
4132 Each CMMA value takes up one byte.
4136 struct kvm_s390_cmma_log {
4147 start_gfn is the number of the first guest frame whose CMMA values are
4150 count is the length of the buffer in bytes,
4152 values points to the buffer where the result will be written to.
4154 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4155 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4158 The result is written in the buffer pointed to by the field values, and
4159 the values of the input parameter are updated as follows.
4161 Depending on the flags, different actions are performed. The only
4162 supported flag so far is KVM_S390_CMMA_PEEK.
4164 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4165 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4166 It is not necessarily the same as the one passed as input, as clean pages
4169 count will indicate the number of bytes actually written in the buffer.
4170 It can (and very often will) be smaller than the input value, since the
4171 buffer is only filled until 16 bytes of clean values are found (which
4172 are then not copied in the buffer). Since a CMMA migration block needs
4173 the base address and the length, for a total of 16 bytes, we will send
4174 back some clean data if there is some dirty data afterwards, as long as
4175 the size of the clean data does not exceed the size of the header. This
4176 allows to minimize the amount of data to be saved or transferred over
4177 the network at the expense of more roundtrips to userspace. The next
4178 invocation of the ioctl will skip over all the clean values, saving
4179 potentially more than just the 16 bytes we found.
4181 If KVM_S390_CMMA_PEEK is set:
4182 the existing storage attributes are read even when not in migration
4183 mode, and no other action is performed;
4185 the output start_gfn will be equal to the input start_gfn,
4187 the output count will be equal to the input count, except if the end of
4188 memory has been reached.
4191 the field "remaining" will indicate the total number of dirty CMMA values
4192 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4197 values points to the userspace buffer where the result will be stored.
4199 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4200 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4201 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4202 -EFAULT if the userspace address is invalid or if no page table is
4203 present for the addresses (e.g. when using hugepages).
4205 4.108 KVM_S390_SET_CMMA_BITS
4206 ----------------------------
4208 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4209 :Architectures: s390
4211 :Parameters: struct kvm_s390_cmma_log (in)
4212 :Returns: 0 on success, a negative value on error
4214 This ioctl is used to set the values of the CMMA bits on the s390
4215 architecture. It is meant to be used during live migration to restore
4216 the CMMA values, but there are no restrictions on its use.
4217 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4218 Each CMMA value takes up one byte.
4222 struct kvm_s390_cmma_log {
4233 start_gfn indicates the starting guest frame number,
4235 count indicates how many values are to be considered in the buffer,
4237 flags is not used and must be 0.
4239 mask indicates which PGSTE bits are to be considered.
4241 remaining is not used.
4243 values points to the buffer in userspace where to store the values.
4245 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4246 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4247 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4248 if the flags field was not 0, with -EFAULT if the userspace address is
4249 invalid, if invalid pages are written to (e.g. after the end of memory)
4250 or if no page table is present for the addresses (e.g. when using
4253 4.109 KVM_PPC_GET_CPU_CHAR
4254 --------------------------
4256 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4257 :Architectures: powerpc
4259 :Parameters: struct kvm_ppc_cpu_char (out)
4260 :Returns: 0 on successful completion,
4261 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4263 This ioctl gives userspace information about certain characteristics
4264 of the CPU relating to speculative execution of instructions and
4265 possible information leakage resulting from speculative execution (see
4266 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4267 returned in struct kvm_ppc_cpu_char, which looks like this::
4269 struct kvm_ppc_cpu_char {
4270 __u64 character; /* characteristics of the CPU */
4271 __u64 behaviour; /* recommended software behaviour */
4272 __u64 character_mask; /* valid bits in character */
4273 __u64 behaviour_mask; /* valid bits in behaviour */
4276 For extensibility, the character_mask and behaviour_mask fields
4277 indicate which bits of character and behaviour have been filled in by
4278 the kernel. If the set of defined bits is extended in future then
4279 userspace will be able to tell whether it is running on a kernel that
4280 knows about the new bits.
4282 The character field describes attributes of the CPU which can help
4283 with preventing inadvertent information disclosure - specifically,
4284 whether there is an instruction to flash-invalidate the L1 data cache
4285 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4286 to a mode where entries can only be used by the thread that created
4287 them, whether the bcctr[l] instruction prevents speculation, and
4288 whether a speculation barrier instruction (ori 31,31,0) is provided.
4290 The behaviour field describes actions that software should take to
4291 prevent inadvertent information disclosure, and thus describes which
4292 vulnerabilities the hardware is subject to; specifically whether the
4293 L1 data cache should be flushed when returning to user mode from the
4294 kernel, and whether a speculation barrier should be placed between an
4295 array bounds check and the array access.
4297 These fields use the same bit definitions as the new
4298 H_GET_CPU_CHARACTERISTICS hypercall.
4300 4.110 KVM_MEMORY_ENCRYPT_OP
4301 ---------------------------
4306 :Parameters: an opaque platform specific structure (in/out)
4307 :Returns: 0 on success; -1 on error
4309 If the platform supports creating encrypted VMs then this ioctl can be used
4310 for issuing platform-specific memory encryption commands to manage those
4313 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4314 (SEV) commands on AMD Processors. The SEV commands are defined in
4315 Documentation/virt/kvm/amd-memory-encryption.rst.
4317 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4318 -----------------------------------
4323 :Parameters: struct kvm_enc_region (in)
4324 :Returns: 0 on success; -1 on error
4326 This ioctl can be used to register a guest memory region which may
4327 contain encrypted data (e.g. guest RAM, SMRAM etc).
4329 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4330 memory region may contain encrypted data. The SEV memory encryption
4331 engine uses a tweak such that two identical plaintext pages, each at
4332 different locations will have differing ciphertexts. So swapping or
4333 moving ciphertext of those pages will not result in plaintext being
4334 swapped. So relocating (or migrating) physical backing pages for the SEV
4335 guest will require some additional steps.
4337 Note: The current SEV key management spec does not provide commands to
4338 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4339 memory region registered with the ioctl.
4341 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4342 -------------------------------------
4347 :Parameters: struct kvm_enc_region (in)
4348 :Returns: 0 on success; -1 on error
4350 This ioctl can be used to unregister the guest memory region registered
4351 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4353 4.113 KVM_HYPERV_EVENTFD
4354 ------------------------
4356 :Capability: KVM_CAP_HYPERV_EVENTFD
4359 :Parameters: struct kvm_hyperv_eventfd (in)
4361 This ioctl (un)registers an eventfd to receive notifications from the guest on
4362 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4363 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4364 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4368 struct kvm_hyperv_eventfd {
4375 The conn_id field should fit within 24 bits::
4377 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4379 The acceptable values for the flags field are::
4381 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4383 :Returns: 0 on success,
4384 -EINVAL if conn_id or flags is outside the allowed range,
4385 -ENOENT on deassign if the conn_id isn't registered,
4386 -EEXIST on assign if the conn_id is already registered
4388 4.114 KVM_GET_NESTED_STATE
4389 --------------------------
4391 :Capability: KVM_CAP_NESTED_STATE
4394 :Parameters: struct kvm_nested_state (in/out)
4395 :Returns: 0 on success, -1 on error
4399 ===== =============================================================
4400 E2BIG the total state size exceeds the value of 'size' specified by
4401 the user; the size required will be written into size.
4402 ===== =============================================================
4406 struct kvm_nested_state {
4412 struct kvm_vmx_nested_state_hdr vmx;
4413 struct kvm_svm_nested_state_hdr svm;
4415 /* Pad the header to 128 bytes. */
4420 struct kvm_vmx_nested_state_data vmx[0];
4421 struct kvm_svm_nested_state_data svm[0];
4425 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4426 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4427 #define KVM_STATE_NESTED_EVMCS 0x00000004
4429 #define KVM_STATE_NESTED_FORMAT_VMX 0
4430 #define KVM_STATE_NESTED_FORMAT_SVM 1
4432 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4434 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4435 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4437 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4439 struct kvm_vmx_nested_state_hdr {
4448 __u64 preemption_timer_deadline;
4451 struct kvm_vmx_nested_state_data {
4452 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4453 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4456 This ioctl copies the vcpu's nested virtualization state from the kernel to
4459 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4460 to the KVM_CHECK_EXTENSION ioctl().
4462 4.115 KVM_SET_NESTED_STATE
4463 --------------------------
4465 :Capability: KVM_CAP_NESTED_STATE
4468 :Parameters: struct kvm_nested_state (in)
4469 :Returns: 0 on success, -1 on error
4471 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4472 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4474 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4475 -------------------------------------
4477 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4478 KVM_CAP_COALESCED_PIO (for coalesced pio)
4481 :Parameters: struct kvm_coalesced_mmio_zone
4482 :Returns: 0 on success, < 0 on error
4484 Coalesced I/O is a performance optimization that defers hardware
4485 register write emulation so that userspace exits are avoided. It is
4486 typically used to reduce the overhead of emulating frequently accessed
4489 When a hardware register is configured for coalesced I/O, write accesses
4490 do not exit to userspace and their value is recorded in a ring buffer
4491 that is shared between kernel and userspace.
4493 Coalesced I/O is used if one or more write accesses to a hardware
4494 register can be deferred until a read or a write to another hardware
4495 register on the same device. This last access will cause a vmexit and
4496 userspace will process accesses from the ring buffer before emulating
4497 it. That will avoid exiting to userspace on repeated writes.
4499 Coalesced pio is based on coalesced mmio. There is little difference
4500 between coalesced mmio and pio except that coalesced pio records accesses
4503 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4504 ------------------------------------
4506 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4507 :Architectures: x86, arm, arm64, mips
4509 :Parameters: struct kvm_clear_dirty_log (in)
4510 :Returns: 0 on success, -1 on error
4514 /* for KVM_CLEAR_DIRTY_LOG */
4515 struct kvm_clear_dirty_log {
4520 void __user *dirty_bitmap; /* one bit per page */
4525 The ioctl clears the dirty status of pages in a memory slot, according to
4526 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4527 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4528 memory slot, and num_pages is the size in bits of the input bitmap.
4529 first_page must be a multiple of 64; num_pages must also be a multiple of
4530 64 unless first_page + num_pages is the size of the memory slot. For each
4531 bit that is set in the input bitmap, the corresponding page is marked "clean"
4532 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4533 (for example via write-protection, or by clearing the dirty bit in
4534 a page table entry).
4536 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4537 the address space for which you want to clear the dirty status. See
4538 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4540 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4541 is enabled; for more information, see the description of the capability.
4542 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4543 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4545 4.118 KVM_GET_SUPPORTED_HV_CPUID
4546 --------------------------------
4548 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4550 :Type: system ioctl, vcpu ioctl
4551 :Parameters: struct kvm_cpuid2 (in/out)
4552 :Returns: 0 on success, -1 on error
4559 struct kvm_cpuid_entry2 entries[0];
4562 struct kvm_cpuid_entry2 {
4573 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4574 KVM. Userspace can use the information returned by this ioctl to construct
4575 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4576 Windows or Hyper-V guests).
4578 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4579 Functional Specification (TLFS). These leaves can't be obtained with
4580 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4581 leaves (0x40000000, 0x40000001).
4583 Currently, the following list of CPUID leaves are returned:
4585 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4586 - HYPERV_CPUID_INTERFACE
4587 - HYPERV_CPUID_VERSION
4588 - HYPERV_CPUID_FEATURES
4589 - HYPERV_CPUID_ENLIGHTMENT_INFO
4590 - HYPERV_CPUID_IMPLEMENT_LIMITS
4591 - HYPERV_CPUID_NESTED_FEATURES
4592 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4593 - HYPERV_CPUID_SYNDBG_INTERFACE
4594 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4596 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4597 with the 'nent' field indicating the number of entries in the variable-size
4598 array 'entries'. If the number of entries is too low to describe all Hyper-V
4599 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4600 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4601 number of valid entries in the 'entries' array, which is then filled.
4603 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4604 userspace should not expect to get any particular value there.
4606 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4607 system ioctl which exposes all supported feature bits unconditionally, vcpu
4608 version has the following quirks:
4610 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4611 feature bit are only exposed when Enlightened VMCS was previously enabled
4612 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4613 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4614 (presumes KVM_CREATE_IRQCHIP has already been called).
4616 4.119 KVM_ARM_VCPU_FINALIZE
4617 ---------------------------
4619 :Architectures: arm, arm64
4621 :Parameters: int feature (in)
4622 :Returns: 0 on success, -1 on error
4626 ====== ==============================================================
4627 EPERM feature not enabled, needs configuration, or already finalized
4628 EINVAL feature unknown or not present
4629 ====== ==============================================================
4631 Recognised values for feature:
4633 ===== ===========================================
4634 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4635 ===== ===========================================
4637 Finalizes the configuration of the specified vcpu feature.
4639 The vcpu must already have been initialised, enabling the affected feature, by
4640 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4643 For affected vcpu features, this is a mandatory step that must be performed
4644 before the vcpu is fully usable.
4646 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4647 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4648 that should be performaned and how to do it are feature-dependent.
4650 Other calls that depend on a particular feature being finalized, such as
4651 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4652 -EPERM unless the feature has already been finalized by means of a
4653 KVM_ARM_VCPU_FINALIZE call.
4655 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4658 4.120 KVM_SET_PMU_EVENT_FILTER
4659 ------------------------------
4661 :Capability: KVM_CAP_PMU_EVENT_FILTER
4664 :Parameters: struct kvm_pmu_event_filter (in)
4665 :Returns: 0 on success, -1 on error
4669 struct kvm_pmu_event_filter {
4672 __u32 fixed_counter_bitmap;
4678 This ioctl restricts the set of PMU events that the guest can program.
4679 The argument holds a list of events which will be allowed or denied.
4680 The eventsel+umask of each event the guest attempts to program is compared
4681 against the events field to determine whether the guest should have access.
4682 The events field only controls general purpose counters; fixed purpose
4683 counters are controlled by the fixed_counter_bitmap.
4685 No flags are defined yet, the field must be zero.
4687 Valid values for 'action'::
4689 #define KVM_PMU_EVENT_ALLOW 0
4690 #define KVM_PMU_EVENT_DENY 1
4692 4.121 KVM_PPC_SVM_OFF
4693 ---------------------
4696 :Architectures: powerpc
4699 :Returns: 0 on successful completion,
4703 ====== ================================================================
4704 EINVAL if ultravisor failed to terminate the secure guest
4705 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4706 ====== ================================================================
4708 This ioctl is used to turn off the secure mode of the guest or transition
4709 the guest from secure mode to normal mode. This is invoked when the guest
4710 is reset. This has no effect if called for a normal guest.
4712 This ioctl issues an ultravisor call to terminate the secure guest,
4713 unpins the VPA pages and releases all the device pages that are used to
4714 track the secure pages by hypervisor.
4716 4.122 KVM_S390_NORMAL_RESET
4717 ---------------------------
4719 :Capability: KVM_CAP_S390_VCPU_RESETS
4720 :Architectures: s390
4725 This ioctl resets VCPU registers and control structures according to
4726 the cpu reset definition in the POP (Principles Of Operation).
4728 4.123 KVM_S390_INITIAL_RESET
4729 ----------------------------
4732 :Architectures: s390
4737 This ioctl resets VCPU registers and control structures according to
4738 the initial cpu reset definition in the POP. However, the cpu is not
4739 put into ESA mode. This reset is a superset of the normal reset.
4741 4.124 KVM_S390_CLEAR_RESET
4742 --------------------------
4744 :Capability: KVM_CAP_S390_VCPU_RESETS
4745 :Architectures: s390
4750 This ioctl resets VCPU registers and control structures according to
4751 the clear cpu reset definition in the POP. However, the cpu is not put
4752 into ESA mode. This reset is a superset of the initial reset.
4755 4.125 KVM_S390_PV_COMMAND
4756 -------------------------
4758 :Capability: KVM_CAP_S390_PROTECTED
4759 :Architectures: s390
4761 :Parameters: struct kvm_pv_cmd
4762 :Returns: 0 on success, < 0 on error
4767 __u32 cmd; /* Command to be executed */
4768 __u16 rc; /* Ultravisor return code */
4769 __u16 rrc; /* Ultravisor return reason code */
4770 __u64 data; /* Data or address */
4771 __u32 flags; /* flags for future extensions. Must be 0 for now */
4778 Allocate memory and register the VM with the Ultravisor, thereby
4779 donating memory to the Ultravisor that will become inaccessible to
4780 KVM. All existing CPUs are converted to protected ones. After this
4781 command has succeeded, any CPU added via hotplug will become
4782 protected during its creation as well.
4786 ===== =============================
4787 EINTR an unmasked signal is pending
4788 ===== =============================
4792 Deregister the VM from the Ultravisor and reclaim the memory that
4793 had been donated to the Ultravisor, making it usable by the kernel
4794 again. All registered VCPUs are converted back to non-protected
4797 KVM_PV_VM_SET_SEC_PARMS
4798 Pass the image header from VM memory to the Ultravisor in
4799 preparation of image unpacking and verification.
4802 Unpack (protect and decrypt) a page of the encrypted boot image.
4805 Verify the integrity of the unpacked image. Only if this succeeds,
4806 KVM is allowed to start protected VCPUs.
4808 4.126 KVM_X86_SET_MSR_FILTER
4809 ----------------------------
4811 :Capability: KVM_CAP_X86_MSR_FILTER
4814 :Parameters: struct kvm_msr_filter
4815 :Returns: 0 on success, < 0 on error
4819 struct kvm_msr_filter_range {
4820 #define KVM_MSR_FILTER_READ (1 << 0)
4821 #define KVM_MSR_FILTER_WRITE (1 << 1)
4823 __u32 nmsrs; /* number of msrs in bitmap */
4824 __u32 base; /* MSR index the bitmap starts at */
4825 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4828 #define KVM_MSR_FILTER_MAX_RANGES 16
4829 struct kvm_msr_filter {
4830 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4831 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4833 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4836 flags values for ``struct kvm_msr_filter_range``:
4838 ``KVM_MSR_FILTER_READ``
4840 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4841 indicates that a read should immediately fail, while a 1 indicates that
4842 a read for a particular MSR should be handled regardless of the default
4845 ``KVM_MSR_FILTER_WRITE``
4847 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4848 indicates that a write should immediately fail, while a 1 indicates that
4849 a write for a particular MSR should be handled regardless of the default
4852 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4854 Filter both read and write accesses to MSRs using the given bitmap. A 0
4855 in the bitmap indicates that both reads and writes should immediately fail,
4856 while a 1 indicates that reads and writes for a particular MSR are not
4857 filtered by this range.
4859 flags values for ``struct kvm_msr_filter``:
4861 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4863 If no filter range matches an MSR index that is getting accessed, KVM will
4864 fall back to allowing access to the MSR.
4866 ``KVM_MSR_FILTER_DEFAULT_DENY``
4868 If no filter range matches an MSR index that is getting accessed, KVM will
4869 fall back to rejecting access to the MSR. In this mode, all MSRs that should
4870 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4872 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4873 specify whether a certain MSR access should be explicitly filtered for or not.
4875 If this ioctl has never been invoked, MSR accesses are not guarded and the
4876 default KVM in-kernel emulation behavior is fully preserved.
4878 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4879 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4882 As soon as the filtering is in place, every MSR access is processed through
4883 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4884 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4885 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4888 If a bit is within one of the defined ranges, read and write accesses are
4889 guarded by the bitmap's value for the MSR index if the kind of access
4890 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4891 cover this particular access, the behavior is determined by the flags
4892 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4893 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4895 Each bitmap range specifies a range of MSRs to potentially allow access on.
4896 The range goes from MSR index [base .. base+nmsrs]. The flags field
4897 indicates whether reads, writes or both reads and writes are filtered
4898 by setting a 1 bit in the bitmap for the corresponding MSR index.
4900 If an MSR access is not permitted through the filtering, it generates a
4901 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4902 allows user space to deflect and potentially handle various MSR accesses
4905 Note, invoking this ioctl with a vCPU is running is inherently racy. However,
4906 KVM does guarantee that vCPUs will see either the previous filter or the new
4907 filter, e.g. MSRs with identical settings in both the old and new filter will
4908 have deterministic behavior.
4910 4.127 KVM_XEN_HVM_SET_ATTR
4911 --------------------------
4913 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4916 :Parameters: struct kvm_xen_hvm_attr
4917 :Returns: 0 on success, < 0 on error
4921 struct kvm_xen_hvm_attr {
4936 KVM_XEN_ATTR_TYPE_LONG_MODE
4937 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
4938 determines the layout of the shared info pages exposed to the VM.
4940 KVM_XEN_ATTR_TYPE_SHARED_INFO
4941 Sets the guest physical frame number at which the Xen "shared info"
4942 page resides. Note that although Xen places vcpu_info for the first
4943 32 vCPUs in the shared_info page, KVM does not automatically do so
4944 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
4945 explicitly even when the vcpu_info for a given vCPU resides at the
4946 "default" location in the shared_info page. This is because KVM is
4947 not aware of the Xen CPU id which is used as the index into the
4948 vcpu_info[] array, so cannot know the correct default location.
4950 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
4951 Sets the exception vector used to deliver Xen event channel upcalls.
4953 4.127 KVM_XEN_HVM_GET_ATTR
4954 --------------------------
4956 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4959 :Parameters: struct kvm_xen_hvm_attr
4960 :Returns: 0 on success, < 0 on error
4962 Allows Xen VM attributes to be read. For the structure and types,
4963 see KVM_XEN_HVM_SET_ATTR above.
4965 4.128 KVM_XEN_VCPU_SET_ATTR
4966 ---------------------------
4968 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
4971 :Parameters: struct kvm_xen_vcpu_attr
4972 :Returns: 0 on success, < 0 on error
4976 struct kvm_xen_vcpu_attr {
4984 __u64 state_entry_time;
4986 __u64 time_runnable;
4995 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
4996 Sets the guest physical address of the vcpu_info for a given vCPU.
4998 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
4999 Sets the guest physical address of an additional pvclock structure
5000 for a given vCPU. This is typically used for guest vsyscall support.
5002 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5003 Sets the guest physical address of the vcpu_runstate_info for a given
5004 vCPU. This is how a Xen guest tracks CPU state such as steal time.
5006 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5007 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5008 the given vCPU from the .u.runstate.state member of the structure.
5009 KVM automatically accounts running and runnable time but blocked
5010 and offline states are only entered explicitly.
5012 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5013 Sets all fields of the vCPU runstate data from the .u.runstate member
5014 of the structure, including the current runstate. The state_entry_time
5015 must equal the sum of the other four times.
5017 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5018 This *adds* the contents of the .u.runstate members of the structure
5019 to the corresponding members of the given vCPU's runstate data, thus
5020 permitting atomic adjustments to the runstate times. The adjustment
5021 to the state_entry_time must equal the sum of the adjustments to the
5022 other four times. The state field must be set to -1, or to a valid
5023 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5024 or RUNSTATE_offline) to set the current accounted state as of the
5025 adjusted state_entry_time.
5027 4.129 KVM_XEN_VCPU_GET_ATTR
5028 ---------------------------
5030 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5033 :Parameters: struct kvm_xen_vcpu_attr
5034 :Returns: 0 on success, < 0 on error
5036 Allows Xen vCPU attributes to be read. For the structure and types,
5037 see KVM_XEN_VCPU_SET_ATTR above.
5039 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5040 with the KVM_XEN_VCPU_GET_ATTR ioctl.
5042 4.130 KVM_ARM_MTE_COPY_TAGS
5043 ---------------------------
5045 :Capability: KVM_CAP_ARM_MTE
5046 :Architectures: arm64
5048 :Parameters: struct kvm_arm_copy_mte_tags
5049 :Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5050 arguments, -EFAULT if memory cannot be accessed).
5054 struct kvm_arm_copy_mte_tags {
5062 Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5063 ``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr``
5064 field must point to a buffer which the tags will be copied to or from.
5066 ``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5067 ``KVM_ARM_TAGS_FROM_GUEST``.
5069 The size of the buffer to store the tags is ``(length / 16)`` bytes
5070 (granules in MTE are 16 bytes long). Each byte contains a single tag
5071 value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5072 ``PTRACE_POKEMTETAGS``.
5074 If an error occurs before any data is copied then a negative error code is
5075 returned. If some tags have been copied before an error occurs then the number
5076 of bytes successfully copied is returned. If the call completes successfully
5077 then ``length`` is returned.
5079 4.131 KVM_GET_SREGS2
5082 :Capability: KVM_CAP_SREGS2
5085 :Parameters: struct kvm_sregs2 (out)
5086 :Returns: 0 on success, -1 on error
5088 Reads special registers from the vcpu.
5089 This ioctl (when supported) replaces the KVM_GET_SREGS.
5094 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5095 struct kvm_segment cs, ds, es, fs, gs, ss;
5096 struct kvm_segment tr, ldt;
5097 struct kvm_dtable gdt, idt;
5098 __u64 cr0, cr2, cr3, cr4, cr8;
5105 flags values for ``kvm_sregs2``:
5107 ``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5109 Indicates thats the struct contain valid PDPTR values.
5112 4.132 KVM_SET_SREGS2
5115 :Capability: KVM_CAP_SREGS2
5118 :Parameters: struct kvm_sregs2 (in)
5119 :Returns: 0 on success, -1 on error
5121 Writes special registers into the vcpu.
5122 See KVM_GET_SREGS2 for the data structures.
5123 This ioctl (when supported) replaces the KVM_SET_SREGS.
5125 4.133 KVM_GET_STATS_FD
5126 ----------------------
5128 :Capability: KVM_CAP_STATS_BINARY_FD
5130 :Type: vm ioctl, vcpu ioctl
5132 :Returns: statistics file descriptor on success, < 0 on error
5136 ====== ======================================================
5137 ENOMEM if the fd could not be created due to lack of memory
5138 EMFILE if the number of opened files exceeds the limit
5139 ====== ======================================================
5141 The returned file descriptor can be used to read VM/vCPU statistics data in
5142 binary format. The data in the file descriptor consists of four blocks
5143 organized as follows:
5155 Apart from the header starting at offset 0, please be aware that it is
5156 not guaranteed that the four blocks are adjacent or in the above order;
5157 the offsets of the id, descriptors and data blocks are found in the
5158 header. However, all four blocks are aligned to 64 bit offsets in the
5159 file and they do not overlap.
5161 All blocks except the data block are immutable. Userspace can read them
5162 only one time after retrieving the file descriptor, and then use ``pread`` or
5163 ``lseek`` to read the statistics repeatedly.
5165 All data is in system endianness.
5167 The format of the header is as follows::
5169 struct kvm_stats_header {
5178 The ``flags`` field is not used at the moment. It is always read as 0.
5180 The ``name_size`` field is the size (in byte) of the statistics name string
5181 (including trailing '\0') which is contained in the "id string" block and
5182 appended at the end of every descriptor.
5184 The ``num_desc`` field is the number of descriptors that are included in the
5185 descriptor block. (The actual number of values in the data block may be
5186 larger, since each descriptor may comprise more than one value).
5188 The ``id_offset`` field is the offset of the id string from the start of the
5189 file indicated by the file descriptor. It is a multiple of 8.
5191 The ``desc_offset`` field is the offset of the Descriptors block from the start
5192 of the file indicated by the file descriptor. It is a multiple of 8.
5194 The ``data_offset`` field is the offset of the Stats Data block from the start
5195 of the file indicated by the file descriptor. It is a multiple of 8.
5197 The id string block contains a string which identifies the file descriptor on
5198 which KVM_GET_STATS_FD was invoked. The size of the block, including the
5199 trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5201 The descriptors block is only needed to be read once for the lifetime of the
5202 file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5203 by a string of size ``name_size``.
5205 #define KVM_STATS_TYPE_SHIFT 0
5206 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT)
5207 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT)
5208 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT)
5209 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT)
5211 #define KVM_STATS_UNIT_SHIFT 4
5212 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT)
5213 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT)
5214 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT)
5215 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT)
5216 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT)
5218 #define KVM_STATS_BASE_SHIFT 8
5219 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT)
5220 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT)
5221 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT)
5223 struct kvm_stats_desc {
5232 The ``flags`` field contains the type and unit of the statistics data described
5233 by this descriptor. Its endianness is CPU native.
5234 The following flags are supported:
5236 Bits 0-3 of ``flags`` encode the type:
5237 * ``KVM_STATS_TYPE_CUMULATIVE``
5238 The statistics data is cumulative. The value of data can only be increased.
5239 Most of the counters used in KVM are of this type.
5240 The corresponding ``size`` field for this type is always 1.
5241 All cumulative statistics data are read/write.
5242 * ``KVM_STATS_TYPE_INSTANT``
5243 The statistics data is instantaneous. Its value can be increased or
5244 decreased. This type is usually used as a measurement of some resources,
5245 like the number of dirty pages, the number of large pages, etc.
5246 All instant statistics are read only.
5247 The corresponding ``size`` field for this type is always 1.
5248 * ``KVM_STATS_TYPE_PEAK``
5249 The statistics data is peak. The value of data can only be increased, and
5250 represents a peak value for a measurement, for example the maximum number
5251 of items in a hash table bucket, the longest time waited and so on.
5252 The corresponding ``size`` field for this type is always 1.
5254 Bits 4-7 of ``flags`` encode the unit:
5255 * ``KVM_STATS_UNIT_NONE``
5256 There is no unit for the value of statistics data. This usually means that
5257 the value is a simple counter of an event.
5258 * ``KVM_STATS_UNIT_BYTES``
5259 It indicates that the statistics data is used to measure memory size, in the
5260 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
5261 determined by the ``exponent`` field in the descriptor.
5262 * ``KVM_STATS_UNIT_SECONDS``
5263 It indicates that the statistics data is used to measure time or latency.
5264 * ``KVM_STATS_UNIT_CYCLES``
5265 It indicates that the statistics data is used to measure CPU clock cycles.
5267 Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
5269 * ``KVM_STATS_BASE_POW10``
5270 The scale is based on power of 10. It is used for measurement of time and
5271 CPU clock cycles. For example, an exponent of -9 can be used with
5272 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
5273 * ``KVM_STATS_BASE_POW2``
5274 The scale is based on power of 2. It is used for measurement of memory size.
5275 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
5276 express that the unit is MiB.
5278 The ``size`` field is the number of values of this statistics data. Its
5279 value is usually 1 for most of simple statistics. 1 means it contains an
5280 unsigned 64bit data.
5282 The ``offset`` field is the offset from the start of Data Block to the start of
5283 the corresponding statistics data.
5285 The ``unused`` field is reserved for future support for other types of
5286 statistics data, like log/linear histogram. Its value is always 0 for the types
5289 The ``name`` field is the name string of the statistics data. The name string
5290 starts at the end of ``struct kvm_stats_desc``. The maximum length including
5291 the trailing ``'\0'``, is indicated by ``name_size`` in the header.
5293 The Stats Data block contains an array of 64-bit values in the same order
5294 as the descriptors in Descriptors block.
5296 5. The kvm_run structure
5297 ========================
5299 Application code obtains a pointer to the kvm_run structure by
5300 mmap()ing a vcpu fd. From that point, application code can control
5301 execution by changing fields in kvm_run prior to calling the KVM_RUN
5302 ioctl, and obtain information about the reason KVM_RUN returned by
5303 looking up structure members.
5309 __u8 request_interrupt_window;
5311 Request that KVM_RUN return when it becomes possible to inject external
5312 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
5316 __u8 immediate_exit;
5318 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
5319 exits immediately, returning -EINTR. In the common scenario where a
5320 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
5321 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
5322 Rather than blocking the signal outside KVM_RUN, userspace can set up
5323 a signal handler that sets run->immediate_exit to a non-zero value.
5325 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
5334 When KVM_RUN has returned successfully (return value 0), this informs
5335 application code why KVM_RUN has returned. Allowable values for this
5336 field are detailed below.
5340 __u8 ready_for_interrupt_injection;
5342 If request_interrupt_window has been specified, this field indicates
5343 an interrupt can be injected now with KVM_INTERRUPT.
5349 The value of the current interrupt flag. Only valid if in-kernel
5350 local APIC is not used.
5356 More architecture-specific flags detailing state of the VCPU that may
5357 affect the device's behavior. Current defined flags::
5359 /* x86, set if the VCPU is in system management mode */
5360 #define KVM_RUN_X86_SMM (1 << 0)
5361 /* x86, set if bus lock detected in VM */
5362 #define KVM_RUN_BUS_LOCK (1 << 1)
5366 /* in (pre_kvm_run), out (post_kvm_run) */
5369 The value of the cr8 register. Only valid if in-kernel local APIC is
5370 not used. Both input and output.
5376 The value of the APIC BASE msr. Only valid if in-kernel local
5377 APIC is not used. Both input and output.
5382 /* KVM_EXIT_UNKNOWN */
5384 __u64 hardware_exit_reason;
5387 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
5388 reasons. Further architecture-specific information is available in
5389 hardware_exit_reason.
5393 /* KVM_EXIT_FAIL_ENTRY */
5395 __u64 hardware_entry_failure_reason;
5396 __u32 cpu; /* if KVM_LAST_CPU */
5399 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
5400 to unknown reasons. Further architecture-specific information is
5401 available in hardware_entry_failure_reason.
5405 /* KVM_EXIT_EXCEPTION */
5417 #define KVM_EXIT_IO_IN 0
5418 #define KVM_EXIT_IO_OUT 1
5420 __u8 size; /* bytes */
5423 __u64 data_offset; /* relative to kvm_run start */
5426 If exit_reason is KVM_EXIT_IO, then the vcpu has
5427 executed a port I/O instruction which could not be satisfied by kvm.
5428 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
5429 where kvm expects application code to place the data for the next
5430 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
5434 /* KVM_EXIT_DEBUG */
5436 struct kvm_debug_exit_arch arch;
5439 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
5440 for which architecture specific information is returned.
5452 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
5453 executed a memory-mapped I/O instruction which could not be satisfied
5454 by kvm. The 'data' member contains the written data if 'is_write' is
5455 true, and should be filled by application code otherwise.
5457 The 'data' member contains, in its first 'len' bytes, the value as it would
5458 appear if the VCPU performed a load or store of the appropriate width directly
5463 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
5464 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5465 operations are complete (and guest state is consistent) only after userspace
5466 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5467 incomplete operations and then check for pending signals.
5469 The pending state of the operation is not preserved in state which is
5470 visible to userspace, thus userspace should ensure that the operation is
5471 completed before performing a live migration. Userspace can re-enter the
5472 guest with an unmasked signal pending or with the immediate_exit field set
5473 to complete pending operations without allowing any further instructions
5478 /* KVM_EXIT_HYPERCALL */
5487 Unused. This was once used for 'hypercall to userspace'. To implement
5488 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5490 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5494 /* KVM_EXIT_TPR_ACCESS */
5501 To be documented (KVM_TPR_ACCESS_REPORTING).
5505 /* KVM_EXIT_S390_SIEIC */
5508 __u64 mask; /* psw upper half */
5509 __u64 addr; /* psw lower half */
5518 /* KVM_EXIT_S390_RESET */
5519 #define KVM_S390_RESET_POR 1
5520 #define KVM_S390_RESET_CLEAR 2
5521 #define KVM_S390_RESET_SUBSYSTEM 4
5522 #define KVM_S390_RESET_CPU_INIT 8
5523 #define KVM_S390_RESET_IPL 16
5524 __u64 s390_reset_flags;
5530 /* KVM_EXIT_S390_UCONTROL */
5532 __u64 trans_exc_code;
5536 s390 specific. A page fault has occurred for a user controlled virtual
5537 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5538 resolved by the kernel.
5539 The program code and the translation exception code that were placed
5540 in the cpu's lowcore are presented here as defined by the z Architecture
5541 Principles of Operation Book in the Chapter for Dynamic Address Translation
5553 Deprecated - was used for 440 KVM.
5562 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5563 hypercalls and exit with this exit struct that contains all the guest gprs.
5565 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5566 Userspace can now handle the hypercall and when it's done modify the gprs as
5567 necessary. Upon guest entry all guest GPRs will then be replaced by the values
5572 /* KVM_EXIT_PAPR_HCALL */
5579 This is used on 64-bit PowerPC when emulating a pSeries partition,
5580 e.g. with the 'pseries' machine type in qemu. It occurs when the
5581 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
5582 contains the hypercall number (from the guest R3), and 'args' contains
5583 the arguments (from the guest R4 - R12). Userspace should put the
5584 return code in 'ret' and any extra returned values in args[].
5585 The possible hypercalls are defined in the Power Architecture Platform
5586 Requirements (PAPR) document available from www.power.org (free
5587 developer registration required to access it).
5591 /* KVM_EXIT_S390_TSCH */
5593 __u16 subchannel_id;
5594 __u16 subchannel_nr;
5601 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5602 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5603 interrupt for the target subchannel has been dequeued and subchannel_id,
5604 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5605 interrupt. ipb is needed for instruction parameter decoding.
5614 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5615 interrupt acknowledge path to the core. When the core successfully
5616 delivers an interrupt, it automatically populates the EPR register with
5617 the interrupt vector number and acknowledges the interrupt inside
5618 the interrupt controller.
5620 In case the interrupt controller lives in user space, we need to do
5621 the interrupt acknowledge cycle through it to fetch the next to be
5622 delivered interrupt vector using this exit.
5624 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5625 external interrupt has just been delivered into the guest. User space
5626 should put the acknowledged interrupt vector into the 'epr' field.
5630 /* KVM_EXIT_SYSTEM_EVENT */
5632 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5633 #define KVM_SYSTEM_EVENT_RESET 2
5634 #define KVM_SYSTEM_EVENT_CRASH 3
5639 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5640 a system-level event using some architecture specific mechanism (hypercall
5641 or some special instruction). In case of ARM/ARM64, this is triggered using
5642 HVC instruction based PSCI call from the vcpu. The 'type' field describes
5643 the system-level event type. The 'flags' field describes architecture
5644 specific flags for the system-level event.
5646 Valid values for 'type' are:
5648 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5649 VM. Userspace is not obliged to honour this, and if it does honour
5650 this does not need to destroy the VM synchronously (ie it may call
5651 KVM_RUN again before shutdown finally occurs).
5652 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5653 As with SHUTDOWN, userspace can choose to ignore the request, or
5654 to schedule the reset to occur in the future and may call KVM_RUN again.
5655 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5656 has requested a crash condition maintenance. Userspace can choose
5657 to ignore the request, or to gather VM memory core dump and/or
5658 reset/shutdown of the VM.
5662 /* KVM_EXIT_IOAPIC_EOI */
5667 Indicates that the VCPU's in-kernel local APIC received an EOI for a
5668 level-triggered IOAPIC interrupt. This exit only triggers when the
5669 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5670 the userspace IOAPIC should process the EOI and retrigger the interrupt if
5671 it is still asserted. Vector is the LAPIC interrupt vector for which the
5676 struct kvm_hyperv_exit {
5677 #define KVM_EXIT_HYPERV_SYNIC 1
5678 #define KVM_EXIT_HYPERV_HCALL 2
5679 #define KVM_EXIT_HYPERV_SYNDBG 3
5706 /* KVM_EXIT_HYPERV */
5707 struct kvm_hyperv_exit hyperv;
5709 Indicates that the VCPU exits into userspace to process some tasks
5710 related to Hyper-V emulation.
5712 Valid values for 'type' are:
5714 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5716 Hyper-V SynIC state change. Notification is used to remap SynIC
5717 event/message pages and to enable/disable SynIC messages/events processing
5720 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
5722 Hyper-V Synthetic debugger state change. Notification is used to either update
5723 the pending_page location or to send a control command (send the buffer located
5724 in send_page or recv a buffer to recv_page).
5728 /* KVM_EXIT_ARM_NISV */
5734 Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
5735 KVM will typically return to userspace and ask it to do MMIO emulation on its
5736 behalf. However, for certain classes of instructions, no instruction decode
5737 (direction, length of memory access) is provided, and fetching and decoding
5738 the instruction from the VM is overly complicated to live in the kernel.
5740 Historically, when this situation occurred, KVM would print a warning and kill
5741 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
5742 trying to do I/O, which just couldn't be emulated, and the warning message was
5743 phrased accordingly. However, what happened more often was that a guest bug
5744 caused access outside the guest memory areas which should lead to a more
5745 meaningful warning message and an external abort in the guest, if the access
5746 did not fall within an I/O window.
5748 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
5749 this capability at VM creation. Once this is done, these types of errors will
5750 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
5751 the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
5752 in the fault_ipa field. Userspace can either fix up the access if it's
5753 actually an I/O access by decoding the instruction from guest memory (if it's
5754 very brave) and continue executing the guest, or it can decide to suspend,
5755 dump, or restart the guest.
5757 Note that KVM does not skip the faulting instruction as it does for
5758 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
5759 if it decides to decode and emulate the instruction.
5763 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
5765 __u8 error; /* user -> kernel */
5767 __u32 reason; /* kernel -> user */
5768 __u32 index; /* kernel -> user */
5769 __u64 data; /* kernel <-> user */
5772 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
5773 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
5774 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
5777 The "reason" field specifies why the MSR trap occurred. User space will only
5778 receive MSR exit traps when a particular reason was requested during through
5779 ENABLE_CAP. Currently valid exit reasons are:
5781 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
5782 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
5783 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
5785 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
5786 wants to read. To respond to this request with a successful read, user space
5787 writes the respective data into the "data" field and must continue guest
5788 execution to ensure the read data is transferred into guest register state.
5790 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
5791 the "error" field. This will inject a #GP into the guest when the VCPU is
5794 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
5795 wants to write. Once finished processing the event, user space must continue
5796 vCPU execution. If the MSR write was unsuccessful, user space also sets the
5797 "error" field to "1".
5802 struct kvm_xen_exit {
5803 #define KVM_EXIT_XEN_HCALL 1
5816 struct kvm_hyperv_exit xen;
5818 Indicates that the VCPU exits into userspace to process some tasks
5819 related to Xen emulation.
5821 Valid values for 'type' are:
5823 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
5824 Userspace is expected to place the hypercall result into the appropriate
5825 field before invoking KVM_RUN again.
5829 /* Fix the size of the union. */
5834 * shared registers between kvm and userspace.
5835 * kvm_valid_regs specifies the register classes set by the host
5836 * kvm_dirty_regs specified the register classes dirtied by userspace
5837 * struct kvm_sync_regs is architecture specific, as well as the
5838 * bits for kvm_valid_regs and kvm_dirty_regs
5840 __u64 kvm_valid_regs;
5841 __u64 kvm_dirty_regs;
5843 struct kvm_sync_regs regs;
5844 char padding[SYNC_REGS_SIZE_BYTES];
5847 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
5848 certain guest registers without having to call SET/GET_*REGS. Thus we can
5849 avoid some system call overhead if userspace has to handle the exit.
5850 Userspace can query the validity of the structure by checking
5851 kvm_valid_regs for specific bits. These bits are architecture specific
5852 and usually define the validity of a groups of registers. (e.g. one bit
5853 for general purpose registers)
5855 Please note that the kernel is allowed to use the kvm_run structure as the
5856 primary storage for certain register types. Therefore, the kernel may use the
5857 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
5865 6. Capabilities that can be enabled on vCPUs
5866 ============================================
5868 There are certain capabilities that change the behavior of the virtual CPU or
5869 the virtual machine when enabled. To enable them, please see section 4.37.
5870 Below you can find a list of capabilities and what their effect on the vCPU or
5871 the virtual machine is when enabling them.
5873 The following information is provided along with the description:
5876 which instruction set architectures provide this ioctl.
5877 x86 includes both i386 and x86_64.
5880 whether this is a per-vcpu or per-vm capability.
5883 what parameters are accepted by the capability.
5886 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
5887 are not detailed, but errors with specific meanings are.
5896 :Returns: 0 on success; -1 on error
5898 This capability enables interception of OSI hypercalls that otherwise would
5899 be treated as normal system calls to be injected into the guest. OSI hypercalls
5900 were invented by Mac-on-Linux to have a standardized communication mechanism
5901 between the guest and the host.
5903 When this capability is enabled, KVM_EXIT_OSI can occur.
5906 6.2 KVM_CAP_PPC_PAPR
5907 --------------------
5912 :Returns: 0 on success; -1 on error
5914 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
5915 done using the hypercall instruction "sc 1".
5917 It also sets the guest privilege level to "supervisor" mode. Usually the guest
5918 runs in "hypervisor" privilege mode with a few missing features.
5920 In addition to the above, it changes the semantics of SDR1. In this mode, the
5921 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
5922 HTAB invisible to the guest.
5924 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
5932 :Parameters: args[0] is the address of a struct kvm_config_tlb
5933 :Returns: 0 on success; -1 on error
5937 struct kvm_config_tlb {
5944 Configures the virtual CPU's TLB array, establishing a shared memory area
5945 between userspace and KVM. The "params" and "array" fields are userspace
5946 addresses of mmu-type-specific data structures. The "array_len" field is an
5947 safety mechanism, and should be set to the size in bytes of the memory that
5948 userspace has reserved for the array. It must be at least the size dictated
5949 by "mmu_type" and "params".
5951 While KVM_RUN is active, the shared region is under control of KVM. Its
5952 contents are undefined, and any modification by userspace results in
5953 boundedly undefined behavior.
5955 On return from KVM_RUN, the shared region will reflect the current state of
5956 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
5957 to tell KVM which entries have been changed, prior to calling KVM_RUN again
5960 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
5962 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
5963 - The "array" field points to an array of type "struct
5964 kvm_book3e_206_tlb_entry".
5965 - The array consists of all entries in the first TLB, followed by all
5966 entries in the second TLB.
5967 - Within a TLB, entries are ordered first by increasing set number. Within a
5968 set, entries are ordered by way (increasing ESEL).
5969 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
5970 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
5971 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
5972 hardware ignores this value for TLB0.
5974 6.4 KVM_CAP_S390_CSS_SUPPORT
5975 ----------------------------
5977 :Architectures: s390
5980 :Returns: 0 on success; -1 on error
5982 This capability enables support for handling of channel I/O instructions.
5984 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
5985 handled in-kernel, while the other I/O instructions are passed to userspace.
5987 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
5988 SUBCHANNEL intercepts.
5990 Note that even though this capability is enabled per-vcpu, the complete
5991 virtual machine is affected.
5998 :Parameters: args[0] defines whether the proxy facility is active
5999 :Returns: 0 on success; -1 on error
6001 This capability enables or disables the delivery of interrupts through the
6002 external proxy facility.
6004 When enabled (args[0] != 0), every time the guest gets an external interrupt
6005 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
6006 to receive the topmost interrupt vector.
6008 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
6010 When this capability is enabled, KVM_EXIT_EPR can occur.
6012 6.6 KVM_CAP_IRQ_MPIC
6013 --------------------
6016 :Parameters: args[0] is the MPIC device fd;
6017 args[1] is the MPIC CPU number for this vcpu
6019 This capability connects the vcpu to an in-kernel MPIC device.
6021 6.7 KVM_CAP_IRQ_XICS
6022 --------------------
6026 :Parameters: args[0] is the XICS device fd;
6027 args[1] is the XICS CPU number (server ID) for this vcpu
6029 This capability connects the vcpu to an in-kernel XICS device.
6031 6.8 KVM_CAP_S390_IRQCHIP
6032 ------------------------
6034 :Architectures: s390
6038 This capability enables the in-kernel irqchip for s390. Please refer to
6039 "4.24 KVM_CREATE_IRQCHIP" for details.
6041 6.9 KVM_CAP_MIPS_FPU
6042 --------------------
6044 :Architectures: mips
6046 :Parameters: args[0] is reserved for future use (should be 0).
6048 This capability allows the use of the host Floating Point Unit by the guest. It
6049 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
6050 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
6051 accessed (depending on the current guest FPU register mode), and the Status.FR,
6052 Config5.FRE bits are accessible via the KVM API and also from the guest,
6053 depending on them being supported by the FPU.
6055 6.10 KVM_CAP_MIPS_MSA
6056 ---------------------
6058 :Architectures: mips
6060 :Parameters: args[0] is reserved for future use (should be 0).
6062 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
6063 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
6064 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
6065 registers can be accessed, and the Config5.MSAEn bit is accessible via the
6066 KVM API and also from the guest.
6068 6.74 KVM_CAP_SYNC_REGS
6069 ----------------------
6071 :Architectures: s390, x86
6072 :Target: s390: always enabled, x86: vcpu
6074 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
6076 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
6078 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
6079 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
6080 without having to call SET/GET_*REGS". This reduces overhead by eliminating
6081 repeated ioctl calls for setting and/or getting register values. This is
6082 particularly important when userspace is making synchronous guest state
6083 modifications, e.g. when emulating and/or intercepting instructions in
6086 For s390 specifics, please refer to the source code.
6090 - the register sets to be copied out to kvm_run are selectable
6091 by userspace (rather that all sets being copied out for every exit).
6092 - vcpu_events are available in addition to regs and sregs.
6094 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
6095 function as an input bit-array field set by userspace to indicate the
6096 specific register sets to be copied out on the next exit.
6098 To indicate when userspace has modified values that should be copied into
6099 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
6100 This is done using the same bitflags as for the 'kvm_valid_regs' field.
6101 If the dirty bit is not set, then the register set values will not be copied
6102 into the vCPU even if they've been modified.
6104 Unused bitfields in the bitarrays must be set to zero.
6108 struct kvm_sync_regs {
6109 struct kvm_regs regs;
6110 struct kvm_sregs sregs;
6111 struct kvm_vcpu_events events;
6114 6.75 KVM_CAP_PPC_IRQ_XIVE
6115 -------------------------
6119 :Parameters: args[0] is the XIVE device fd;
6120 args[1] is the XIVE CPU number (server ID) for this vcpu
6122 This capability connects the vcpu to an in-kernel XIVE device.
6124 7. Capabilities that can be enabled on VMs
6125 ==========================================
6127 There are certain capabilities that change the behavior of the virtual
6128 machine when enabled. To enable them, please see section 4.37. Below
6129 you can find a list of capabilities and what their effect on the VM
6130 is when enabling them.
6132 The following information is provided along with the description:
6135 which instruction set architectures provide this ioctl.
6136 x86 includes both i386 and x86_64.
6139 what parameters are accepted by the capability.
6142 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6143 are not detailed, but errors with specific meanings are.
6146 7.1 KVM_CAP_PPC_ENABLE_HCALL
6147 ----------------------------
6150 :Parameters: args[0] is the sPAPR hcall number;
6151 args[1] is 0 to disable, 1 to enable in-kernel handling
6153 This capability controls whether individual sPAPR hypercalls (hcalls)
6154 get handled by the kernel or not. Enabling or disabling in-kernel
6155 handling of an hcall is effective across the VM. On creation, an
6156 initial set of hcalls are enabled for in-kernel handling, which
6157 consists of those hcalls for which in-kernel handlers were implemented
6158 before this capability was implemented. If disabled, the kernel will
6159 not to attempt to handle the hcall, but will always exit to userspace
6160 to handle it. Note that it may not make sense to enable some and
6161 disable others of a group of related hcalls, but KVM does not prevent
6162 userspace from doing that.
6164 If the hcall number specified is not one that has an in-kernel
6165 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
6168 7.2 KVM_CAP_S390_USER_SIGP
6169 --------------------------
6171 :Architectures: s390
6174 This capability controls which SIGP orders will be handled completely in user
6175 space. With this capability enabled, all fast orders will be handled completely
6182 - CONDITIONAL EMERGENCY SIGNAL
6184 All other orders will be handled completely in user space.
6186 Only privileged operation exceptions will be checked for in the kernel (or even
6187 in the hardware prior to interception). If this capability is not enabled, the
6188 old way of handling SIGP orders is used (partially in kernel and user space).
6190 7.3 KVM_CAP_S390_VECTOR_REGISTERS
6191 ---------------------------------
6193 :Architectures: s390
6195 :Returns: 0 on success, negative value on error
6197 Allows use of the vector registers introduced with z13 processor, and
6198 provides for the synchronization between host and user space. Will
6199 return -EINVAL if the machine does not support vectors.
6201 7.4 KVM_CAP_S390_USER_STSI
6202 --------------------------
6204 :Architectures: s390
6207 This capability allows post-handlers for the STSI instruction. After
6208 initial handling in the kernel, KVM exits to user space with
6209 KVM_EXIT_S390_STSI to allow user space to insert further data.
6211 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
6223 @addr - guest address of STSI SYSIB
6227 @ar - access register number
6229 KVM handlers should exit to userspace with rc = -EREMOTE.
6231 7.5 KVM_CAP_SPLIT_IRQCHIP
6232 -------------------------
6235 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
6236 :Returns: 0 on success, -1 on error
6238 Create a local apic for each processor in the kernel. This can be used
6239 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
6240 IOAPIC and PIC (and also the PIT, even though this has to be enabled
6243 This capability also enables in kernel routing of interrupt requests;
6244 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
6245 used in the IRQ routing table. The first args[0] MSI routes are reserved
6246 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
6247 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
6249 Fails if VCPU has already been created, or if the irqchip is already in the
6250 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
6255 :Architectures: s390
6258 Allows use of runtime-instrumentation introduced with zEC12 processor.
6259 Will return -EINVAL if the machine does not support runtime-instrumentation.
6260 Will return -EBUSY if a VCPU has already been created.
6262 7.7 KVM_CAP_X2APIC_API
6263 ----------------------
6266 :Parameters: args[0] - features that should be enabled
6267 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
6269 Valid feature flags in args[0] are::
6271 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
6272 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
6274 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
6275 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
6276 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
6277 respective sections.
6279 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
6280 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
6281 as a broadcast even in x2APIC mode in order to support physical x2APIC
6282 without interrupt remapping. This is undesirable in logical mode,
6283 where 0xff represents CPUs 0-7 in cluster 0.
6285 7.8 KVM_CAP_S390_USER_INSTR0
6286 ----------------------------
6288 :Architectures: s390
6291 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
6292 be intercepted and forwarded to user space. User space can use this
6293 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
6294 not inject an operating exception for these instructions, user space has
6295 to take care of that.
6297 This capability can be enabled dynamically even if VCPUs were already
6298 created and are running.
6303 :Architectures: s390
6305 :Returns: 0 on success; -EINVAL if the machine does not support
6306 guarded storage; -EBUSY if a VCPU has already been created.
6308 Allows use of guarded storage for the KVM guest.
6310 7.10 KVM_CAP_S390_AIS
6311 ---------------------
6313 :Architectures: s390
6316 Allow use of adapter-interruption suppression.
6317 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
6319 7.11 KVM_CAP_PPC_SMT
6320 --------------------
6323 :Parameters: vsmt_mode, flags
6325 Enabling this capability on a VM provides userspace with a way to set
6326 the desired virtual SMT mode (i.e. the number of virtual CPUs per
6327 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
6328 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
6329 the number of threads per subcore for the host. Currently flags must
6330 be 0. A successful call to enable this capability will result in
6331 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
6332 subsequently queried for the VM. This capability is only supported by
6333 HV KVM, and can only be set before any VCPUs have been created.
6334 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
6335 modes are available.
6337 7.12 KVM_CAP_PPC_FWNMI
6338 ----------------------
6343 With this capability a machine check exception in the guest address
6344 space will cause KVM to exit the guest with NMI exit reason. This
6345 enables QEMU to build error log and branch to guest kernel registered
6346 machine check handling routine. Without this capability KVM will
6347 branch to guests' 0x200 interrupt vector.
6349 7.13 KVM_CAP_X86_DISABLE_EXITS
6350 ------------------------------
6353 :Parameters: args[0] defines which exits are disabled
6354 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
6356 Valid bits in args[0] are::
6358 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
6359 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
6360 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
6361 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
6363 Enabling this capability on a VM provides userspace with a way to no
6364 longer intercept some instructions for improved latency in some
6365 workloads, and is suggested when vCPUs are associated to dedicated
6366 physical CPUs. More bits can be added in the future; userspace can
6367 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
6370 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
6372 7.14 KVM_CAP_S390_HPAGE_1M
6373 --------------------------
6375 :Architectures: s390
6377 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
6378 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
6381 With this capability the KVM support for memory backing with 1m pages
6382 through hugetlbfs can be enabled for a VM. After the capability is
6383 enabled, cmma can't be enabled anymore and pfmfi and the storage key
6384 interpretation are disabled. If cmma has already been enabled or the
6385 hpage module parameter is not set to 1, -EINVAL is returned.
6387 While it is generally possible to create a huge page backed VM without
6388 this capability, the VM will not be able to run.
6390 7.15 KVM_CAP_MSR_PLATFORM_INFO
6391 ------------------------------
6394 :Parameters: args[0] whether feature should be enabled or not
6396 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
6397 a #GP would be raised when the guest tries to access. Currently, this
6398 capability does not enable write permissions of this MSR for the guest.
6400 7.16 KVM_CAP_PPC_NESTED_HV
6401 --------------------------
6405 :Returns: 0 on success, -EINVAL when the implementation doesn't support
6406 nested-HV virtualization.
6408 HV-KVM on POWER9 and later systems allows for "nested-HV"
6409 virtualization, which provides a way for a guest VM to run guests that
6410 can run using the CPU's supervisor mode (privileged non-hypervisor
6411 state). Enabling this capability on a VM depends on the CPU having
6412 the necessary functionality and on the facility being enabled with a
6413 kvm-hv module parameter.
6415 7.17 KVM_CAP_EXCEPTION_PAYLOAD
6416 ------------------------------
6419 :Parameters: args[0] whether feature should be enabled or not
6421 With this capability enabled, CR2 will not be modified prior to the
6422 emulated VM-exit when L1 intercepts a #PF exception that occurs in
6423 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
6424 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
6425 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
6426 #DB) exception for L2, exception.has_payload will be set and the
6427 faulting address (or the new DR6 bits*) will be reported in the
6428 exception_payload field. Similarly, when userspace injects a #PF (or
6429 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
6430 exception.has_payload and to put the faulting address - or the new DR6
6431 bits\ [#]_ - in the exception_payload field.
6433 This capability also enables exception.pending in struct
6434 kvm_vcpu_events, which allows userspace to distinguish between pending
6435 and injected exceptions.
6438 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
6441 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
6443 :Architectures: x86, arm, arm64, mips
6444 :Parameters: args[0] whether feature should be enabled or not
6448 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
6449 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
6451 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
6452 automatically clear and write-protect all pages that are returned as dirty.
6453 Rather, userspace will have to do this operation separately using
6454 KVM_CLEAR_DIRTY_LOG.
6456 At the cost of a slightly more complicated operation, this provides better
6457 scalability and responsiveness for two reasons. First,
6458 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
6459 than requiring to sync a full memslot; this ensures that KVM does not
6460 take spinlocks for an extended period of time. Second, in some cases a
6461 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
6462 userspace actually using the data in the page. Pages can be modified
6463 during this time, which is inefficient for both the guest and userspace:
6464 the guest will incur a higher penalty due to write protection faults,
6465 while userspace can see false reports of dirty pages. Manual reprotection
6466 helps reducing this time, improving guest performance and reducing the
6467 number of dirty log false positives.
6469 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
6470 will be initialized to 1 when created. This also improves performance because
6471 dirty logging can be enabled gradually in small chunks on the first call
6472 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
6473 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
6474 x86 and arm64 for now).
6476 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
6477 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
6478 it hard or impossible to use it correctly. The availability of
6479 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
6480 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
6482 7.19 KVM_CAP_PPC_SECURE_GUEST
6483 ------------------------------
6487 This capability indicates that KVM is running on a host that has
6488 ultravisor firmware and thus can support a secure guest. On such a
6489 system, a guest can ask the ultravisor to make it a secure guest,
6490 one whose memory is inaccessible to the host except for pages which
6491 are explicitly requested to be shared with the host. The ultravisor
6492 notifies KVM when a guest requests to become a secure guest, and KVM
6493 has the opportunity to veto the transition.
6495 If present, this capability can be enabled for a VM, meaning that KVM
6496 will allow the transition to secure guest mode. Otherwise KVM will
6497 veto the transition.
6499 7.20 KVM_CAP_HALT_POLL
6500 ----------------------
6504 :Parameters: args[0] is the maximum poll time in nanoseconds
6505 :Returns: 0 on success; -1 on error
6507 This capability overrides the kvm module parameter halt_poll_ns for the
6510 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6511 scheduling during guest halts. The maximum time a VCPU can spend polling is
6512 controlled by the kvm module parameter halt_poll_ns. This capability allows
6513 the maximum halt time to specified on a per-VM basis, effectively overriding
6514 the module parameter for the target VM.
6516 7.21 KVM_CAP_X86_USER_SPACE_MSR
6517 -------------------------------
6521 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6522 :Returns: 0 on success; -1 on error
6524 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6527 When a guest requests to read or write an MSR, KVM may not implement all MSRs
6528 that are relevant to a respective system. It also does not differentiate by
6531 To allow more fine grained control over MSR handling, user space may enable
6532 this capability. With it enabled, MSR accesses that match the mask specified in
6533 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6534 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6535 can then handle to implement model specific MSR handling and/or user notifications
6536 to inform a user that an MSR was not handled.
6538 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
6539 -------------------------------
6543 :Parameters: args[0] defines the policy used when bus locks detected in guest
6544 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
6546 Valid bits in args[0] are::
6548 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
6549 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
6551 Enabling this capability on a VM provides userspace with a way to select
6552 a policy to handle the bus locks detected in guest. Userspace can obtain
6553 the supported modes from the result of KVM_CHECK_EXTENSION and define it
6554 through the KVM_ENABLE_CAP.
6556 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
6557 currently and mutually exclusive with each other. More bits can be added in
6560 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
6561 so that no additional actions are needed. This is the default mode.
6563 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
6564 in VM. KVM just exits to userspace when handling them. Userspace can enforce
6565 its own throttling or other policy based mitigations.
6567 This capability is aimed to address the thread that VM can exploit bus locks to
6568 degree the performance of the whole system. Once the userspace enable this
6569 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
6570 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
6571 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
6572 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
6573 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
6575 7.23 KVM_CAP_PPC_DAWR1
6576 ----------------------
6580 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
6582 This capability can be used to check / enable 2nd DAWR feature provided
6583 by POWER10 processor.
6586 7.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
6587 -------------------------------------
6589 Architectures: x86 SEV enabled
6591 Parameters: args[0] is the fd of the source vm
6592 Returns: 0 on success; ENOTTY on error
6594 This capability enables userspace to copy encryption context from the vm
6595 indicated by the fd to the vm this is called on.
6597 This is intended to support in-guest workloads scheduled by the host. This
6598 allows the in-guest workload to maintain its own NPTs and keeps the two vms
6599 from accidentally clobbering each other with interrupts and the like (separate
6602 7.25 KVM_CAP_SGX_ATTRIBUTE
6603 --------------------------
6607 :Parameters: args[0] is a file handle of a SGX attribute file in securityfs
6608 :Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
6609 attribute is not supported by KVM.
6611 KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
6612 more priveleged enclave attributes. args[0] must hold a file handle to a valid
6613 SGX attribute file corresponding to an attribute that is supported/restricted
6614 by KVM (currently only PROVISIONKEY).
6616 The SGX subsystem restricts access to a subset of enclave attributes to provide
6617 additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
6618 is restricted to deter malware from using the PROVISIONKEY to obtain a stable
6619 system fingerprint. To prevent userspace from circumventing such restrictions
6620 by running an enclave in a VM, KVM prevents access to privileged attributes by
6623 See Documentation/x86/sgx.rst for more details.
6625 7.26 KVM_CAP_PPC_RPT_INVALIDATE
6626 -------------------------------
6628 :Capability: KVM_CAP_PPC_RPT_INVALIDATE
6632 This capability indicates that the kernel is capable of handling
6633 H_RPT_INVALIDATE hcall.
6635 In order to enable the use of H_RPT_INVALIDATE in the guest,
6636 user space might have to advertise it for the guest. For example,
6637 IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
6638 present in the "ibm,hypertas-functions" device-tree property.
6640 This capability is enabled for hypervisors on platforms like POWER9
6641 that support radix MMU.
6643 7.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
6644 --------------------------------------
6647 :Parameters: args[0] whether the feature should be enabled or not
6649 When this capability is enabled, an emulation failure will result in an exit
6650 to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
6651 to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
6652 to 15 instruction bytes for any exit to userspace resulting from an emulation
6653 failure. When these exits to userspace occur use the emulation_failure struct
6654 instead of the internal struct. They both have the same layout, but the
6655 emulation_failure struct matches the content better. It also explicitly
6656 defines the 'flags' field which is used to describe the fields in the struct
6657 that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
6658 set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
6661 7.28 KVM_CAP_ARM_MTE
6662 --------------------
6664 :Architectures: arm64
6667 This capability indicates that KVM (and the hardware) supports exposing the
6668 Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
6669 VMM before creating any VCPUs to allow the guest access. Note that MTE is only
6670 available to a guest running in AArch64 mode and enabling this capability will
6671 cause attempts to create AArch32 VCPUs to fail.
6673 When enabled the guest is able to access tags associated with any memory given
6674 to the guest. KVM will ensure that the tags are maintained during swap or
6675 hibernation of the host; however the VMM needs to manually save/restore the
6676 tags as appropriate if the VM is migrated.
6678 When this capability is enabled all memory in memslots must be mapped as
6679 not-shareable (no MAP_SHARED), attempts to create a memslot with a
6680 MAP_SHARED mmap will result in an -EINVAL return.
6682 When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
6683 perform a bulk copy of tags to/from the guest.
6685 8. Other capabilities.
6686 ======================
6688 This section lists capabilities that give information about other
6689 features of the KVM implementation.
6691 8.1 KVM_CAP_PPC_HWRNG
6692 ---------------------
6696 This capability, if KVM_CHECK_EXTENSION indicates that it is
6697 available, means that the kernel has an implementation of the
6698 H_RANDOM hypercall backed by a hardware random-number generator.
6699 If present, the kernel H_RANDOM handler can be enabled for guest use
6700 with the KVM_CAP_PPC_ENABLE_HCALL capability.
6702 8.2 KVM_CAP_HYPERV_SYNIC
6703 ------------------------
6707 This capability, if KVM_CHECK_EXTENSION indicates that it is
6708 available, means that the kernel has an implementation of the
6709 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
6710 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
6712 In order to use SynIC, it has to be activated by setting this
6713 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
6714 will disable the use of APIC hardware virtualization even if supported
6715 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
6717 8.3 KVM_CAP_PPC_RADIX_MMU
6718 -------------------------
6722 This capability, if KVM_CHECK_EXTENSION indicates that it is
6723 available, means that the kernel can support guests using the
6724 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
6727 8.4 KVM_CAP_PPC_HASH_MMU_V3
6728 ---------------------------
6732 This capability, if KVM_CHECK_EXTENSION indicates that it is
6733 available, means that the kernel can support guests using the
6734 hashed page table MMU defined in Power ISA V3.00 (as implemented in
6735 the POWER9 processor), including in-memory segment tables.
6740 :Architectures: mips
6742 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6743 it is available, means that full hardware assisted virtualization capabilities
6744 of the hardware are available for use through KVM. An appropriate
6745 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
6748 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6749 available, it means that the VM is using full hardware assisted virtualization
6750 capabilities of the hardware. This is useful to check after creating a VM with
6751 KVM_VM_MIPS_DEFAULT.
6753 The value returned by KVM_CHECK_EXTENSION should be compared against known
6754 values (see below). All other values are reserved. This is to allow for the
6755 possibility of other hardware assisted virtualization implementations which
6756 may be incompatible with the MIPS VZ ASE.
6758 == ==========================================================================
6759 0 The trap & emulate implementation is in use to run guest code in user
6760 mode. Guest virtual memory segments are rearranged to fit the guest in the
6761 user mode address space.
6763 1 The MIPS VZ ASE is in use, providing full hardware assisted
6764 virtualization, including standard guest virtual memory segments.
6765 == ==========================================================================
6770 :Architectures: mips
6772 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6773 it is available, means that the trap & emulate implementation is available to
6774 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
6775 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
6776 to KVM_CREATE_VM to create a VM which utilises it.
6778 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6779 available, it means that the VM is using trap & emulate.
6781 8.7 KVM_CAP_MIPS_64BIT
6782 ----------------------
6784 :Architectures: mips
6786 This capability indicates the supported architecture type of the guest, i.e. the
6787 supported register and address width.
6789 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
6790 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
6791 be checked specifically against known values (see below). All other values are
6794 == ========================================================================
6795 0 MIPS32 or microMIPS32.
6796 Both registers and addresses are 32-bits wide.
6797 It will only be possible to run 32-bit guest code.
6799 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
6800 Registers are 64-bits wide, but addresses are 32-bits wide.
6801 64-bit guest code may run but cannot access MIPS64 memory segments.
6802 It will also be possible to run 32-bit guest code.
6804 2 MIPS64 or microMIPS64 with access to all address segments.
6805 Both registers and addresses are 64-bits wide.
6806 It will be possible to run 64-bit or 32-bit guest code.
6807 == ========================================================================
6809 8.9 KVM_CAP_ARM_USER_IRQ
6810 ------------------------
6812 :Architectures: arm, arm64
6814 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
6815 that if userspace creates a VM without an in-kernel interrupt controller, it
6816 will be notified of changes to the output level of in-kernel emulated devices,
6817 which can generate virtual interrupts, presented to the VM.
6818 For such VMs, on every return to userspace, the kernel
6819 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
6820 output level of the device.
6822 Whenever kvm detects a change in the device output level, kvm guarantees at
6823 least one return to userspace before running the VM. This exit could either
6824 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
6825 userspace can always sample the device output level and re-compute the state of
6826 the userspace interrupt controller. Userspace should always check the state
6827 of run->s.regs.device_irq_level on every kvm exit.
6828 The value in run->s.regs.device_irq_level can represent both level and edge
6829 triggered interrupt signals, depending on the device. Edge triggered interrupt
6830 signals will exit to userspace with the bit in run->s.regs.device_irq_level
6831 set exactly once per edge signal.
6833 The field run->s.regs.device_irq_level is available independent of
6834 run->kvm_valid_regs or run->kvm_dirty_regs bits.
6836 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
6837 number larger than 0 indicating the version of this capability is implemented
6838 and thereby which bits in run->s.regs.device_irq_level can signal values.
6840 Currently the following bits are defined for the device_irq_level bitmap::
6842 KVM_CAP_ARM_USER_IRQ >= 1:
6844 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
6845 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
6846 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
6848 Future versions of kvm may implement additional events. These will get
6849 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
6852 8.10 KVM_CAP_PPC_SMT_POSSIBLE
6853 -----------------------------
6857 Querying this capability returns a bitmap indicating the possible
6858 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
6859 (counting from the right) is set, then a virtual SMT mode of 2^N is
6862 8.11 KVM_CAP_HYPERV_SYNIC2
6863 --------------------------
6867 This capability enables a newer version of Hyper-V Synthetic interrupt
6868 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
6869 doesn't clear SynIC message and event flags pages when they are enabled by
6870 writing to the respective MSRs.
6872 8.12 KVM_CAP_HYPERV_VP_INDEX
6873 ----------------------------
6877 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
6878 value is used to denote the target vcpu for a SynIC interrupt. For
6879 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
6880 capability is absent, userspace can still query this msr's value.
6882 8.13 KVM_CAP_S390_AIS_MIGRATION
6883 -------------------------------
6885 :Architectures: s390
6888 This capability indicates if the flic device will be able to get/set the
6889 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
6890 to discover this without having to create a flic device.
6892 8.14 KVM_CAP_S390_PSW
6893 ---------------------
6895 :Architectures: s390
6897 This capability indicates that the PSW is exposed via the kvm_run structure.
6899 8.15 KVM_CAP_S390_GMAP
6900 ----------------------
6902 :Architectures: s390
6904 This capability indicates that the user space memory used as guest mapping can
6905 be anywhere in the user memory address space, as long as the memory slots are
6906 aligned and sized to a segment (1MB) boundary.
6908 8.16 KVM_CAP_S390_COW
6909 ---------------------
6911 :Architectures: s390
6913 This capability indicates that the user space memory used as guest mapping can
6914 use copy-on-write semantics as well as dirty pages tracking via read-only page
6917 8.17 KVM_CAP_S390_BPB
6918 ---------------------
6920 :Architectures: s390
6922 This capability indicates that kvm will implement the interfaces to handle
6923 reset, migration and nested KVM for branch prediction blocking. The stfle
6924 facility 82 should not be provided to the guest without this capability.
6926 8.18 KVM_CAP_HYPERV_TLBFLUSH
6927 ----------------------------
6931 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
6933 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
6934 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
6936 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
6937 ----------------------------------
6939 :Architectures: arm, arm64
6941 This capability indicates that userspace can specify (via the
6942 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
6943 takes a virtual SError interrupt exception.
6944 If KVM advertises this capability, userspace can only specify the ISS field for
6945 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
6946 CPU when the exception is taken. If this virtual SError is taken to EL1 using
6947 AArch64, this value will be reported in the ISS field of ESR_ELx.
6949 See KVM_CAP_VCPU_EVENTS for more details.
6951 8.20 KVM_CAP_HYPERV_SEND_IPI
6952 ----------------------------
6956 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
6958 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
6960 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
6961 -----------------------------------
6965 This capability indicates that KVM running on top of Hyper-V hypervisor
6966 enables Direct TLB flush for its guests meaning that TLB flush
6967 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
6968 Due to the different ABI for hypercall parameters between Hyper-V and
6969 KVM, enabling this capability effectively disables all hypercall
6970 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
6971 flush hypercalls by Hyper-V) so userspace should disable KVM identification
6972 in CPUID and only exposes Hyper-V identification. In this case, guest
6973 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
6975 8.22 KVM_CAP_S390_VCPU_RESETS
6976 -----------------------------
6978 :Architectures: s390
6980 This capability indicates that the KVM_S390_NORMAL_RESET and
6981 KVM_S390_CLEAR_RESET ioctls are available.
6983 8.23 KVM_CAP_S390_PROTECTED
6984 ---------------------------
6986 :Architectures: s390
6988 This capability indicates that the Ultravisor has been initialized and
6989 KVM can therefore start protected VMs.
6990 This capability governs the KVM_S390_PV_COMMAND ioctl and the
6991 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
6992 guests when the state change is invalid.
6994 8.24 KVM_CAP_STEAL_TIME
6995 -----------------------
6997 :Architectures: arm64, x86
6999 This capability indicates that KVM supports steal time accounting.
7000 When steal time accounting is supported it may be enabled with
7001 architecture-specific interfaces. This capability and the architecture-
7002 specific interfaces must be consistent, i.e. if one says the feature
7003 is supported, than the other should as well and vice versa. For arm64
7004 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
7005 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
7007 8.25 KVM_CAP_S390_DIAG318
7008 -------------------------
7010 :Architectures: s390
7012 This capability enables a guest to set information about its control program
7013 (i.e. guest kernel type and version). The information is helpful during
7014 system/firmware service events, providing additional data about the guest
7015 environments running on the machine.
7017 The information is associated with the DIAGNOSE 0x318 instruction, which sets
7018 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
7019 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
7020 environment the control program is running in (e.g. Linux, z/VM...), and the
7021 CPVC is used for information specific to OS (e.g. Linux version, Linux
7024 If this capability is available, then the CPNC and CPVC can be synchronized
7025 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
7027 8.26 KVM_CAP_X86_USER_SPACE_MSR
7028 -------------------------------
7032 This capability indicates that KVM supports deflection of MSR reads and
7033 writes to user space. It can be enabled on a VM level. If enabled, MSR
7034 accesses that would usually trigger a #GP by KVM into the guest will
7035 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
7036 KVM_EXIT_X86_WRMSR exit notifications.
7038 8.27 KVM_CAP_X86_MSR_FILTER
7039 ---------------------------
7043 This capability indicates that KVM supports that accesses to user defined MSRs
7044 may be rejected. With this capability exposed, KVM exports new VM ioctl
7045 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
7046 ranges that KVM should reject access to.
7048 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
7049 trap and emulate MSRs that are outside of the scope of KVM as well as
7050 limit the attack surface on KVM's MSR emulation code.
7052 8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
7053 -----------------------------
7057 When enabled, KVM will disable paravirtual features provided to the
7058 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
7059 (0x40000001). Otherwise, a guest may use the paravirtual features
7060 regardless of what has actually been exposed through the CPUID leaf.
7062 8.29 KVM_CAP_DIRTY_LOG_RING
7063 ---------------------------
7066 :Parameters: args[0] - size of the dirty log ring
7068 KVM is capable of tracking dirty memory using ring buffers that are
7069 mmaped into userspace; there is one dirty ring per vcpu.
7071 The dirty ring is available to userspace as an array of
7072 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
7074 struct kvm_dirty_gfn {
7076 __u32 slot; /* as_id | slot_id */
7080 The following values are defined for the flags field to define the
7081 current state of the entry::
7083 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
7084 #define KVM_DIRTY_GFN_F_RESET BIT(1)
7085 #define KVM_DIRTY_GFN_F_MASK 0x3
7087 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
7088 ioctl to enable this capability for the new guest and set the size of
7089 the rings. Enabling the capability is only allowed before creating any
7090 vCPU, and the size of the ring must be a power of two. The larger the
7091 ring buffer, the less likely the ring is full and the VM is forced to
7092 exit to userspace. The optimal size depends on the workload, but it is
7093 recommended that it be at least 64 KiB (4096 entries).
7095 Just like for dirty page bitmaps, the buffer tracks writes to
7096 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
7097 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
7098 with the flag set, userspace can start harvesting dirty pages from the
7101 An entry in the ring buffer can be unused (flag bits ``00``),
7102 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
7103 state machine for the entry is as follows::
7105 dirtied harvested reset
7106 00 -----------> 01 -------------> 1X -------+
7109 +------------------------------------------+
7111 To harvest the dirty pages, userspace accesses the mmaped ring buffer
7112 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
7113 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
7114 The userspace should harvest this GFN and mark the flags from state
7115 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
7116 to show that this GFN is harvested and waiting for a reset), and move
7117 on to the next GFN. The userspace should continue to do this until the
7118 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
7119 all the dirty GFNs that were available.
7121 It's not necessary for userspace to harvest the all dirty GFNs at once.
7122 However it must collect the dirty GFNs in sequence, i.e., the userspace
7123 program cannot skip one dirty GFN to collect the one next to it.
7125 After processing one or more entries in the ring buffer, userspace
7126 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
7127 it, so that the kernel will reprotect those collected GFNs.
7128 Therefore, the ioctl must be called *before* reading the content of
7131 The dirty ring can get full. When it happens, the KVM_RUN of the
7132 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
7134 The dirty ring interface has a major difference comparing to the
7135 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
7136 userspace, it's still possible that the kernel has not yet flushed the
7137 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
7138 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
7139 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
7140 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
7142 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
7143 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
7144 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
7145 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
7146 machine will switch to ring-buffer dirty page tracking and further
7147 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
7149 8.30 KVM_CAP_XEN_HVM
7150 --------------------
7154 This capability indicates the features that Xen supports for hosting Xen
7155 PVHVM guests. Valid flags are::
7157 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
7158 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
7159 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
7160 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2)
7162 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
7163 ioctl is available, for the guest to set its hypercall page.
7165 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
7166 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
7167 contents, to request that KVM generate hypercall page content automatically
7168 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
7170 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
7171 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
7172 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
7173 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
7176 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
7177 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
7178 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
7180 8.31 KVM_CAP_PPC_MULTITCE
7181 -------------------------
7183 :Capability: KVM_CAP_PPC_MULTITCE
7187 This capability means the kernel is capable of handling hypercalls
7188 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
7189 space. This significantly accelerates DMA operations for PPC KVM guests.
7190 User space should expect that its handlers for these hypercalls
7191 are not going to be called if user space previously registered LIOBN
7192 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
7194 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
7195 user space might have to advertise it for the guest. For example,
7196 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
7197 present in the "ibm,hypertas-functions" device-tree property.
7199 The hypercalls mentioned above may or may not be processed successfully
7200 in the kernel based fast path. If they can not be handled by the kernel,
7201 they will get passed on to user space. So user space still has to have
7202 an implementation for these despite the in kernel acceleration.
7204 This capability is always enabled.
7206 8.32 KVM_CAP_PTP_KVM
7207 --------------------
7209 :Architectures: arm64
7211 This capability indicates that the KVM virtual PTP service is
7212 supported in the host. A VMM can check whether the service is
7213 available to the guest on migration.
7215 8.33 KVM_CAP_HYPERV_ENFORCE_CPUID
7216 -----------------------------
7220 When enabled, KVM will disable emulated Hyper-V features provided to the
7221 guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
7222 currently implmented Hyper-V features are provided unconditionally when
7223 Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
7226 8.34 KVM_CAP_EXIT_HYPERCALL
7227 ---------------------------
7229 :Capability: KVM_CAP_EXIT_HYPERCALL
7233 This capability, if enabled, will cause KVM to exit to userspace
7234 with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
7236 Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
7237 of hypercalls that can be configured to exit to userspace.
7238 Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
7240 The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
7241 of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace
7242 the hypercalls whose corresponding bit is in the argument, and return
7243 ENOSYS for the others.