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 On arm64, the physical address size for a VM (IPA Size limit) is limited
155 to 40bits by default. The limit can be configured if the host supports the
156 extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
157 KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
158 identifier, where IPA_Bits is the maximum width of any physical
159 address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
160 machine type identifier.
162 e.g, to configure a guest to use 48bit physical address size::
164 vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
166 The requested size (IPA_Bits) must be:
168 == =========================================================
169 0 Implies default size, 40bits (for backward compatibility)
170 N Implies N bits, where N is a positive integer such that,
171 32 <= N <= Host_IPA_Limit
172 == =========================================================
174 Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
175 is dependent on the CPU capability and the kernel configuration. The limit can
176 be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
179 Creation of the VM will fail if the requested IPA size (whether it is
180 implicit or explicit) is unsupported on the host.
182 Please note that configuring the IPA size does not affect the capability
183 exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
184 size of the address translated by the stage2 level (guest physical to
185 host physical address translations).
188 4.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
189 ----------------------------------------------------------
191 :Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
194 :Parameters: struct kvm_msr_list (in/out)
195 :Returns: 0 on success; -1 on error
199 ====== ============================================================
200 EFAULT the msr index list cannot be read from or written to
201 E2BIG the msr index list is too big to fit in the array specified by
203 ====== ============================================================
207 struct kvm_msr_list {
208 __u32 nmsrs; /* number of msrs in entries */
212 The user fills in the size of the indices array in nmsrs, and in return
213 kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
214 indices array with their numbers.
216 KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported. The list
217 varies by kvm version and host processor, but does not change otherwise.
219 Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
220 not returned in the MSR list, as different vcpus can have a different number
221 of banks, as set via the KVM_X86_SETUP_MCE ioctl.
223 KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
224 to the KVM_GET_MSRS system ioctl. This lets userspace probe host capabilities
225 and processor features that are exposed via MSRs (e.g., VMX capabilities).
226 This list also varies by kvm version and host processor, but does not change
230 4.4 KVM_CHECK_EXTENSION
231 -----------------------
233 :Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
235 :Type: system ioctl, vm ioctl
236 :Parameters: extension identifier (KVM_CAP_*)
237 :Returns: 0 if unsupported; 1 (or some other positive integer) if supported
239 The API allows the application to query about extensions to the core
240 kvm API. Userspace passes an extension identifier (an integer) and
241 receives an integer that describes the extension availability.
242 Generally 0 means no and 1 means yes, but some extensions may report
243 additional information in the integer return value.
245 Based on their initialization different VMs may have different capabilities.
246 It is thus encouraged to use the vm ioctl to query for capabilities (available
247 with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
249 4.5 KVM_GET_VCPU_MMAP_SIZE
250 --------------------------
256 :Returns: size of vcpu mmap area, in bytes
258 The KVM_RUN ioctl (cf.) communicates with userspace via a shared
259 memory region. This ioctl returns the size of that region. See the
260 KVM_RUN documentation for details.
262 Besides the size of the KVM_RUN communication region, other areas of
263 the VCPU file descriptor can be mmap-ed, including:
265 - if KVM_CAP_COALESCED_MMIO is available, a page at
266 KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
267 this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
268 KVM_CAP_COALESCED_MMIO is not documented yet.
270 - if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
271 KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE. For more information on
272 KVM_CAP_DIRTY_LOG_RING, see section 8.3.
281 :Parameters: vcpu id (apic id on x86)
282 :Returns: vcpu fd on success, -1 on error
284 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
285 The vcpu id is an integer in the range [0, max_vcpu_id).
287 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
288 the KVM_CHECK_EXTENSION ioctl() at run-time.
289 The maximum possible value for max_vcpus can be retrieved using the
290 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
292 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
294 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
295 same as the value returned from KVM_CAP_NR_VCPUS.
297 The maximum possible value for max_vcpu_id can be retrieved using the
298 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
300 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
301 is the same as the value returned from KVM_CAP_MAX_VCPUS.
303 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
304 threads in one or more virtual CPU cores. (This is because the
305 hardware requires all the hardware threads in a CPU core to be in the
306 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
307 of vcpus per virtual core (vcore). The vcore id is obtained by
308 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
309 given vcore will always be in the same physical core as each other
310 (though that might be a different physical core from time to time).
311 Userspace can control the threading (SMT) mode of the guest by its
312 allocation of vcpu ids. For example, if userspace wants
313 single-threaded guest vcpus, it should make all vcpu ids be a multiple
314 of the number of vcpus per vcore.
316 For virtual cpus that have been created with S390 user controlled virtual
317 machines, the resulting vcpu fd can be memory mapped at page offset
318 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
319 cpu's hardware control block.
322 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
323 --------------------------------
328 :Parameters: struct kvm_dirty_log (in/out)
329 :Returns: 0 on success, -1 on error
333 /* for KVM_GET_DIRTY_LOG */
334 struct kvm_dirty_log {
338 void __user *dirty_bitmap; /* one bit per page */
343 Given a memory slot, return a bitmap containing any pages dirtied
344 since the last call to this ioctl. Bit 0 is the first page in the
345 memory slot. Ensure the entire structure is cleared to avoid padding
348 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
349 the address space for which you want to return the dirty bitmap. See
350 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
352 The bits in the dirty bitmap are cleared before the ioctl returns, unless
353 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
354 see the description of the capability.
356 Note that the Xen shared info page, if configured, shall always be assumed
357 to be dirty. KVM will not explicitly mark it such.
367 :Returns: 0 on success, -1 on error
371 ======= ==============================================================
372 EINTR an unmasked signal is pending
373 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
374 instructions from device memory (arm64)
375 ENOSYS data abort outside memslots with no syndrome info and
376 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
377 EPERM SVE feature set but not finalized (arm64)
378 ======= ==============================================================
380 This ioctl is used to run a guest virtual cpu. While there are no
381 explicit parameters, there is an implicit parameter block that can be
382 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
383 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
384 kvm_run' (see below).
391 :Architectures: all except arm64
393 :Parameters: struct kvm_regs (out)
394 :Returns: 0 on success, -1 on error
396 Reads the general purpose registers from the vcpu.
402 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
403 __u64 rax, rbx, rcx, rdx;
404 __u64 rsi, rdi, rsp, rbp;
405 __u64 r8, r9, r10, r11;
406 __u64 r12, r13, r14, r15;
412 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
424 :Architectures: all except arm64
426 :Parameters: struct kvm_regs (in)
427 :Returns: 0 on success, -1 on error
429 Writes the general purpose registers into the vcpu.
431 See KVM_GET_REGS for the data structure.
438 :Architectures: x86, ppc
440 :Parameters: struct kvm_sregs (out)
441 :Returns: 0 on success, -1 on error
443 Reads special registers from the vcpu.
449 struct kvm_segment cs, ds, es, fs, gs, ss;
450 struct kvm_segment tr, ldt;
451 struct kvm_dtable gdt, idt;
452 __u64 cr0, cr2, cr3, cr4, cr8;
455 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
458 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
460 interrupt_bitmap is a bitmap of pending external interrupts. At most
461 one bit may be set. This interrupt has been acknowledged by the APIC
462 but not yet injected into the cpu core.
469 :Architectures: x86, ppc
471 :Parameters: struct kvm_sregs (in)
472 :Returns: 0 on success, -1 on error
474 Writes special registers into the vcpu. See KVM_GET_SREGS for the
484 :Parameters: struct kvm_translation (in/out)
485 :Returns: 0 on success, -1 on error
487 Translates a virtual address according to the vcpu's current address
492 struct kvm_translation {
494 __u64 linear_address;
497 __u64 physical_address;
509 :Architectures: x86, ppc, mips, riscv
511 :Parameters: struct kvm_interrupt (in)
512 :Returns: 0 on success, negative on failure.
514 Queues a hardware interrupt vector to be injected.
518 /* for KVM_INTERRUPT */
519 struct kvm_interrupt {
529 ========= ===================================
531 -EEXIST if an interrupt is already enqueued
532 -EINVAL the irq number is invalid
533 -ENXIO if the PIC is in the kernel
534 -EFAULT if the pointer is invalid
535 ========= ===================================
537 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
538 ioctl is useful if the in-kernel PIC is not used.
543 Queues an external interrupt to be injected. This ioctl is overleaded
544 with 3 different irq values:
548 This injects an edge type external interrupt into the guest once it's ready
549 to receive interrupts. When injected, the interrupt is done.
551 b) KVM_INTERRUPT_UNSET
553 This unsets any pending interrupt.
555 Only available with KVM_CAP_PPC_UNSET_IRQ.
557 c) KVM_INTERRUPT_SET_LEVEL
559 This injects a level type external interrupt into the guest context. The
560 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
563 Only available with KVM_CAP_PPC_IRQ_LEVEL.
565 Note that any value for 'irq' other than the ones stated above is invalid
566 and incurs unexpected behavior.
568 This is an asynchronous vcpu ioctl and can be invoked from any thread.
573 Queues an external interrupt to be injected into the virtual CPU. A negative
574 interrupt number dequeues the interrupt.
576 This is an asynchronous vcpu ioctl and can be invoked from any thread.
581 Queues an external interrupt to be injected into the virtual CPU. This ioctl
582 is overloaded with 2 different irq values:
586 This sets external interrupt for a virtual CPU and it will receive
589 b) KVM_INTERRUPT_UNSET
591 This clears pending external interrupt for a virtual CPU.
593 This is an asynchronous vcpu ioctl and can be invoked from any thread.
603 :Returns: -1 on error
605 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
611 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
613 :Type: system ioctl, vcpu ioctl
614 :Parameters: struct kvm_msrs (in/out)
615 :Returns: number of msrs successfully returned;
618 When used as a system ioctl:
619 Reads the values of MSR-based features that are available for the VM. This
620 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
621 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
624 When used as a vcpu ioctl:
625 Reads model-specific registers from the vcpu. Supported msr indices can
626 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
631 __u32 nmsrs; /* number of msrs in entries */
634 struct kvm_msr_entry entries[0];
637 struct kvm_msr_entry {
643 Application code should set the 'nmsrs' member (which indicates the
644 size of the entries array) and the 'index' member of each array entry.
645 kvm will fill in the 'data' member.
654 :Parameters: struct kvm_msrs (in)
655 :Returns: number of msrs successfully set (see below), -1 on error
657 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
660 Application code should set the 'nmsrs' member (which indicates the
661 size of the entries array), and the 'index' and 'data' members of each
664 It tries to set the MSRs in array entries[] one by one. If setting an MSR
665 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
666 by KVM, etc..., it stops processing the MSR list and returns the number of
667 MSRs that have been set successfully.
676 :Parameters: struct kvm_cpuid (in)
677 :Returns: 0 on success, -1 on error
679 Defines the vcpu responses to the cpuid instruction. Applications
680 should use the KVM_SET_CPUID2 ioctl if available.
683 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID
684 configuration (if there is) is not corrupted. Userspace can get a copy
685 of the resulting CPUID configuration through KVM_GET_CPUID2 in case.
686 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model
687 after running the guest, may cause guest instability.
688 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc...
689 may cause guest instability.
693 struct kvm_cpuid_entry {
702 /* for KVM_SET_CPUID */
706 struct kvm_cpuid_entry entries[0];
710 4.21 KVM_SET_SIGNAL_MASK
711 ------------------------
716 :Parameters: struct kvm_signal_mask (in)
717 :Returns: 0 on success, -1 on error
719 Defines which signals are blocked during execution of KVM_RUN. This
720 signal mask temporarily overrides the threads signal mask. Any
721 unblocked signal received (except SIGKILL and SIGSTOP, which retain
722 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
724 Note the signal will only be delivered if not blocked by the original
729 /* for KVM_SET_SIGNAL_MASK */
730 struct kvm_signal_mask {
742 :Parameters: struct kvm_fpu (out)
743 :Returns: 0 on success, -1 on error
745 Reads the floating point state from the vcpu.
749 /* for KVM_GET_FPU and KVM_SET_FPU */
754 __u8 ftwx; /* in fxsave format */
771 :Parameters: struct kvm_fpu (in)
772 :Returns: 0 on success, -1 on error
774 Writes the floating point state to the vcpu.
778 /* for KVM_GET_FPU and KVM_SET_FPU */
783 __u8 ftwx; /* in fxsave format */
794 4.24 KVM_CREATE_IRQCHIP
795 -----------------------
797 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
798 :Architectures: x86, arm64, s390
801 :Returns: 0 on success, -1 on error
803 Creates an interrupt controller model in the kernel.
804 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
805 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
806 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
807 On arm64, a GICv2 is created. Any other GIC versions require the usage of
808 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
809 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
810 On s390, a dummy irq routing table is created.
812 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
813 before KVM_CREATE_IRQCHIP can be used.
819 :Capability: KVM_CAP_IRQCHIP
820 :Architectures: x86, arm64
822 :Parameters: struct kvm_irq_level
823 :Returns: 0 on success, -1 on error
825 Sets the level of a GSI input to the interrupt controller model in the kernel.
826 On some architectures it is required that an interrupt controller model has
827 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
828 interrupts require the level to be set to 1 and then back to 0.
830 On real hardware, interrupt pins can be active-low or active-high. This
831 does not matter for the level field of struct kvm_irq_level: 1 always
832 means active (asserted), 0 means inactive (deasserted).
834 x86 allows the operating system to program the interrupt polarity
835 (active-low/active-high) for level-triggered interrupts, and KVM used
836 to consider the polarity. However, due to bitrot in the handling of
837 active-low interrupts, the above convention is now valid on x86 too.
838 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
839 should not present interrupts to the guest as active-low unless this
840 capability is present (or unless it is not using the in-kernel irqchip,
844 arm64 can signal an interrupt either at the CPU level, or at the
845 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
846 use PPIs designated for specific cpus. The irq field is interpreted
849 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
850 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
852 The irq_type field has the following values:
855 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
857 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
858 (the vcpu_index field is ignored)
860 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
862 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
864 In both cases, level is used to assert/deassert the line.
866 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
867 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
870 Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions
871 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
872 be used for a userspace interrupt controller.
876 struct kvm_irq_level {
879 __s32 status; /* not used for KVM_IRQ_LEVEL */
881 __u32 level; /* 0 or 1 */
888 :Capability: KVM_CAP_IRQCHIP
891 :Parameters: struct kvm_irqchip (in/out)
892 :Returns: 0 on success, -1 on error
894 Reads the state of a kernel interrupt controller created with
895 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
900 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
903 char dummy[512]; /* reserving space */
904 struct kvm_pic_state pic;
905 struct kvm_ioapic_state ioapic;
913 :Capability: KVM_CAP_IRQCHIP
916 :Parameters: struct kvm_irqchip (in)
917 :Returns: 0 on success, -1 on error
919 Sets the state of a kernel interrupt controller created with
920 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
925 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
928 char dummy[512]; /* reserving space */
929 struct kvm_pic_state pic;
930 struct kvm_ioapic_state ioapic;
935 4.28 KVM_XEN_HVM_CONFIG
936 -----------------------
938 :Capability: KVM_CAP_XEN_HVM
941 :Parameters: struct kvm_xen_hvm_config (in)
942 :Returns: 0 on success, -1 on error
944 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
945 page, and provides the starting address and size of the hypercall
946 blobs in userspace. When the guest writes the MSR, kvm copies one
947 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
952 struct kvm_xen_hvm_config {
962 If certain flags are returned from the KVM_CAP_XEN_HVM check, they may
963 be set in the flags field of this ioctl:
965 The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate
966 the contents of the hypercall page automatically; hypercalls will be
967 intercepted and passed to userspace through KVM_EXIT_XEN. In this
968 ase, all of the blob size and address fields must be zero.
970 The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace
971 will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event
972 channel interrupts rather than manipulating the guest's shared_info
973 structures directly. This, in turn, may allow KVM to enable features
974 such as intercepting the SCHEDOP_poll hypercall to accelerate PV
975 spinlock operation for the guest. Userspace may still use the ioctl
976 to deliver events if it was advertised, even if userspace does not
977 send this indication that it will always do so
979 No other flags are currently valid in the struct kvm_xen_hvm_config.
984 :Capability: KVM_CAP_ADJUST_CLOCK
987 :Parameters: struct kvm_clock_data (out)
988 :Returns: 0 on success, -1 on error
990 Gets the current timestamp of kvmclock as seen by the current guest. In
991 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
994 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
995 set of bits that KVM can return in struct kvm_clock_data's flag member.
997 The following flags are defined:
1000 If set, the returned value is the exact kvmclock
1001 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called.
1002 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant
1003 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try
1004 to make all VCPUs follow this clock, but the exact value read by each
1005 VCPU could differ, because the host TSC is not stable.
1008 If set, the `realtime` field in the kvm_clock_data
1009 structure is populated with the value of the host's real time
1010 clocksource at the instant when KVM_GET_CLOCK was called. If clear,
1011 the `realtime` field does not contain a value.
1014 If set, the `host_tsc` field in the kvm_clock_data
1015 structure is populated with the value of the host's timestamp counter (TSC)
1016 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field
1017 does not contain a value.
1021 struct kvm_clock_data {
1022 __u64 clock; /* kvmclock current value */
1034 :Capability: KVM_CAP_ADJUST_CLOCK
1037 :Parameters: struct kvm_clock_data (in)
1038 :Returns: 0 on success, -1 on error
1040 Sets the current timestamp of kvmclock to the value specified in its parameter.
1041 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1044 The following flags can be passed:
1047 If set, KVM will compare the value of the `realtime` field
1048 with the value of the host's real time clocksource at the instant when
1049 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final
1050 kvmclock value that will be provided to guests.
1052 Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored.
1056 struct kvm_clock_data {
1057 __u64 clock; /* kvmclock current value */
1066 4.31 KVM_GET_VCPU_EVENTS
1067 ------------------------
1069 :Capability: KVM_CAP_VCPU_EVENTS
1070 :Extended by: KVM_CAP_INTR_SHADOW
1071 :Architectures: x86, arm64
1073 :Parameters: struct kvm_vcpu_event (out)
1074 :Returns: 0 on success, -1 on error
1079 Gets currently pending exceptions, interrupts, and NMIs as well as related
1084 struct kvm_vcpu_events {
1088 __u8 has_error_code;
1109 __u8 smm_inside_nmi;
1113 __u8 exception_has_payload;
1114 __u64 exception_payload;
1117 The following bits are defined in the flags field:
1119 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1120 interrupt.shadow contains a valid state.
1122 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1125 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1126 exception_has_payload, exception_payload, and exception.pending
1127 fields contain a valid state. This bit will be set whenever
1128 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1130 - KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the
1131 triple_fault_pending field contains a valid state. This bit will
1132 be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled.
1137 If the guest accesses a device that is being emulated by the host kernel in
1138 such a way that a real device would generate a physical SError, KVM may make
1139 a virtual SError pending for that VCPU. This system error interrupt remains
1140 pending until the guest takes the exception by unmasking PSTATE.A.
1142 Running the VCPU may cause it to take a pending SError, or make an access that
1143 causes an SError to become pending. The event's description is only valid while
1144 the VPCU is not running.
1146 This API provides a way to read and write the pending 'event' state that is not
1147 visible to the guest. To save, restore or migrate a VCPU the struct representing
1148 the state can be read then written using this GET/SET API, along with the other
1149 guest-visible registers. It is not possible to 'cancel' an SError that has been
1152 A device being emulated in user-space may also wish to generate an SError. To do
1153 this the events structure can be populated by user-space. The current state
1154 should be read first, to ensure no existing SError is pending. If an existing
1155 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1156 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1157 Serviceability (RAS) Specification").
1159 SError exceptions always have an ESR value. Some CPUs have the ability to
1160 specify what the virtual SError's ESR value should be. These systems will
1161 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1162 always have a non-zero value when read, and the agent making an SError pending
1163 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1164 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1165 with exception.has_esr as zero, KVM will choose an ESR.
1167 Specifying exception.has_esr on a system that does not support it will return
1168 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1169 will return -EINVAL.
1171 It is not possible to read back a pending external abort (injected via
1172 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1173 directly to the virtual CPU).
1177 struct kvm_vcpu_events {
1179 __u8 serror_pending;
1180 __u8 serror_has_esr;
1181 __u8 ext_dabt_pending;
1182 /* Align it to 8 bytes */
1189 4.32 KVM_SET_VCPU_EVENTS
1190 ------------------------
1192 :Capability: KVM_CAP_VCPU_EVENTS
1193 :Extended by: KVM_CAP_INTR_SHADOW
1194 :Architectures: x86, arm64
1196 :Parameters: struct kvm_vcpu_event (in)
1197 :Returns: 0 on success, -1 on error
1202 Set pending exceptions, interrupts, and NMIs as well as related states of the
1205 See KVM_GET_VCPU_EVENTS for the data structure.
1207 Fields that may be modified asynchronously by running VCPUs can be excluded
1208 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1209 smi.pending. Keep the corresponding bits in the flags field cleared to
1210 suppress overwriting the current in-kernel state. The bits are:
1212 =============================== ==================================
1213 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1214 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1215 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1216 =============================== ==================================
1218 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1219 the flags field to signal that interrupt.shadow contains a valid state and
1220 shall be written into the VCPU.
1222 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1224 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1225 can be set in the flags field to signal that the
1226 exception_has_payload, exception_payload, and exception.pending fields
1227 contain a valid state and shall be written into the VCPU.
1229 If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT
1230 can be set in flags field to signal that the triple_fault field contains
1231 a valid state and shall be written into the VCPU.
1236 User space may need to inject several types of events to the guest.
1238 Set the pending SError exception state for this VCPU. It is not possible to
1239 'cancel' an Serror that has been made pending.
1241 If the guest performed an access to I/O memory which could not be handled by
1242 userspace, for example because of missing instruction syndrome decode
1243 information or because there is no device mapped at the accessed IPA, then
1244 userspace can ask the kernel to inject an external abort using the address
1245 from the exiting fault on the VCPU. It is a programming error to set
1246 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1247 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1248 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1249 how userspace reports accesses for the above cases to guests, across different
1250 userspace implementations. Nevertheless, userspace can still emulate all Arm
1251 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1253 See KVM_GET_VCPU_EVENTS for the data structure.
1256 4.33 KVM_GET_DEBUGREGS
1257 ----------------------
1259 :Capability: KVM_CAP_DEBUGREGS
1262 :Parameters: struct kvm_debugregs (out)
1263 :Returns: 0 on success, -1 on error
1265 Reads debug registers from the vcpu.
1269 struct kvm_debugregs {
1278 4.34 KVM_SET_DEBUGREGS
1279 ----------------------
1281 :Capability: KVM_CAP_DEBUGREGS
1284 :Parameters: struct kvm_debugregs (in)
1285 :Returns: 0 on success, -1 on error
1287 Writes debug registers into the vcpu.
1289 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1290 yet and must be cleared on entry.
1293 4.35 KVM_SET_USER_MEMORY_REGION
1294 -------------------------------
1296 :Capability: KVM_CAP_USER_MEMORY
1299 :Parameters: struct kvm_userspace_memory_region (in)
1300 :Returns: 0 on success, -1 on error
1304 struct kvm_userspace_memory_region {
1307 __u64 guest_phys_addr;
1308 __u64 memory_size; /* bytes */
1309 __u64 userspace_addr; /* start of the userspace allocated memory */
1312 /* for kvm_userspace_memory_region::flags */
1313 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1314 #define KVM_MEM_READONLY (1UL << 1)
1316 This ioctl allows the user to create, modify or delete a guest physical
1317 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1318 should be less than the maximum number of user memory slots supported per
1319 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1320 Slots may not overlap in guest physical address space.
1322 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1323 specifies the address space which is being modified. They must be
1324 less than the value that KVM_CHECK_EXTENSION returns for the
1325 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1326 are unrelated; the restriction on overlapping slots only applies within
1329 Deleting a slot is done by passing zero for memory_size. When changing
1330 an existing slot, it may be moved in the guest physical memory space,
1331 or its flags may be modified, but it may not be resized.
1333 Memory for the region is taken starting at the address denoted by the
1334 field userspace_addr, which must point at user addressable memory for
1335 the entire memory slot size. Any object may back this memory, including
1336 anonymous memory, ordinary files, and hugetlbfs.
1338 On architectures that support a form of address tagging, userspace_addr must
1339 be an untagged address.
1341 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1342 be identical. This allows large pages in the guest to be backed by large
1345 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1346 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1347 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1348 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1349 to make a new slot read-only. In this case, writes to this memory will be
1350 posted to userspace as KVM_EXIT_MMIO exits.
1352 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1353 the memory region are automatically reflected into the guest. For example, an
1354 mmap() that affects the region will be made visible immediately. Another
1355 example is madvise(MADV_DROP).
1357 Note: On arm64, a write generated by the page-table walker (to update
1358 the Access and Dirty flags, for example) never results in a
1359 KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This
1360 is because KVM cannot provide the data that would be written by the
1361 page-table walker, making it impossible to emulate the access.
1362 Instead, an abort (data abort if the cause of the page-table update
1363 was a load or a store, instruction abort if it was an instruction
1364 fetch) is injected in the guest.
1366 4.36 KVM_SET_TSS_ADDR
1367 ---------------------
1369 :Capability: KVM_CAP_SET_TSS_ADDR
1372 :Parameters: unsigned long tss_address (in)
1373 :Returns: 0 on success, -1 on error
1375 This ioctl defines the physical address of a three-page region in the guest
1376 physical address space. The region must be within the first 4GB of the
1377 guest physical address space and must not conflict with any memory slot
1378 or any mmio address. The guest may malfunction if it accesses this memory
1381 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1382 because of a quirk in the virtualization implementation (see the internals
1383 documentation when it pops into existence).
1389 :Capability: KVM_CAP_ENABLE_CAP
1390 :Architectures: mips, ppc, s390, x86
1392 :Parameters: struct kvm_enable_cap (in)
1393 :Returns: 0 on success; -1 on error
1395 :Capability: KVM_CAP_ENABLE_CAP_VM
1398 :Parameters: struct kvm_enable_cap (in)
1399 :Returns: 0 on success; -1 on error
1403 Not all extensions are enabled by default. Using this ioctl the application
1404 can enable an extension, making it available to the guest.
1406 On systems that do not support this ioctl, it always fails. On systems that
1407 do support it, it only works for extensions that are supported for enablement.
1409 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1414 struct kvm_enable_cap {
1418 The capability that is supposed to get enabled.
1424 A bitfield indicating future enhancements. Has to be 0 for now.
1430 Arguments for enabling a feature. If a feature needs initial values to
1431 function properly, this is the place to put them.
1438 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1439 for vm-wide capabilities.
1441 4.38 KVM_GET_MP_STATE
1442 ---------------------
1444 :Capability: KVM_CAP_MP_STATE
1445 :Architectures: x86, s390, arm64, riscv
1447 :Parameters: struct kvm_mp_state (out)
1448 :Returns: 0 on success; -1 on error
1452 struct kvm_mp_state {
1456 Returns the vcpu's current "multiprocessing state" (though also valid on
1457 uniprocessor guests).
1459 Possible values are:
1461 ========================== ===============================================
1462 KVM_MP_STATE_RUNNABLE the vcpu is currently running
1464 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1465 which has not yet received an INIT signal [x86]
1466 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1467 now ready for a SIPI [x86]
1468 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1469 is waiting for an interrupt [x86]
1470 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1471 accessible via KVM_GET_VCPU_EVENTS) [x86]
1472 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv]
1473 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1474 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1476 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1478 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting
1479 for a wakeup event [arm64]
1480 ========================== ===============================================
1482 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1483 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1484 these architectures.
1489 If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the
1490 architectural execution of a WFI instruction.
1492 If a wakeup event is recognized, KVM will exit to userspace with a
1493 KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If
1494 userspace wants to honor the wakeup, it must set the vCPU's MP state to
1495 KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup
1496 event in subsequent calls to KVM_RUN.
1500 If userspace intends to keep the vCPU in a SUSPENDED state, it is
1501 strongly recommended that userspace take action to suppress the
1502 wakeup event (such as masking an interrupt). Otherwise, subsequent
1503 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP
1504 event and inadvertently waste CPU cycles.
1506 Additionally, if userspace takes action to suppress a wakeup event,
1507 it is strongly recommended that it also restores the vCPU to its
1508 original state when the vCPU is made RUNNABLE again. For example,
1509 if userspace masked a pending interrupt to suppress the wakeup,
1510 the interrupt should be unmasked before returning control to the
1516 The only states that are valid are KVM_MP_STATE_STOPPED and
1517 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1519 4.39 KVM_SET_MP_STATE
1520 ---------------------
1522 :Capability: KVM_CAP_MP_STATE
1523 :Architectures: x86, s390, arm64, riscv
1525 :Parameters: struct kvm_mp_state (in)
1526 :Returns: 0 on success; -1 on error
1528 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1531 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1532 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1533 these architectures.
1538 The only states that are valid are KVM_MP_STATE_STOPPED and
1539 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1541 4.40 KVM_SET_IDENTITY_MAP_ADDR
1542 ------------------------------
1544 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1547 :Parameters: unsigned long identity (in)
1548 :Returns: 0 on success, -1 on error
1550 This ioctl defines the physical address of a one-page region in the guest
1551 physical address space. The region must be within the first 4GB of the
1552 guest physical address space and must not conflict with any memory slot
1553 or any mmio address. The guest may malfunction if it accesses this memory
1556 Setting the address to 0 will result in resetting the address to its default
1559 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1560 because of a quirk in the virtualization implementation (see the internals
1561 documentation when it pops into existence).
1563 Fails if any VCPU has already been created.
1565 4.41 KVM_SET_BOOT_CPU_ID
1566 ------------------------
1568 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1571 :Parameters: unsigned long vcpu_id
1572 :Returns: 0 on success, -1 on error
1574 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1575 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1576 is vcpu 0. This ioctl has to be called before vcpu creation,
1577 otherwise it will return EBUSY error.
1583 :Capability: KVM_CAP_XSAVE
1586 :Parameters: struct kvm_xsave (out)
1587 :Returns: 0 on success, -1 on error
1597 This ioctl would copy current vcpu's xsave struct to the userspace.
1603 :Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2
1606 :Parameters: struct kvm_xsave (in)
1607 :Returns: 0 on success, -1 on error
1617 This ioctl would copy userspace's xsave struct to the kernel. It copies
1618 as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2),
1619 when invoked on the vm file descriptor. The size value returned by
1620 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
1621 Currently, it is only greater than 4096 if a dynamic feature has been
1622 enabled with ``arch_prctl()``, but this may change in the future.
1624 The offsets of the state save areas in struct kvm_xsave follow the
1625 contents of CPUID leaf 0xD on the host.
1631 :Capability: KVM_CAP_XCRS
1634 :Parameters: struct kvm_xcrs (out)
1635 :Returns: 0 on success, -1 on error
1648 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1652 This ioctl would copy current vcpu's xcrs to the userspace.
1658 :Capability: KVM_CAP_XCRS
1661 :Parameters: struct kvm_xcrs (in)
1662 :Returns: 0 on success, -1 on error
1675 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1679 This ioctl would set vcpu's xcr to the value userspace specified.
1682 4.46 KVM_GET_SUPPORTED_CPUID
1683 ----------------------------
1685 :Capability: KVM_CAP_EXT_CPUID
1688 :Parameters: struct kvm_cpuid2 (in/out)
1689 :Returns: 0 on success, -1 on error
1696 struct kvm_cpuid_entry2 entries[0];
1699 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1700 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1701 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1703 struct kvm_cpuid_entry2 {
1714 This ioctl returns x86 cpuid features which are supported by both the
1715 hardware and kvm in its default configuration. Userspace can use the
1716 information returned by this ioctl to construct cpuid information (for
1717 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1718 userspace capabilities, and with user requirements (for example, the
1719 user may wish to constrain cpuid to emulate older hardware, or for
1720 feature consistency across a cluster).
1722 Dynamically-enabled feature bits need to be requested with
1723 ``arch_prctl()`` before calling this ioctl. Feature bits that have not
1724 been requested are excluded from the result.
1726 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1727 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1728 its default configuration. If userspace enables such capabilities, it
1729 is responsible for modifying the results of this ioctl appropriately.
1731 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1732 with the 'nent' field indicating the number of entries in the variable-size
1733 array 'entries'. If the number of entries is too low to describe the cpu
1734 capabilities, an error (E2BIG) is returned. If the number is too high,
1735 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1736 number is just right, the 'nent' field is adjusted to the number of valid
1737 entries in the 'entries' array, which is then filled.
1739 The entries returned are the host cpuid as returned by the cpuid instruction,
1740 with unknown or unsupported features masked out. Some features (for example,
1741 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1742 emulate them efficiently. The fields in each entry are defined as follows:
1745 the eax value used to obtain the entry
1748 the ecx value used to obtain the entry (for entries that are
1752 an OR of zero or more of the following:
1754 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1755 if the index field is valid
1758 the values returned by the cpuid instruction for
1759 this function/index combination
1761 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1762 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1763 support. Instead it is reported via::
1765 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1767 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1768 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1771 4.47 KVM_PPC_GET_PVINFO
1772 -----------------------
1774 :Capability: KVM_CAP_PPC_GET_PVINFO
1777 :Parameters: struct kvm_ppc_pvinfo (out)
1778 :Returns: 0 on success, !0 on error
1782 struct kvm_ppc_pvinfo {
1788 This ioctl fetches PV specific information that need to be passed to the guest
1789 using the device tree or other means from vm context.
1791 The hcall array defines 4 instructions that make up a hypercall.
1793 If any additional field gets added to this structure later on, a bit for that
1794 additional piece of information will be set in the flags bitmap.
1796 The flags bitmap is defined as::
1798 /* the host supports the ePAPR idle hcall
1799 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1801 4.52 KVM_SET_GSI_ROUTING
1802 ------------------------
1804 :Capability: KVM_CAP_IRQ_ROUTING
1805 :Architectures: x86 s390 arm64
1807 :Parameters: struct kvm_irq_routing (in)
1808 :Returns: 0 on success, -1 on error
1810 Sets the GSI routing table entries, overwriting any previously set entries.
1812 On arm64, GSI routing has the following limitation:
1814 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1818 struct kvm_irq_routing {
1821 struct kvm_irq_routing_entry entries[0];
1824 No flags are specified so far, the corresponding field must be set to zero.
1828 struct kvm_irq_routing_entry {
1834 struct kvm_irq_routing_irqchip irqchip;
1835 struct kvm_irq_routing_msi msi;
1836 struct kvm_irq_routing_s390_adapter adapter;
1837 struct kvm_irq_routing_hv_sint hv_sint;
1838 struct kvm_irq_routing_xen_evtchn xen_evtchn;
1843 /* gsi routing entry types */
1844 #define KVM_IRQ_ROUTING_IRQCHIP 1
1845 #define KVM_IRQ_ROUTING_MSI 2
1846 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1847 #define KVM_IRQ_ROUTING_HV_SINT 4
1848 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5
1852 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1853 type, specifies that the devid field contains a valid value. The per-VM
1854 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1855 the device ID. If this capability is not available, userspace should
1856 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1861 struct kvm_irq_routing_irqchip {
1866 struct kvm_irq_routing_msi {
1876 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1877 for the device that wrote the MSI message. For PCI, this is usually a
1878 BFD identifier in the lower 16 bits.
1880 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1881 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1882 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1883 address_hi must be zero.
1887 struct kvm_irq_routing_s390_adapter {
1891 __u32 summary_offset;
1895 struct kvm_irq_routing_hv_sint {
1900 struct kvm_irq_routing_xen_evtchn {
1907 When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit
1908 in its indication of supported features, routing to Xen event channels
1909 is supported. Although the priority field is present, only the value
1910 KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by
1911 2 level event channels. FIFO event channel support may be added in
1915 4.55 KVM_SET_TSC_KHZ
1916 --------------------
1918 :Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL
1920 :Type: vcpu ioctl / vm ioctl
1921 :Parameters: virtual tsc_khz
1922 :Returns: 0 on success, -1 on error
1924 Specifies the tsc frequency for the virtual machine. The unit of the
1927 If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also
1928 be used as a vm ioctl to set the initial tsc frequency of subsequently
1931 4.56 KVM_GET_TSC_KHZ
1932 --------------------
1934 :Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL
1936 :Type: vcpu ioctl / vm ioctl
1938 :Returns: virtual tsc-khz on success, negative value on error
1940 Returns the tsc frequency of the guest. The unit of the return value is
1941 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1948 :Capability: KVM_CAP_IRQCHIP
1951 :Parameters: struct kvm_lapic_state (out)
1952 :Returns: 0 on success, -1 on error
1956 #define KVM_APIC_REG_SIZE 0x400
1957 struct kvm_lapic_state {
1958 char regs[KVM_APIC_REG_SIZE];
1961 Reads the Local APIC registers and copies them into the input argument. The
1962 data format and layout are the same as documented in the architecture manual.
1964 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1965 enabled, then the format of APIC_ID register depends on the APIC mode
1966 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1967 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1968 which is stored in bits 31-24 of the APIC register, or equivalently in
1969 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1970 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1972 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1973 always uses xAPIC format.
1979 :Capability: KVM_CAP_IRQCHIP
1982 :Parameters: struct kvm_lapic_state (in)
1983 :Returns: 0 on success, -1 on error
1987 #define KVM_APIC_REG_SIZE 0x400
1988 struct kvm_lapic_state {
1989 char regs[KVM_APIC_REG_SIZE];
1992 Copies the input argument into the Local APIC registers. The data format
1993 and layout are the same as documented in the architecture manual.
1995 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1996 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1997 See the note in KVM_GET_LAPIC.
2003 :Capability: KVM_CAP_IOEVENTFD
2006 :Parameters: struct kvm_ioeventfd (in)
2007 :Returns: 0 on success, !0 on error
2009 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
2010 within the guest. A guest write in the registered address will signal the
2011 provided event instead of triggering an exit.
2015 struct kvm_ioeventfd {
2017 __u64 addr; /* legal pio/mmio address */
2018 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
2024 For the special case of virtio-ccw devices on s390, the ioevent is matched
2025 to a subchannel/virtqueue tuple instead.
2027 The following flags are defined::
2029 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
2030 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
2031 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
2032 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2033 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2035 If datamatch flag is set, the event will be signaled only if the written value
2036 to the registered address is equal to datamatch in struct kvm_ioeventfd.
2038 For virtio-ccw devices, addr contains the subchannel id and datamatch the
2041 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2042 the kernel will ignore the length of guest write and may get a faster vmexit.
2043 The speedup may only apply to specific architectures, but the ioeventfd will
2049 :Capability: KVM_CAP_SW_TLB
2052 :Parameters: struct kvm_dirty_tlb (in)
2053 :Returns: 0 on success, -1 on error
2057 struct kvm_dirty_tlb {
2062 This must be called whenever userspace has changed an entry in the shared
2063 TLB, prior to calling KVM_RUN on the associated vcpu.
2065 The "bitmap" field is the userspace address of an array. This array
2066 consists of a number of bits, equal to the total number of TLB entries as
2067 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
2068 nearest multiple of 64.
2070 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2073 The array is little-endian: the bit 0 is the least significant bit of the
2074 first byte, bit 8 is the least significant bit of the second byte, etc.
2075 This avoids any complications with differing word sizes.
2077 The "num_dirty" field is a performance hint for KVM to determine whether it
2078 should skip processing the bitmap and just invalidate everything. It must
2079 be set to the number of set bits in the bitmap.
2082 4.62 KVM_CREATE_SPAPR_TCE
2083 -------------------------
2085 :Capability: KVM_CAP_SPAPR_TCE
2086 :Architectures: powerpc
2088 :Parameters: struct kvm_create_spapr_tce (in)
2089 :Returns: file descriptor for manipulating the created TCE table
2091 This creates a virtual TCE (translation control entry) table, which
2092 is an IOMMU for PAPR-style virtual I/O. It is used to translate
2093 logical addresses used in virtual I/O into guest physical addresses,
2094 and provides a scatter/gather capability for PAPR virtual I/O.
2098 /* for KVM_CAP_SPAPR_TCE */
2099 struct kvm_create_spapr_tce {
2104 The liobn field gives the logical IO bus number for which to create a
2105 TCE table. The window_size field specifies the size of the DMA window
2106 which this TCE table will translate - the table will contain one 64
2107 bit TCE entry for every 4kiB of the DMA window.
2109 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2110 table has been created using this ioctl(), the kernel will handle it
2111 in real mode, updating the TCE table. H_PUT_TCE calls for other
2112 liobns will cause a vm exit and must be handled by userspace.
2114 The return value is a file descriptor which can be passed to mmap(2)
2115 to map the created TCE table into userspace. This lets userspace read
2116 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2117 userspace update the TCE table directly which is useful in some
2121 4.63 KVM_ALLOCATE_RMA
2122 ---------------------
2124 :Capability: KVM_CAP_PPC_RMA
2125 :Architectures: powerpc
2127 :Parameters: struct kvm_allocate_rma (out)
2128 :Returns: file descriptor for mapping the allocated RMA
2130 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2131 time by the kernel. An RMA is a physically-contiguous, aligned region
2132 of memory used on older POWER processors to provide the memory which
2133 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2134 POWER processors support a set of sizes for the RMA that usually
2135 includes 64MB, 128MB, 256MB and some larger powers of two.
2139 /* for KVM_ALLOCATE_RMA */
2140 struct kvm_allocate_rma {
2144 The return value is a file descriptor which can be passed to mmap(2)
2145 to map the allocated RMA into userspace. The mapped area can then be
2146 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2147 RMA for a virtual machine. The size of the RMA in bytes (which is
2148 fixed at host kernel boot time) is returned in the rma_size field of
2149 the argument structure.
2151 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2152 is supported; 2 if the processor requires all virtual machines to have
2153 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2154 because it supports the Virtual RMA (VRMA) facility.
2160 :Capability: KVM_CAP_USER_NMI
2164 :Returns: 0 on success, -1 on error
2166 Queues an NMI on the thread's vcpu. Note this is well defined only
2167 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2168 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2169 has been called, this interface is completely emulated within the kernel.
2171 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2172 following algorithm:
2175 - read the local APIC's state (KVM_GET_LAPIC)
2176 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2177 - if so, issue KVM_NMI
2180 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2184 4.65 KVM_S390_UCAS_MAP
2185 ----------------------
2187 :Capability: KVM_CAP_S390_UCONTROL
2188 :Architectures: s390
2190 :Parameters: struct kvm_s390_ucas_mapping (in)
2191 :Returns: 0 in case of success
2193 The parameter is defined like this::
2195 struct kvm_s390_ucas_mapping {
2201 This ioctl maps the memory at "user_addr" with the length "length" to
2202 the vcpu's address space starting at "vcpu_addr". All parameters need to
2203 be aligned by 1 megabyte.
2206 4.66 KVM_S390_UCAS_UNMAP
2207 ------------------------
2209 :Capability: KVM_CAP_S390_UCONTROL
2210 :Architectures: s390
2212 :Parameters: struct kvm_s390_ucas_mapping (in)
2213 :Returns: 0 in case of success
2215 The parameter is defined like this::
2217 struct kvm_s390_ucas_mapping {
2223 This ioctl unmaps the memory in the vcpu's address space starting at
2224 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2225 All parameters need to be aligned by 1 megabyte.
2228 4.67 KVM_S390_VCPU_FAULT
2229 ------------------------
2231 :Capability: KVM_CAP_S390_UCONTROL
2232 :Architectures: s390
2234 :Parameters: vcpu absolute address (in)
2235 :Returns: 0 in case of success
2237 This call creates a page table entry on the virtual cpu's address space
2238 (for user controlled virtual machines) or the virtual machine's address
2239 space (for regular virtual machines). This only works for minor faults,
2240 thus it's recommended to access subject memory page via the user page
2241 table upfront. This is useful to handle validity intercepts for user
2242 controlled virtual machines to fault in the virtual cpu's lowcore pages
2243 prior to calling the KVM_RUN ioctl.
2246 4.68 KVM_SET_ONE_REG
2247 --------------------
2249 :Capability: KVM_CAP_ONE_REG
2252 :Parameters: struct kvm_one_reg (in)
2253 :Returns: 0 on success, negative value on failure
2257 ====== ============================================================
2258 ENOENT no such register
2259 EINVAL invalid register ID, or no such register or used with VMs in
2260 protected virtualization mode on s390
2261 EPERM (arm64) register access not allowed before vcpu finalization
2262 EBUSY (riscv) changing register value not allowed after the vcpu
2263 has run at least once
2264 ====== ============================================================
2266 (These error codes are indicative only: do not rely on a specific error
2267 code being returned in a specific situation.)
2271 struct kvm_one_reg {
2276 Using this ioctl, a single vcpu register can be set to a specific value
2277 defined by user space with the passed in struct kvm_one_reg, where id
2278 refers to the register identifier as described below and addr is a pointer
2279 to a variable with the respective size. There can be architecture agnostic
2280 and architecture specific registers. Each have their own range of operation
2281 and their own constants and width. To keep track of the implemented
2282 registers, find a list below:
2284 ======= =============================== ============
2285 Arch Register Width (bits)
2286 ======= =============================== ============
2287 PPC KVM_REG_PPC_HIOR 64
2288 PPC KVM_REG_PPC_IAC1 64
2289 PPC KVM_REG_PPC_IAC2 64
2290 PPC KVM_REG_PPC_IAC3 64
2291 PPC KVM_REG_PPC_IAC4 64
2292 PPC KVM_REG_PPC_DAC1 64
2293 PPC KVM_REG_PPC_DAC2 64
2294 PPC KVM_REG_PPC_DABR 64
2295 PPC KVM_REG_PPC_DSCR 64
2296 PPC KVM_REG_PPC_PURR 64
2297 PPC KVM_REG_PPC_SPURR 64
2298 PPC KVM_REG_PPC_DAR 64
2299 PPC KVM_REG_PPC_DSISR 32
2300 PPC KVM_REG_PPC_AMR 64
2301 PPC KVM_REG_PPC_UAMOR 64
2302 PPC KVM_REG_PPC_MMCR0 64
2303 PPC KVM_REG_PPC_MMCR1 64
2304 PPC KVM_REG_PPC_MMCRA 64
2305 PPC KVM_REG_PPC_MMCR2 64
2306 PPC KVM_REG_PPC_MMCRS 64
2307 PPC KVM_REG_PPC_MMCR3 64
2308 PPC KVM_REG_PPC_SIAR 64
2309 PPC KVM_REG_PPC_SDAR 64
2310 PPC KVM_REG_PPC_SIER 64
2311 PPC KVM_REG_PPC_SIER2 64
2312 PPC KVM_REG_PPC_SIER3 64
2313 PPC KVM_REG_PPC_PMC1 32
2314 PPC KVM_REG_PPC_PMC2 32
2315 PPC KVM_REG_PPC_PMC3 32
2316 PPC KVM_REG_PPC_PMC4 32
2317 PPC KVM_REG_PPC_PMC5 32
2318 PPC KVM_REG_PPC_PMC6 32
2319 PPC KVM_REG_PPC_PMC7 32
2320 PPC KVM_REG_PPC_PMC8 32
2321 PPC KVM_REG_PPC_FPR0 64
2323 PPC KVM_REG_PPC_FPR31 64
2324 PPC KVM_REG_PPC_VR0 128
2326 PPC KVM_REG_PPC_VR31 128
2327 PPC KVM_REG_PPC_VSR0 128
2329 PPC KVM_REG_PPC_VSR31 128
2330 PPC KVM_REG_PPC_FPSCR 64
2331 PPC KVM_REG_PPC_VSCR 32
2332 PPC KVM_REG_PPC_VPA_ADDR 64
2333 PPC KVM_REG_PPC_VPA_SLB 128
2334 PPC KVM_REG_PPC_VPA_DTL 128
2335 PPC KVM_REG_PPC_EPCR 32
2336 PPC KVM_REG_PPC_EPR 32
2337 PPC KVM_REG_PPC_TCR 32
2338 PPC KVM_REG_PPC_TSR 32
2339 PPC KVM_REG_PPC_OR_TSR 32
2340 PPC KVM_REG_PPC_CLEAR_TSR 32
2341 PPC KVM_REG_PPC_MAS0 32
2342 PPC KVM_REG_PPC_MAS1 32
2343 PPC KVM_REG_PPC_MAS2 64
2344 PPC KVM_REG_PPC_MAS7_3 64
2345 PPC KVM_REG_PPC_MAS4 32
2346 PPC KVM_REG_PPC_MAS6 32
2347 PPC KVM_REG_PPC_MMUCFG 32
2348 PPC KVM_REG_PPC_TLB0CFG 32
2349 PPC KVM_REG_PPC_TLB1CFG 32
2350 PPC KVM_REG_PPC_TLB2CFG 32
2351 PPC KVM_REG_PPC_TLB3CFG 32
2352 PPC KVM_REG_PPC_TLB0PS 32
2353 PPC KVM_REG_PPC_TLB1PS 32
2354 PPC KVM_REG_PPC_TLB2PS 32
2355 PPC KVM_REG_PPC_TLB3PS 32
2356 PPC KVM_REG_PPC_EPTCFG 32
2357 PPC KVM_REG_PPC_ICP_STATE 64
2358 PPC KVM_REG_PPC_VP_STATE 128
2359 PPC KVM_REG_PPC_TB_OFFSET 64
2360 PPC KVM_REG_PPC_SPMC1 32
2361 PPC KVM_REG_PPC_SPMC2 32
2362 PPC KVM_REG_PPC_IAMR 64
2363 PPC KVM_REG_PPC_TFHAR 64
2364 PPC KVM_REG_PPC_TFIAR 64
2365 PPC KVM_REG_PPC_TEXASR 64
2366 PPC KVM_REG_PPC_FSCR 64
2367 PPC KVM_REG_PPC_PSPB 32
2368 PPC KVM_REG_PPC_EBBHR 64
2369 PPC KVM_REG_PPC_EBBRR 64
2370 PPC KVM_REG_PPC_BESCR 64
2371 PPC KVM_REG_PPC_TAR 64
2372 PPC KVM_REG_PPC_DPDES 64
2373 PPC KVM_REG_PPC_DAWR 64
2374 PPC KVM_REG_PPC_DAWRX 64
2375 PPC KVM_REG_PPC_CIABR 64
2376 PPC KVM_REG_PPC_IC 64
2377 PPC KVM_REG_PPC_VTB 64
2378 PPC KVM_REG_PPC_CSIGR 64
2379 PPC KVM_REG_PPC_TACR 64
2380 PPC KVM_REG_PPC_TCSCR 64
2381 PPC KVM_REG_PPC_PID 64
2382 PPC KVM_REG_PPC_ACOP 64
2383 PPC KVM_REG_PPC_VRSAVE 32
2384 PPC KVM_REG_PPC_LPCR 32
2385 PPC KVM_REG_PPC_LPCR_64 64
2386 PPC KVM_REG_PPC_PPR 64
2387 PPC KVM_REG_PPC_ARCH_COMPAT 32
2388 PPC KVM_REG_PPC_DABRX 32
2389 PPC KVM_REG_PPC_WORT 64
2390 PPC KVM_REG_PPC_SPRG9 64
2391 PPC KVM_REG_PPC_DBSR 32
2392 PPC KVM_REG_PPC_TIDR 64
2393 PPC KVM_REG_PPC_PSSCR 64
2394 PPC KVM_REG_PPC_DEC_EXPIRY 64
2395 PPC KVM_REG_PPC_PTCR 64
2396 PPC KVM_REG_PPC_DAWR1 64
2397 PPC KVM_REG_PPC_DAWRX1 64
2398 PPC KVM_REG_PPC_TM_GPR0 64
2400 PPC KVM_REG_PPC_TM_GPR31 64
2401 PPC KVM_REG_PPC_TM_VSR0 128
2403 PPC KVM_REG_PPC_TM_VSR63 128
2404 PPC KVM_REG_PPC_TM_CR 64
2405 PPC KVM_REG_PPC_TM_LR 64
2406 PPC KVM_REG_PPC_TM_CTR 64
2407 PPC KVM_REG_PPC_TM_FPSCR 64
2408 PPC KVM_REG_PPC_TM_AMR 64
2409 PPC KVM_REG_PPC_TM_PPR 64
2410 PPC KVM_REG_PPC_TM_VRSAVE 64
2411 PPC KVM_REG_PPC_TM_VSCR 32
2412 PPC KVM_REG_PPC_TM_DSCR 64
2413 PPC KVM_REG_PPC_TM_TAR 64
2414 PPC KVM_REG_PPC_TM_XER 64
2416 MIPS KVM_REG_MIPS_R0 64
2418 MIPS KVM_REG_MIPS_R31 64
2419 MIPS KVM_REG_MIPS_HI 64
2420 MIPS KVM_REG_MIPS_LO 64
2421 MIPS KVM_REG_MIPS_PC 64
2422 MIPS KVM_REG_MIPS_CP0_INDEX 32
2423 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2424 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2425 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2426 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2427 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2428 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2429 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2430 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2431 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2432 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2433 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2434 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2435 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2436 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2437 MIPS KVM_REG_MIPS_CP0_WIRED 32
2438 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2439 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2440 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2441 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2442 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2443 MIPS KVM_REG_MIPS_CP0_COUNT 32
2444 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2445 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2446 MIPS KVM_REG_MIPS_CP0_STATUS 32
2447 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2448 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2449 MIPS KVM_REG_MIPS_CP0_EPC 64
2450 MIPS KVM_REG_MIPS_CP0_PRID 32
2451 MIPS KVM_REG_MIPS_CP0_EBASE 64
2452 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2453 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2454 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2455 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2456 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2457 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2458 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2459 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2460 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2461 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2462 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2463 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2464 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2465 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2466 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2467 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2468 MIPS KVM_REG_MIPS_COUNT_CTL 64
2469 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2470 MIPS KVM_REG_MIPS_COUNT_HZ 64
2471 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2472 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2473 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2474 MIPS KVM_REG_MIPS_FCR_IR 32
2475 MIPS KVM_REG_MIPS_FCR_CSR 32
2476 MIPS KVM_REG_MIPS_MSA_IR 32
2477 MIPS KVM_REG_MIPS_MSA_CSR 32
2478 ======= =============================== ============
2480 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2481 is the register group type, or coprocessor number:
2483 ARM core registers have the following id bit patterns::
2485 0x4020 0000 0010 <index into the kvm_regs struct:16>
2487 ARM 32-bit CP15 registers have the following id bit patterns::
2489 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2491 ARM 64-bit CP15 registers have the following id bit patterns::
2493 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2495 ARM CCSIDR registers are demultiplexed by CSSELR value::
2497 0x4020 0000 0011 00 <csselr:8>
2499 ARM 32-bit VFP control registers have the following id bit patterns::
2501 0x4020 0000 0012 1 <regno:12>
2503 ARM 64-bit FP registers have the following id bit patterns::
2505 0x4030 0000 0012 0 <regno:12>
2507 ARM firmware pseudo-registers have the following bit pattern::
2509 0x4030 0000 0014 <regno:16>
2512 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2513 that is the register group type, or coprocessor number:
2515 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2516 that the size of the access is variable, as the kvm_regs structure
2517 contains elements ranging from 32 to 128 bits. The index is a 32bit
2518 value in the kvm_regs structure seen as a 32bit array::
2520 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2524 ======================= ========= ===== =======================================
2525 Encoding Register Bits kvm_regs member
2526 ======================= ========= ===== =======================================
2527 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2528 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2530 0x6030 0000 0010 003c X30 64 regs.regs[30]
2531 0x6030 0000 0010 003e SP 64 regs.sp
2532 0x6030 0000 0010 0040 PC 64 regs.pc
2533 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2534 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2535 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2536 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2537 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2538 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2539 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2540 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2541 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2542 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2544 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2545 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2546 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2547 ======================= ========= ===== =======================================
2549 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2552 The equivalent register content can be accessed via bits [127:0] of
2553 the corresponding SVE Zn registers instead for vcpus that have SVE
2554 enabled (see below).
2556 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2558 0x6020 0000 0011 00 <csselr:8>
2560 arm64 system registers have the following id bit patterns::
2562 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2566 Two system register IDs do not follow the specified pattern. These
2567 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2568 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2569 two had their values accidentally swapped, which means TIMER_CVAL is
2570 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2571 derived from the register encoding for CNTV_CVAL_EL0. As this is
2572 API, it must remain this way.
2574 arm64 firmware pseudo-registers have the following bit pattern::
2576 0x6030 0000 0014 <regno:16>
2578 arm64 SVE registers have the following bit patterns::
2580 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2581 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2582 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2583 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2585 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2586 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2587 quadwords: see [2]_ below.
2589 These registers are only accessible on vcpus for which SVE is enabled.
2590 See KVM_ARM_VCPU_INIT for details.
2592 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2593 accessible until the vcpu's SVE configuration has been finalized
2594 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2595 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2597 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2598 lengths supported by the vcpu to be discovered and configured by
2599 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2600 or KVM_SET_ONE_REG, the value of this register is of type
2601 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2604 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2606 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2607 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2608 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2609 /* Vector length vq * 16 bytes supported */
2611 /* Vector length vq * 16 bytes not supported */
2613 .. [2] The maximum value vq for which the above condition is true is
2614 max_vq. This is the maximum vector length available to the guest on
2615 this vcpu, and determines which register slices are visible through
2616 this ioctl interface.
2618 (See Documentation/arch/arm64/sve.rst for an explanation of the "vq"
2621 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2622 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2625 Userspace may subsequently modify it if desired until the vcpu's SVE
2626 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2628 Apart from simply removing all vector lengths from the host set that
2629 exceed some value, support for arbitrarily chosen sets of vector lengths
2630 is hardware-dependent and may not be available. Attempting to configure
2631 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2634 After the vcpu's SVE configuration is finalized, further attempts to
2635 write this register will fail with EPERM.
2637 arm64 bitmap feature firmware pseudo-registers have the following bit pattern::
2639 0x6030 0000 0016 <regno:16>
2641 The bitmap feature firmware registers exposes the hypercall services that
2642 are available for userspace to configure. The set bits corresponds to the
2643 services that are available for the guests to access. By default, KVM
2644 sets all the supported bits during VM initialization. The userspace can
2645 discover the available services via KVM_GET_ONE_REG, and write back the
2646 bitmap corresponding to the features that it wishes guests to see via
2649 Note: These registers are immutable once any of the vCPUs of the VM has
2650 run at least once. A KVM_SET_ONE_REG in such a scenario will return
2651 a -EBUSY to userspace.
2653 (See Documentation/virt/kvm/arm/hypercalls.rst for more details.)
2656 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2657 the register group type:
2659 MIPS core registers (see above) have the following id bit patterns::
2661 0x7030 0000 0000 <reg:16>
2663 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2664 patterns depending on whether they're 32-bit or 64-bit registers::
2666 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2667 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2669 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2670 versions of the EntryLo registers regardless of the word size of the host
2671 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2672 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2673 the PFNX field starting at bit 30.
2675 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2678 0x7030 0000 0001 01 <reg:8>
2680 MIPS KVM control registers (see above) have the following id bit patterns::
2682 0x7030 0000 0002 <reg:16>
2684 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2685 id bit patterns depending on the size of the register being accessed. They are
2686 always accessed according to the current guest FPU mode (Status.FR and
2687 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2688 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2689 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2690 overlap the FPU registers::
2692 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2693 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2694 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2696 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2697 following id bit patterns::
2699 0x7020 0000 0003 01 <0:3> <reg:5>
2701 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2702 following id bit patterns::
2704 0x7020 0000 0003 02 <0:3> <reg:5>
2706 RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of
2707 that is the register group type.
2709 RISC-V config registers are meant for configuring a Guest VCPU and it has
2710 the following id bit patterns::
2712 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host)
2713 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host)
2715 Following are the RISC-V config registers:
2717 ======================= ========= =============================================
2718 Encoding Register Description
2719 ======================= ========= =============================================
2720 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU
2721 ======================= ========= =============================================
2723 The isa config register can be read anytime but can only be written before
2724 a Guest VCPU runs. It will have ISA feature bits matching underlying host
2727 RISC-V core registers represent the general execution state of a Guest VCPU
2728 and it has the following id bit patterns::
2730 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host)
2731 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host)
2733 Following are the RISC-V core registers:
2735 ======================= ========= =============================================
2736 Encoding Register Description
2737 ======================= ========= =============================================
2738 0x80x0 0000 0200 0000 regs.pc Program counter
2739 0x80x0 0000 0200 0001 regs.ra Return address
2740 0x80x0 0000 0200 0002 regs.sp Stack pointer
2741 0x80x0 0000 0200 0003 regs.gp Global pointer
2742 0x80x0 0000 0200 0004 regs.tp Task pointer
2743 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0
2744 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1
2745 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2
2746 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0
2747 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1
2748 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0
2749 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1
2750 0x80x0 0000 0200 000c regs.a2 Function argument 2
2751 0x80x0 0000 0200 000d regs.a3 Function argument 3
2752 0x80x0 0000 0200 000e regs.a4 Function argument 4
2753 0x80x0 0000 0200 000f regs.a5 Function argument 5
2754 0x80x0 0000 0200 0010 regs.a6 Function argument 6
2755 0x80x0 0000 0200 0011 regs.a7 Function argument 7
2756 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2
2757 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3
2758 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4
2759 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5
2760 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6
2761 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7
2762 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8
2763 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9
2764 0x80x0 0000 0200 001a regs.s10 Callee saved register 10
2765 0x80x0 0000 0200 001b regs.s11 Callee saved register 11
2766 0x80x0 0000 0200 001c regs.t3 Caller saved register 3
2767 0x80x0 0000 0200 001d regs.t4 Caller saved register 4
2768 0x80x0 0000 0200 001e regs.t5 Caller saved register 5
2769 0x80x0 0000 0200 001f regs.t6 Caller saved register 6
2770 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode)
2771 ======================= ========= =============================================
2773 RISC-V csr registers represent the supervisor mode control/status registers
2774 of a Guest VCPU and it has the following id bit patterns::
2776 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host)
2777 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host)
2779 Following are the RISC-V csr registers:
2781 ======================= ========= =============================================
2782 Encoding Register Description
2783 ======================= ========= =============================================
2784 0x80x0 0000 0300 0000 sstatus Supervisor status
2785 0x80x0 0000 0300 0001 sie Supervisor interrupt enable
2786 0x80x0 0000 0300 0002 stvec Supervisor trap vector base
2787 0x80x0 0000 0300 0003 sscratch Supervisor scratch register
2788 0x80x0 0000 0300 0004 sepc Supervisor exception program counter
2789 0x80x0 0000 0300 0005 scause Supervisor trap cause
2790 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction
2791 0x80x0 0000 0300 0007 sip Supervisor interrupt pending
2792 0x80x0 0000 0300 0008 satp Supervisor address translation and protection
2793 ======================= ========= =============================================
2795 RISC-V timer registers represent the timer state of a Guest VCPU and it has
2796 the following id bit patterns::
2798 0x8030 0000 04 <index into the kvm_riscv_timer struct:24>
2800 Following are the RISC-V timer registers:
2802 ======================= ========= =============================================
2803 Encoding Register Description
2804 ======================= ========= =============================================
2805 0x8030 0000 0400 0000 frequency Time base frequency (read-only)
2806 0x8030 0000 0400 0001 time Time value visible to Guest
2807 0x8030 0000 0400 0002 compare Time compare programmed by Guest
2808 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF)
2809 ======================= ========= =============================================
2811 RISC-V F-extension registers represent the single precision floating point
2812 state of a Guest VCPU and it has the following id bit patterns::
2814 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24>
2816 Following are the RISC-V F-extension registers:
2818 ======================= ========= =============================================
2819 Encoding Register Description
2820 ======================= ========= =============================================
2821 0x8020 0000 0500 0000 f[0] Floating point register 0
2823 0x8020 0000 0500 001f f[31] Floating point register 31
2824 0x8020 0000 0500 0020 fcsr Floating point control and status register
2825 ======================= ========= =============================================
2827 RISC-V D-extension registers represent the double precision floating point
2828 state of a Guest VCPU and it has the following id bit patterns::
2830 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr)
2831 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr)
2833 Following are the RISC-V D-extension registers:
2835 ======================= ========= =============================================
2836 Encoding Register Description
2837 ======================= ========= =============================================
2838 0x8030 0000 0600 0000 f[0] Floating point register 0
2840 0x8030 0000 0600 001f f[31] Floating point register 31
2841 0x8020 0000 0600 0020 fcsr Floating point control and status register
2842 ======================= ========= =============================================
2845 4.69 KVM_GET_ONE_REG
2846 --------------------
2848 :Capability: KVM_CAP_ONE_REG
2851 :Parameters: struct kvm_one_reg (in and out)
2852 :Returns: 0 on success, negative value on failure
2856 ======== ============================================================
2857 ENOENT no such register
2858 EINVAL invalid register ID, or no such register or used with VMs in
2859 protected virtualization mode on s390
2860 EPERM (arm64) register access not allowed before vcpu finalization
2861 ======== ============================================================
2863 (These error codes are indicative only: do not rely on a specific error
2864 code being returned in a specific situation.)
2866 This ioctl allows to receive the value of a single register implemented
2867 in a vcpu. The register to read is indicated by the "id" field of the
2868 kvm_one_reg struct passed in. On success, the register value can be found
2869 at the memory location pointed to by "addr".
2871 The list of registers accessible using this interface is identical to the
2875 4.70 KVM_KVMCLOCK_CTRL
2876 ----------------------
2878 :Capability: KVM_CAP_KVMCLOCK_CTRL
2879 :Architectures: Any that implement pvclocks (currently x86 only)
2882 :Returns: 0 on success, -1 on error
2884 This ioctl sets a flag accessible to the guest indicating that the specified
2885 vCPU has been paused by the host userspace.
2887 The host will set a flag in the pvclock structure that is checked from the
2888 soft lockup watchdog. The flag is part of the pvclock structure that is
2889 shared between guest and host, specifically the second bit of the flags
2890 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2891 the host and read/cleared exclusively by the guest. The guest operation of
2892 checking and clearing the flag must be an atomic operation so
2893 load-link/store-conditional, or equivalent must be used. There are two cases
2894 where the guest will clear the flag: when the soft lockup watchdog timer resets
2895 itself or when a soft lockup is detected. This ioctl can be called any time
2896 after pausing the vcpu, but before it is resumed.
2902 :Capability: KVM_CAP_SIGNAL_MSI
2903 :Architectures: x86 arm64
2905 :Parameters: struct kvm_msi (in)
2906 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2908 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2923 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2924 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2925 the device ID. If this capability is not available, userspace
2926 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2928 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2929 for the device that wrote the MSI message. For PCI, this is usually a
2930 BFD identifier in the lower 16 bits.
2932 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2933 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2934 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2935 address_hi must be zero.
2938 4.71 KVM_CREATE_PIT2
2939 --------------------
2941 :Capability: KVM_CAP_PIT2
2944 :Parameters: struct kvm_pit_config (in)
2945 :Returns: 0 on success, -1 on error
2947 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2948 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2949 parameters have to be passed::
2951 struct kvm_pit_config {
2958 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2960 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2961 exists, this thread will have a name of the following pattern::
2963 kvm-pit/<owner-process-pid>
2965 When running a guest with elevated priorities, the scheduling parameters of
2966 this thread may have to be adjusted accordingly.
2968 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2974 :Capability: KVM_CAP_PIT_STATE2
2977 :Parameters: struct kvm_pit_state2 (out)
2978 :Returns: 0 on success, -1 on error
2980 Retrieves the state of the in-kernel PIT model. Only valid after
2981 KVM_CREATE_PIT2. The state is returned in the following structure::
2983 struct kvm_pit_state2 {
2984 struct kvm_pit_channel_state channels[3];
2991 /* disable PIT in HPET legacy mode */
2992 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2993 /* speaker port data bit enabled */
2994 #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002
2996 This IOCTL replaces the obsolete KVM_GET_PIT.
3002 :Capability: KVM_CAP_PIT_STATE2
3005 :Parameters: struct kvm_pit_state2 (in)
3006 :Returns: 0 on success, -1 on error
3008 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
3009 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
3011 This IOCTL replaces the obsolete KVM_SET_PIT.
3014 4.74 KVM_PPC_GET_SMMU_INFO
3015 --------------------------
3017 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
3018 :Architectures: powerpc
3021 :Returns: 0 on success, -1 on error
3023 This populates and returns a structure describing the features of
3024 the "Server" class MMU emulation supported by KVM.
3025 This can in turn be used by userspace to generate the appropriate
3026 device-tree properties for the guest operating system.
3028 The structure contains some global information, followed by an
3029 array of supported segment page sizes::
3031 struct kvm_ppc_smmu_info {
3035 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
3038 The supported flags are:
3040 - KVM_PPC_PAGE_SIZES_REAL:
3041 When that flag is set, guest page sizes must "fit" the backing
3042 store page sizes. When not set, any page size in the list can
3043 be used regardless of how they are backed by userspace.
3045 - KVM_PPC_1T_SEGMENTS
3046 The emulated MMU supports 1T segments in addition to the
3050 This flag indicates that HPT guests are not supported by KVM,
3051 thus all guests must use radix MMU mode.
3053 The "slb_size" field indicates how many SLB entries are supported
3055 The "sps" array contains 8 entries indicating the supported base
3056 page sizes for a segment in increasing order. Each entry is defined
3059 struct kvm_ppc_one_seg_page_size {
3060 __u32 page_shift; /* Base page shift of segment (or 0) */
3061 __u32 slb_enc; /* SLB encoding for BookS */
3062 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3065 An entry with a "page_shift" of 0 is unused. Because the array is
3066 organized in increasing order, a lookup can stop when encoutering
3069 The "slb_enc" field provides the encoding to use in the SLB for the
3070 page size. The bits are in positions such as the value can directly
3071 be OR'ed into the "vsid" argument of the slbmte instruction.
3073 The "enc" array is a list which for each of those segment base page
3074 size provides the list of supported actual page sizes (which can be
3075 only larger or equal to the base page size), along with the
3076 corresponding encoding in the hash PTE. Similarly, the array is
3077 8 entries sorted by increasing sizes and an entry with a "0" shift
3078 is an empty entry and a terminator::
3080 struct kvm_ppc_one_page_size {
3081 __u32 page_shift; /* Page shift (or 0) */
3082 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
3085 The "pte_enc" field provides a value that can OR'ed into the hash
3086 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3087 into the hash PTE second double word).
3092 :Capability: KVM_CAP_IRQFD
3093 :Architectures: x86 s390 arm64
3095 :Parameters: struct kvm_irqfd (in)
3096 :Returns: 0 on success, -1 on error
3098 Allows setting an eventfd to directly trigger a guest interrupt.
3099 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3100 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
3101 an event is triggered on the eventfd, an interrupt is injected into
3102 the guest using the specified gsi pin. The irqfd is removed using
3103 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3106 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3107 mechanism allowing emulation of level-triggered, irqfd-based
3108 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3109 additional eventfd in the kvm_irqfd.resamplefd field. When operating
3110 in resample mode, posting of an interrupt through kvm_irq.fd asserts
3111 the specified gsi in the irqchip. When the irqchip is resampled, such
3112 as from an EOI, the gsi is de-asserted and the user is notified via
3113 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
3114 the interrupt if the device making use of it still requires service.
3115 Note that closing the resamplefd is not sufficient to disable the
3116 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3117 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3119 On arm64, gsi routing being supported, the following can happen:
3121 - in case no routing entry is associated to this gsi, injection fails
3122 - in case the gsi is associated to an irqchip routing entry,
3123 irqchip.pin + 32 corresponds to the injected SPI ID.
3124 - in case the gsi is associated to an MSI routing entry, the MSI
3125 message and device ID are translated into an LPI (support restricted
3126 to GICv3 ITS in-kernel emulation).
3128 4.76 KVM_PPC_ALLOCATE_HTAB
3129 --------------------------
3131 :Capability: KVM_CAP_PPC_ALLOC_HTAB
3132 :Architectures: powerpc
3134 :Parameters: Pointer to u32 containing hash table order (in/out)
3135 :Returns: 0 on success, -1 on error
3137 This requests the host kernel to allocate an MMU hash table for a
3138 guest using the PAPR paravirtualization interface. This only does
3139 anything if the kernel is configured to use the Book 3S HV style of
3140 virtualization. Otherwise the capability doesn't exist and the ioctl
3141 returns an ENOTTY error. The rest of this description assumes Book 3S
3144 There must be no vcpus running when this ioctl is called; if there
3145 are, it will do nothing and return an EBUSY error.
3147 The parameter is a pointer to a 32-bit unsigned integer variable
3148 containing the order (log base 2) of the desired size of the hash
3149 table, which must be between 18 and 46. On successful return from the
3150 ioctl, the value will not be changed by the kernel.
3152 If no hash table has been allocated when any vcpu is asked to run
3153 (with the KVM_RUN ioctl), the host kernel will allocate a
3154 default-sized hash table (16 MB).
3156 If this ioctl is called when a hash table has already been allocated,
3157 with a different order from the existing hash table, the existing hash
3158 table will be freed and a new one allocated. If this is ioctl is
3159 called when a hash table has already been allocated of the same order
3160 as specified, the kernel will clear out the existing hash table (zero
3161 all HPTEs). In either case, if the guest is using the virtualized
3162 real-mode area (VRMA) facility, the kernel will re-create the VMRA
3163 HPTEs on the next KVM_RUN of any vcpu.
3165 4.77 KVM_S390_INTERRUPT
3166 -----------------------
3169 :Architectures: s390
3170 :Type: vm ioctl, vcpu ioctl
3171 :Parameters: struct kvm_s390_interrupt (in)
3172 :Returns: 0 on success, -1 on error
3174 Allows to inject an interrupt to the guest. Interrupts can be floating
3175 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3177 Interrupt parameters are passed via kvm_s390_interrupt::
3179 struct kvm_s390_interrupt {
3185 type can be one of the following:
3187 KVM_S390_SIGP_STOP (vcpu)
3188 - sigp stop; optional flags in parm
3189 KVM_S390_PROGRAM_INT (vcpu)
3190 - program check; code in parm
3191 KVM_S390_SIGP_SET_PREFIX (vcpu)
3192 - sigp set prefix; prefix address in parm
3193 KVM_S390_RESTART (vcpu)
3195 KVM_S390_INT_CLOCK_COMP (vcpu)
3196 - clock comparator interrupt
3197 KVM_S390_INT_CPU_TIMER (vcpu)
3198 - CPU timer interrupt
3199 KVM_S390_INT_VIRTIO (vm)
3200 - virtio external interrupt; external interrupt
3201 parameters in parm and parm64
3202 KVM_S390_INT_SERVICE (vm)
3203 - sclp external interrupt; sclp parameter in parm
3204 KVM_S390_INT_EMERGENCY (vcpu)
3205 - sigp emergency; source cpu in parm
3206 KVM_S390_INT_EXTERNAL_CALL (vcpu)
3207 - sigp external call; source cpu in parm
3208 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3209 - compound value to indicate an
3210 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3211 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3212 interruption subclass)
3213 KVM_S390_MCHK (vm, vcpu)
3214 - machine check interrupt; cr 14 bits in parm, machine check interrupt
3215 code in parm64 (note that machine checks needing further payload are not
3216 supported by this ioctl)
3218 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3220 4.78 KVM_PPC_GET_HTAB_FD
3221 ------------------------
3223 :Capability: KVM_CAP_PPC_HTAB_FD
3224 :Architectures: powerpc
3226 :Parameters: Pointer to struct kvm_get_htab_fd (in)
3227 :Returns: file descriptor number (>= 0) on success, -1 on error
3229 This returns a file descriptor that can be used either to read out the
3230 entries in the guest's hashed page table (HPT), or to write entries to
3231 initialize the HPT. The returned fd can only be written to if the
3232 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3233 can only be read if that bit is clear. The argument struct looks like
3236 /* For KVM_PPC_GET_HTAB_FD */
3237 struct kvm_get_htab_fd {
3243 /* Values for kvm_get_htab_fd.flags */
3244 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
3245 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
3247 The 'start_index' field gives the index in the HPT of the entry at
3248 which to start reading. It is ignored when writing.
3250 Reads on the fd will initially supply information about all
3251 "interesting" HPT entries. Interesting entries are those with the
3252 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3253 all entries. When the end of the HPT is reached, the read() will
3254 return. If read() is called again on the fd, it will start again from
3255 the beginning of the HPT, but will only return HPT entries that have
3256 changed since they were last read.
3258 Data read or written is structured as a header (8 bytes) followed by a
3259 series of valid HPT entries (16 bytes) each. The header indicates how
3260 many valid HPT entries there are and how many invalid entries follow
3261 the valid entries. The invalid entries are not represented explicitly
3262 in the stream. The header format is::
3264 struct kvm_get_htab_header {
3270 Writes to the fd create HPT entries starting at the index given in the
3271 header; first 'n_valid' valid entries with contents from the data
3272 written, then 'n_invalid' invalid entries, invalidating any previously
3273 valid entries found.
3275 4.79 KVM_CREATE_DEVICE
3276 ----------------------
3278 :Capability: KVM_CAP_DEVICE_CTRL
3281 :Parameters: struct kvm_create_device (in/out)
3282 :Returns: 0 on success, -1 on error
3286 ====== =======================================================
3287 ENODEV The device type is unknown or unsupported
3288 EEXIST Device already created, and this type of device may not
3289 be instantiated multiple times
3290 ====== =======================================================
3292 Other error conditions may be defined by individual device types or
3293 have their standard meanings.
3295 Creates an emulated device in the kernel. The file descriptor returned
3296 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3298 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3299 device type is supported (not necessarily whether it can be created
3302 Individual devices should not define flags. Attributes should be used
3303 for specifying any behavior that is not implied by the device type
3308 struct kvm_create_device {
3309 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3310 __u32 fd; /* out: device handle */
3311 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3314 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3315 --------------------------------------------
3317 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3318 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3319 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set)
3320 :Architectures: x86, arm64, s390
3321 :Type: device ioctl, vm ioctl, vcpu ioctl
3322 :Parameters: struct kvm_device_attr
3323 :Returns: 0 on success, -1 on error
3327 ===== =============================================================
3328 ENXIO The group or attribute is unknown/unsupported for this device
3329 or hardware support is missing.
3330 EPERM The attribute cannot (currently) be accessed this way
3331 (e.g. read-only attribute, or attribute that only makes
3332 sense when the device is in a different state)
3333 ===== =============================================================
3335 Other error conditions may be defined by individual device types.
3337 Gets/sets a specified piece of device configuration and/or state. The
3338 semantics are device-specific. See individual device documentation in
3339 the "devices" directory. As with ONE_REG, the size of the data
3340 transferred is defined by the particular attribute.
3344 struct kvm_device_attr {
3345 __u32 flags; /* no flags currently defined */
3346 __u32 group; /* device-defined */
3347 __u64 attr; /* group-defined */
3348 __u64 addr; /* userspace address of attr data */
3351 4.81 KVM_HAS_DEVICE_ATTR
3352 ------------------------
3354 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3355 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3356 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device
3357 :Type: device ioctl, vm ioctl, vcpu ioctl
3358 :Parameters: struct kvm_device_attr
3359 :Returns: 0 on success, -1 on error
3363 ===== =============================================================
3364 ENXIO The group or attribute is unknown/unsupported for this device
3365 or hardware support is missing.
3366 ===== =============================================================
3368 Tests whether a device supports a particular attribute. A successful
3369 return indicates the attribute is implemented. It does not necessarily
3370 indicate that the attribute can be read or written in the device's
3371 current state. "addr" is ignored.
3373 4.82 KVM_ARM_VCPU_INIT
3374 ----------------------
3377 :Architectures: arm64
3379 :Parameters: struct kvm_vcpu_init (in)
3380 :Returns: 0 on success; -1 on error
3384 ====== =================================================================
3385 EINVAL the target is unknown, or the combination of features is invalid.
3386 ENOENT a features bit specified is unknown.
3387 ====== =================================================================
3389 This tells KVM what type of CPU to present to the guest, and what
3390 optional features it should have. This will cause a reset of the cpu
3391 registers to their initial values. If this is not called, KVM_RUN will
3392 return ENOEXEC for that vcpu.
3394 The initial values are defined as:
3396 * AArch64: EL1h, D, A, I and F bits set. All other bits
3398 * AArch32: SVC, A, I and F bits set. All other bits are
3400 - General Purpose registers, including PC and SP: set to 0
3401 - FPSIMD/NEON registers: set to 0
3402 - SVE registers: set to 0
3403 - System registers: Reset to their architecturally defined
3404 values as for a warm reset to EL1 (resp. SVC)
3406 Note that because some registers reflect machine topology, all vcpus
3407 should be created before this ioctl is invoked.
3409 Userspace can call this function multiple times for a given vcpu, including
3410 after the vcpu has been run. This will reset the vcpu to its initial
3411 state. All calls to this function after the initial call must use the same
3412 target and same set of feature flags, otherwise EINVAL will be returned.
3416 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3417 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3418 and execute guest code when KVM_RUN is called.
3419 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3420 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3421 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3422 backward compatible with v0.2) for the CPU.
3423 Depends on KVM_CAP_ARM_PSCI_0_2.
3424 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3425 Depends on KVM_CAP_ARM_PMU_V3.
3427 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3429 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3430 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3431 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3432 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3435 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3437 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3438 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3439 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3440 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3443 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3444 Depends on KVM_CAP_ARM_SVE.
3445 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3447 * After KVM_ARM_VCPU_INIT:
3449 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3450 initial value of this pseudo-register indicates the best set of
3451 vector lengths possible for a vcpu on this host.
3453 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3455 - KVM_RUN and KVM_GET_REG_LIST are not available;
3457 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3458 the scalable archietctural SVE registers
3459 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3460 KVM_REG_ARM64_SVE_FFR;
3462 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3463 KVM_SET_ONE_REG, to modify the set of vector lengths available
3466 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3468 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3469 no longer be written using KVM_SET_ONE_REG.
3471 4.83 KVM_ARM_PREFERRED_TARGET
3472 -----------------------------
3475 :Architectures: arm64
3477 :Parameters: struct kvm_vcpu_init (out)
3478 :Returns: 0 on success; -1 on error
3482 ====== ==========================================
3483 ENODEV no preferred target available for the host
3484 ====== ==========================================
3486 This queries KVM for preferred CPU target type which can be emulated
3487 by KVM on underlying host.
3489 The ioctl returns struct kvm_vcpu_init instance containing information
3490 about preferred CPU target type and recommended features for it. The
3491 kvm_vcpu_init->features bitmap returned will have feature bits set if
3492 the preferred target recommends setting these features, but this is
3495 The information returned by this ioctl can be used to prepare an instance
3496 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3497 VCPU matching underlying host.
3500 4.84 KVM_GET_REG_LIST
3501 ---------------------
3504 :Architectures: arm64, mips, riscv
3506 :Parameters: struct kvm_reg_list (in/out)
3507 :Returns: 0 on success; -1 on error
3511 ===== ==============================================================
3512 E2BIG the reg index list is too big to fit in the array specified by
3513 the user (the number required will be written into n).
3514 ===== ==============================================================
3518 struct kvm_reg_list {
3519 __u64 n; /* number of registers in reg[] */
3523 This ioctl returns the guest registers that are supported for the
3524 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3527 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3528 -----------------------------------------
3530 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3531 :Architectures: arm64
3533 :Parameters: struct kvm_arm_device_address (in)
3534 :Returns: 0 on success, -1 on error
3538 ====== ============================================
3539 ENODEV The device id is unknown
3540 ENXIO Device not supported on current system
3541 EEXIST Address already set
3542 E2BIG Address outside guest physical address space
3543 EBUSY Address overlaps with other device range
3544 ====== ============================================
3548 struct kvm_arm_device_addr {
3553 Specify a device address in the guest's physical address space where guests
3554 can access emulated or directly exposed devices, which the host kernel needs
3555 to know about. The id field is an architecture specific identifier for a
3558 arm64 divides the id field into two parts, a device id and an
3559 address type id specific to the individual device::
3561 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3562 field: | 0x00000000 | device id | addr type id |
3564 arm64 currently only require this when using the in-kernel GIC
3565 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3566 as the device id. When setting the base address for the guest's
3567 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3568 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3569 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3570 base addresses will return -EEXIST.
3572 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3573 should be used instead.
3576 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3577 ------------------------------
3579 :Capability: KVM_CAP_PPC_RTAS
3582 :Parameters: struct kvm_rtas_token_args
3583 :Returns: 0 on success, -1 on error
3585 Defines a token value for a RTAS (Run Time Abstraction Services)
3586 service in order to allow it to be handled in the kernel. The
3587 argument struct gives the name of the service, which must be the name
3588 of a service that has a kernel-side implementation. If the token
3589 value is non-zero, it will be associated with that service, and
3590 subsequent RTAS calls by the guest specifying that token will be
3591 handled by the kernel. If the token value is 0, then any token
3592 associated with the service will be forgotten, and subsequent RTAS
3593 calls by the guest for that service will be passed to userspace to be
3596 4.87 KVM_SET_GUEST_DEBUG
3597 ------------------------
3599 :Capability: KVM_CAP_SET_GUEST_DEBUG
3600 :Architectures: x86, s390, ppc, arm64
3602 :Parameters: struct kvm_guest_debug (in)
3603 :Returns: 0 on success; -1 on error
3607 struct kvm_guest_debug {
3610 struct kvm_guest_debug_arch arch;
3613 Set up the processor specific debug registers and configure vcpu for
3614 handling guest debug events. There are two parts to the structure, the
3615 first a control bitfield indicates the type of debug events to handle
3616 when running. Common control bits are:
3618 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3619 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3621 The top 16 bits of the control field are architecture specific control
3622 flags which can include the following:
3624 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3625 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390]
3626 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64]
3627 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3628 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3629 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3630 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86]
3632 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3633 are enabled in memory so we need to ensure breakpoint exceptions are
3634 correctly trapped and the KVM run loop exits at the breakpoint and not
3635 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3636 we need to ensure the guest vCPUs architecture specific registers are
3637 updated to the correct (supplied) values.
3639 The second part of the structure is architecture specific and
3640 typically contains a set of debug registers.
3642 For arm64 the number of debug registers is implementation defined and
3643 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3644 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3645 indicating the number of supported registers.
3647 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3648 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3650 Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3651 supported KVM_GUESTDBG_* bits in the control field.
3653 When debug events exit the main run loop with the reason
3654 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3655 structure containing architecture specific debug information.
3657 4.88 KVM_GET_EMULATED_CPUID
3658 ---------------------------
3660 :Capability: KVM_CAP_EXT_EMUL_CPUID
3663 :Parameters: struct kvm_cpuid2 (in/out)
3664 :Returns: 0 on success, -1 on error
3671 struct kvm_cpuid_entry2 entries[0];
3674 The member 'flags' is used for passing flags from userspace.
3678 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3679 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3680 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3682 struct kvm_cpuid_entry2 {
3693 This ioctl returns x86 cpuid features which are emulated by
3694 kvm.Userspace can use the information returned by this ioctl to query
3695 which features are emulated by kvm instead of being present natively.
3697 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3698 structure with the 'nent' field indicating the number of entries in
3699 the variable-size array 'entries'. If the number of entries is too low
3700 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3701 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3702 is returned. If the number is just right, the 'nent' field is adjusted
3703 to the number of valid entries in the 'entries' array, which is then
3706 The entries returned are the set CPUID bits of the respective features
3707 which kvm emulates, as returned by the CPUID instruction, with unknown
3708 or unsupported feature bits cleared.
3710 Features like x2apic, for example, may not be present in the host cpu
3711 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3712 emulated efficiently and thus not included here.
3714 The fields in each entry are defined as follows:
3717 the eax value used to obtain the entry
3719 the ecx value used to obtain the entry (for entries that are
3722 an OR of zero or more of the following:
3724 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3725 if the index field is valid
3729 the values returned by the cpuid instruction for
3730 this function/index combination
3732 4.89 KVM_S390_MEM_OP
3733 --------------------
3735 :Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION
3736 :Architectures: s390
3737 :Type: vm ioctl, vcpu ioctl
3738 :Parameters: struct kvm_s390_mem_op (in)
3739 :Returns: = 0 on success,
3740 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3741 16 bit program exception code if the access causes such an exception
3743 Read or write data from/to the VM's memory.
3744 The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is
3747 Parameters are specified via the following structure::
3749 struct kvm_s390_mem_op {
3750 __u64 gaddr; /* the guest address */
3751 __u64 flags; /* flags */
3752 __u32 size; /* amount of bytes */
3753 __u32 op; /* type of operation */
3754 __u64 buf; /* buffer in userspace */
3757 __u8 ar; /* the access register number */
3758 __u8 key; /* access key, ignored if flag unset */
3759 __u8 pad1[6]; /* ignored */
3760 __u64 old_addr; /* ignored if flag unset */
3762 __u32 sida_offset; /* offset into the sida */
3763 __u8 reserved[32]; /* ignored */
3767 The start address of the memory region has to be specified in the "gaddr"
3768 field, and the length of the region in the "size" field (which must not
3769 be 0). The maximum value for "size" can be obtained by checking the
3770 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3771 userspace application where the read data should be written to for
3772 a read access, or where the data that should be written is stored for
3773 a write access. The "reserved" field is meant for future extensions.
3774 Reserved and unused values are ignored. Future extension that add members must
3775 introduce new flags.
3777 The type of operation is specified in the "op" field. Flags modifying
3778 their behavior can be set in the "flags" field. Undefined flag bits must
3781 Possible operations are:
3782 * ``KVM_S390_MEMOP_LOGICAL_READ``
3783 * ``KVM_S390_MEMOP_LOGICAL_WRITE``
3784 * ``KVM_S390_MEMOP_ABSOLUTE_READ``
3785 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE``
3786 * ``KVM_S390_MEMOP_SIDA_READ``
3787 * ``KVM_S390_MEMOP_SIDA_WRITE``
3788 * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG``
3793 Access logical memory, i.e. translate the given guest address to an absolute
3794 address given the state of the VCPU and use the absolute address as target of
3795 the access. "ar" designates the access register number to be used; the valid
3797 Logical accesses are permitted for the VCPU ioctl only.
3798 Logical accesses are permitted for non-protected guests only.
3801 * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3802 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION``
3803 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3805 The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the
3806 corresponding memory access would cause an access exception; however,
3807 no actual access to the data in memory at the destination is performed.
3808 In this case, "buf" is unused and can be NULL.
3810 In case an access exception occurred during the access (or would occur
3811 in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive
3812 error number indicating the type of exception. This exception is also
3813 raised directly at the corresponding VCPU if the flag
3814 KVM_S390_MEMOP_F_INJECT_EXCEPTION is set.
3815 On protection exceptions, unless specified otherwise, the injected
3816 translation-exception identifier (TEID) indicates suppression.
3818 If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key
3819 protection is also in effect and may cause exceptions if accesses are
3820 prohibited given the access key designated by "key"; the valid range is 0..15.
3821 KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION
3823 Since the accessed memory may span multiple pages and those pages might have
3824 different storage keys, it is possible that a protection exception occurs
3825 after memory has been modified. In this case, if the exception is injected,
3826 the TEID does not indicate suppression.
3828 Absolute read/write:
3829 ^^^^^^^^^^^^^^^^^^^^
3831 Access absolute memory. This operation is intended to be used with the
3832 KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing
3833 the checks required for storage key protection as one operation (as opposed to
3834 user space getting the storage keys, performing the checks, and accessing
3835 memory thereafter, which could lead to a delay between check and access).
3836 Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION
3837 has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set.
3838 Currently absolute accesses are not permitted for VCPU ioctls.
3839 Absolute accesses are permitted for non-protected guests only.
3842 * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3843 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3845 The semantics of the flags common with logical accesses are as for logical
3851 Perform cmpxchg on absolute guest memory. Intended for use with the
3852 KVM_S390_MEMOP_F_SKEY_PROTECTION flag.
3853 Instead of doing an unconditional write, the access occurs only if the target
3854 location contains the value pointed to by "old_addr".
3855 This is performed as an atomic cmpxchg with the length specified by the "size"
3856 parameter. "size" must be a power of two up to and including 16.
3857 If the exchange did not take place because the target value doesn't match the
3858 old value, the value "old_addr" points to is replaced by the target value.
3859 User space can tell if an exchange took place by checking if this replacement
3860 occurred. The cmpxchg op is permitted for the VM ioctl if
3861 KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set.
3864 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3869 Access the secure instruction data area which contains memory operands necessary
3870 for instruction emulation for protected guests.
3871 SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available.
3872 SIDA accesses are permitted for the VCPU ioctl only.
3873 SIDA accesses are permitted for protected guests only.
3875 No flags are supported.
3877 4.90 KVM_S390_GET_SKEYS
3878 -----------------------
3880 :Capability: KVM_CAP_S390_SKEYS
3881 :Architectures: s390
3883 :Parameters: struct kvm_s390_skeys
3884 :Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage
3885 keys, negative value on error
3887 This ioctl is used to get guest storage key values on the s390
3888 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3890 struct kvm_s390_skeys {
3893 __u64 skeydata_addr;
3898 The start_gfn field is the number of the first guest frame whose storage keys
3901 The count field is the number of consecutive frames (starting from start_gfn)
3902 whose storage keys to get. The count field must be at least 1 and the maximum
3903 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3904 will cause the ioctl to return -EINVAL.
3906 The skeydata_addr field is the address to a buffer large enough to hold count
3907 bytes. This buffer will be filled with storage key data by the ioctl.
3909 4.91 KVM_S390_SET_SKEYS
3910 -----------------------
3912 :Capability: KVM_CAP_S390_SKEYS
3913 :Architectures: s390
3915 :Parameters: struct kvm_s390_skeys
3916 :Returns: 0 on success, negative value on error
3918 This ioctl is used to set guest storage key values on the s390
3919 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3920 See section on KVM_S390_GET_SKEYS for struct definition.
3922 The start_gfn field is the number of the first guest frame whose storage keys
3925 The count field is the number of consecutive frames (starting from start_gfn)
3926 whose storage keys to get. The count field must be at least 1 and the maximum
3927 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3928 will cause the ioctl to return -EINVAL.
3930 The skeydata_addr field is the address to a buffer containing count bytes of
3931 storage keys. Each byte in the buffer will be set as the storage key for a
3932 single frame starting at start_gfn for count frames.
3934 Note: If any architecturally invalid key value is found in the given data then
3935 the ioctl will return -EINVAL.
3940 :Capability: KVM_CAP_S390_INJECT_IRQ
3941 :Architectures: s390
3943 :Parameters: struct kvm_s390_irq (in)
3944 :Returns: 0 on success, -1 on error
3949 ====== =================================================================
3950 EINVAL interrupt type is invalid
3951 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3952 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3953 than the maximum of VCPUs
3954 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3955 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3956 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3958 ====== =================================================================
3960 Allows to inject an interrupt to the guest.
3962 Using struct kvm_s390_irq as a parameter allows
3963 to inject additional payload which is not
3964 possible via KVM_S390_INTERRUPT.
3966 Interrupt parameters are passed via kvm_s390_irq::
3968 struct kvm_s390_irq {
3971 struct kvm_s390_io_info io;
3972 struct kvm_s390_ext_info ext;
3973 struct kvm_s390_pgm_info pgm;
3974 struct kvm_s390_emerg_info emerg;
3975 struct kvm_s390_extcall_info extcall;
3976 struct kvm_s390_prefix_info prefix;
3977 struct kvm_s390_stop_info stop;
3978 struct kvm_s390_mchk_info mchk;
3983 type can be one of the following:
3985 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3986 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3987 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3988 - KVM_S390_RESTART - restart; no parameters
3989 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3990 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3991 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3992 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3993 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3995 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3997 4.94 KVM_S390_GET_IRQ_STATE
3998 ---------------------------
4000 :Capability: KVM_CAP_S390_IRQ_STATE
4001 :Architectures: s390
4003 :Parameters: struct kvm_s390_irq_state (out)
4004 :Returns: >= number of bytes copied into buffer,
4005 -EINVAL if buffer size is 0,
4006 -ENOBUFS if buffer size is too small to fit all pending interrupts,
4007 -EFAULT if the buffer address was invalid
4009 This ioctl allows userspace to retrieve the complete state of all currently
4010 pending interrupts in a single buffer. Use cases include migration
4011 and introspection. The parameter structure contains the address of a
4012 userspace buffer and its length::
4014 struct kvm_s390_irq_state {
4016 __u32 flags; /* will stay unused for compatibility reasons */
4018 __u32 reserved[4]; /* will stay unused for compatibility reasons */
4021 Userspace passes in the above struct and for each pending interrupt a
4022 struct kvm_s390_irq is copied to the provided buffer.
4024 The structure contains a flags and a reserved field for future extensions. As
4025 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
4026 reserved, these fields can not be used in the future without breaking
4029 If -ENOBUFS is returned the buffer provided was too small and userspace
4030 may retry with a bigger buffer.
4032 4.95 KVM_S390_SET_IRQ_STATE
4033 ---------------------------
4035 :Capability: KVM_CAP_S390_IRQ_STATE
4036 :Architectures: s390
4038 :Parameters: struct kvm_s390_irq_state (in)
4039 :Returns: 0 on success,
4040 -EFAULT if the buffer address was invalid,
4041 -EINVAL for an invalid buffer length (see below),
4042 -EBUSY if there were already interrupts pending,
4043 errors occurring when actually injecting the
4044 interrupt. See KVM_S390_IRQ.
4046 This ioctl allows userspace to set the complete state of all cpu-local
4047 interrupts currently pending for the vcpu. It is intended for restoring
4048 interrupt state after a migration. The input parameter is a userspace buffer
4049 containing a struct kvm_s390_irq_state::
4051 struct kvm_s390_irq_state {
4053 __u32 flags; /* will stay unused for compatibility reasons */
4055 __u32 reserved[4]; /* will stay unused for compatibility reasons */
4058 The restrictions for flags and reserved apply as well.
4059 (see KVM_S390_GET_IRQ_STATE)
4061 The userspace memory referenced by buf contains a struct kvm_s390_irq
4062 for each interrupt to be injected into the guest.
4063 If one of the interrupts could not be injected for some reason the
4066 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
4067 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
4068 which is the maximum number of possibly pending cpu-local interrupts.
4073 :Capability: KVM_CAP_X86_SMM
4077 :Returns: 0 on success, -1 on error
4079 Queues an SMI on the thread's vcpu.
4081 4.97 KVM_X86_SET_MSR_FILTER
4082 ----------------------------
4084 :Capability: KVM_CAP_X86_MSR_FILTER
4087 :Parameters: struct kvm_msr_filter
4088 :Returns: 0 on success, < 0 on error
4092 struct kvm_msr_filter_range {
4093 #define KVM_MSR_FILTER_READ (1 << 0)
4094 #define KVM_MSR_FILTER_WRITE (1 << 1)
4096 __u32 nmsrs; /* number of msrs in bitmap */
4097 __u32 base; /* MSR index the bitmap starts at */
4098 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4101 #define KVM_MSR_FILTER_MAX_RANGES 16
4102 struct kvm_msr_filter {
4103 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4104 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4106 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4109 flags values for ``struct kvm_msr_filter_range``:
4111 ``KVM_MSR_FILTER_READ``
4113 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4114 indicates that read accesses should be denied, while a 1 indicates that
4115 a read for a particular MSR should be allowed regardless of the default
4118 ``KVM_MSR_FILTER_WRITE``
4120 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4121 indicates that write accesses should be denied, while a 1 indicates that
4122 a write for a particular MSR should be allowed regardless of the default
4125 flags values for ``struct kvm_msr_filter``:
4127 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4129 If no filter range matches an MSR index that is getting accessed, KVM will
4130 allow accesses to all MSRs by default.
4132 ``KVM_MSR_FILTER_DEFAULT_DENY``
4134 If no filter range matches an MSR index that is getting accessed, KVM will
4135 deny accesses to all MSRs by default.
4137 This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny
4138 guest MSR accesses that would normally be allowed by KVM. If an MSR is not
4139 covered by a specific range, the "default" filtering behavior applies. Each
4140 bitmap range covers MSRs from [base .. base+nmsrs).
4142 If an MSR access is denied by userspace, the resulting KVM behavior depends on
4143 whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is
4144 enabled. If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace
4145 on denied accesses, i.e. userspace effectively intercepts the MSR access. If
4146 KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest
4149 If an MSR access is allowed by userspace, KVM will emulate and/or virtualize
4150 the access in accordance with the vCPU model. Note, KVM may still ultimately
4151 inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support
4152 the MSR, or to follow architectural behavior for the MSR.
4154 By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range
4157 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4158 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4162 MSR accesses as part of nested VM-Enter/VM-Exit are not filtered.
4163 This includes both writes to individual VMCS fields and reads/writes
4164 through the MSR lists pointed to by the VMCS.
4166 x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that
4167 cover any x2APIC MSRs).
4169 Note, invoking this ioctl while a vCPU is running is inherently racy. However,
4170 KVM does guarantee that vCPUs will see either the previous filter or the new
4171 filter, e.g. MSRs with identical settings in both the old and new filter will
4172 have deterministic behavior.
4174 Similarly, if userspace wishes to intercept on denied accesses,
4175 KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and
4176 left enabled until after all filters are deactivated. Failure to do so may
4177 result in KVM injecting a #GP instead of exiting to userspace.
4179 4.98 KVM_CREATE_SPAPR_TCE_64
4180 ----------------------------
4182 :Capability: KVM_CAP_SPAPR_TCE_64
4183 :Architectures: powerpc
4185 :Parameters: struct kvm_create_spapr_tce_64 (in)
4186 :Returns: file descriptor for manipulating the created TCE table
4188 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4189 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4191 This capability uses extended struct in ioctl interface::
4193 /* for KVM_CAP_SPAPR_TCE_64 */
4194 struct kvm_create_spapr_tce_64 {
4198 __u64 offset; /* in pages */
4199 __u64 size; /* in pages */
4202 The aim of extension is to support an additional bigger DMA window with
4203 a variable page size.
4204 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4205 a bus offset of the corresponding DMA window, @size and @offset are numbers
4208 @flags are not used at the moment.
4210 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4212 4.99 KVM_REINJECT_CONTROL
4213 -------------------------
4215 :Capability: KVM_CAP_REINJECT_CONTROL
4218 :Parameters: struct kvm_reinject_control (in)
4219 :Returns: 0 on success,
4220 -EFAULT if struct kvm_reinject_control cannot be read,
4221 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4223 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
4224 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4225 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
4226 interrupt whenever there isn't a pending interrupt from i8254.
4227 !reinject mode injects an interrupt as soon as a tick arrives.
4231 struct kvm_reinject_control {
4236 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4237 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4239 4.100 KVM_PPC_CONFIGURE_V3_MMU
4240 ------------------------------
4242 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
4245 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
4246 :Returns: 0 on success,
4247 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4248 -EINVAL if the configuration is invalid
4250 This ioctl controls whether the guest will use radix or HPT (hashed
4251 page table) translation, and sets the pointer to the process table for
4256 struct kvm_ppc_mmuv3_cfg {
4258 __u64 process_table;
4261 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4262 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
4263 to use radix tree translation, and if clear, to use HPT translation.
4264 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4265 to be able to use the global TLB and SLB invalidation instructions;
4266 if clear, the guest may not use these instructions.
4268 The process_table field specifies the address and size of the guest
4269 process table, which is in the guest's space. This field is formatted
4270 as the second doubleword of the partition table entry, as defined in
4271 the Power ISA V3.00, Book III section 5.7.6.1.
4273 4.101 KVM_PPC_GET_RMMU_INFO
4274 ---------------------------
4276 :Capability: KVM_CAP_PPC_RADIX_MMU
4279 :Parameters: struct kvm_ppc_rmmu_info (out)
4280 :Returns: 0 on success,
4281 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4282 -EINVAL if no useful information can be returned
4284 This ioctl returns a structure containing two things: (a) a list
4285 containing supported radix tree geometries, and (b) a list that maps
4286 page sizes to put in the "AP" (actual page size) field for the tlbie
4287 (TLB invalidate entry) instruction.
4291 struct kvm_ppc_rmmu_info {
4292 struct kvm_ppc_radix_geom {
4297 __u32 ap_encodings[8];
4300 The geometries[] field gives up to 8 supported geometries for the
4301 radix page table, in terms of the log base 2 of the smallest page
4302 size, and the number of bits indexed at each level of the tree, from
4303 the PTE level up to the PGD level in that order. Any unused entries
4304 will have 0 in the page_shift field.
4306 The ap_encodings gives the supported page sizes and their AP field
4307 encodings, encoded with the AP value in the top 3 bits and the log
4308 base 2 of the page size in the bottom 6 bits.
4310 4.102 KVM_PPC_RESIZE_HPT_PREPARE
4311 --------------------------------
4313 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4314 :Architectures: powerpc
4316 :Parameters: struct kvm_ppc_resize_hpt (in)
4317 :Returns: 0 on successful completion,
4318 >0 if a new HPT is being prepared, the value is an estimated
4319 number of milliseconds until preparation is complete,
4320 -EFAULT if struct kvm_reinject_control cannot be read,
4321 -EINVAL if the supplied shift or flags are invalid,
4322 -ENOMEM if unable to allocate the new HPT,
4324 Used to implement the PAPR extension for runtime resizing of a guest's
4325 Hashed Page Table (HPT). Specifically this starts, stops or monitors
4326 the preparation of a new potential HPT for the guest, essentially
4327 implementing the H_RESIZE_HPT_PREPARE hypercall.
4331 struct kvm_ppc_resize_hpt {
4337 If called with shift > 0 when there is no pending HPT for the guest,
4338 this begins preparation of a new pending HPT of size 2^(shift) bytes.
4339 It then returns a positive integer with the estimated number of
4340 milliseconds until preparation is complete.
4342 If called when there is a pending HPT whose size does not match that
4343 requested in the parameters, discards the existing pending HPT and
4344 creates a new one as above.
4346 If called when there is a pending HPT of the size requested, will:
4348 * If preparation of the pending HPT is already complete, return 0
4349 * If preparation of the pending HPT has failed, return an error
4350 code, then discard the pending HPT.
4351 * If preparation of the pending HPT is still in progress, return an
4352 estimated number of milliseconds until preparation is complete.
4354 If called with shift == 0, discards any currently pending HPT and
4355 returns 0 (i.e. cancels any in-progress preparation).
4357 flags is reserved for future expansion, currently setting any bits in
4358 flags will result in an -EINVAL.
4360 Normally this will be called repeatedly with the same parameters until
4361 it returns <= 0. The first call will initiate preparation, subsequent
4362 ones will monitor preparation until it completes or fails.
4364 4.103 KVM_PPC_RESIZE_HPT_COMMIT
4365 -------------------------------
4367 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4368 :Architectures: powerpc
4370 :Parameters: struct kvm_ppc_resize_hpt (in)
4371 :Returns: 0 on successful completion,
4372 -EFAULT if struct kvm_reinject_control cannot be read,
4373 -EINVAL if the supplied shift or flags are invalid,
4374 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4375 have the requested size,
4376 -EBUSY if the pending HPT is not fully prepared,
4377 -ENOSPC if there was a hash collision when moving existing
4378 HPT entries to the new HPT,
4379 -EIO on other error conditions
4381 Used to implement the PAPR extension for runtime resizing of a guest's
4382 Hashed Page Table (HPT). Specifically this requests that the guest be
4383 transferred to working with the new HPT, essentially implementing the
4384 H_RESIZE_HPT_COMMIT hypercall.
4388 struct kvm_ppc_resize_hpt {
4394 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4395 returned 0 with the same parameters. In other cases
4396 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4397 -EBUSY, though others may be possible if the preparation was started,
4400 This will have undefined effects on the guest if it has not already
4401 placed itself in a quiescent state where no vcpu will make MMU enabled
4404 On succsful completion, the pending HPT will become the guest's active
4405 HPT and the previous HPT will be discarded.
4407 On failure, the guest will still be operating on its previous HPT.
4409 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4410 -----------------------------------
4412 :Capability: KVM_CAP_MCE
4415 :Parameters: u64 mce_cap (out)
4416 :Returns: 0 on success, -1 on error
4418 Returns supported MCE capabilities. The u64 mce_cap parameter
4419 has the same format as the MSR_IA32_MCG_CAP register. Supported
4420 capabilities will have the corresponding bits set.
4422 4.105 KVM_X86_SETUP_MCE
4423 -----------------------
4425 :Capability: KVM_CAP_MCE
4428 :Parameters: u64 mcg_cap (in)
4429 :Returns: 0 on success,
4430 -EFAULT if u64 mcg_cap cannot be read,
4431 -EINVAL if the requested number of banks is invalid,
4432 -EINVAL if requested MCE capability is not supported.
4434 Initializes MCE support for use. The u64 mcg_cap parameter
4435 has the same format as the MSR_IA32_MCG_CAP register and
4436 specifies which capabilities should be enabled. The maximum
4437 supported number of error-reporting banks can be retrieved when
4438 checking for KVM_CAP_MCE. The supported capabilities can be
4439 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4441 4.106 KVM_X86_SET_MCE
4442 ---------------------
4444 :Capability: KVM_CAP_MCE
4447 :Parameters: struct kvm_x86_mce (in)
4448 :Returns: 0 on success,
4449 -EFAULT if struct kvm_x86_mce cannot be read,
4450 -EINVAL if the bank number is invalid,
4451 -EINVAL if VAL bit is not set in status field.
4453 Inject a machine check error (MCE) into the guest. The input
4456 struct kvm_x86_mce {
4466 If the MCE being reported is an uncorrected error, KVM will
4467 inject it as an MCE exception into the guest. If the guest
4468 MCG_STATUS register reports that an MCE is in progress, KVM
4469 causes an KVM_EXIT_SHUTDOWN vmexit.
4471 Otherwise, if the MCE is a corrected error, KVM will just
4472 store it in the corresponding bank (provided this bank is
4473 not holding a previously reported uncorrected error).
4475 4.107 KVM_S390_GET_CMMA_BITS
4476 ----------------------------
4478 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4479 :Architectures: s390
4481 :Parameters: struct kvm_s390_cmma_log (in, out)
4482 :Returns: 0 on success, a negative value on error
4486 ====== =============================================================
4487 ENOMEM not enough memory can be allocated to complete the task
4488 ENXIO if CMMA is not enabled
4489 EINVAL if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled
4490 EINVAL if KVM_S390_CMMA_PEEK is not set but dirty tracking has been
4491 disabled (and thus migration mode was automatically disabled)
4492 EFAULT if the userspace address is invalid or if no page table is
4493 present for the addresses (e.g. when using hugepages).
4494 ====== =============================================================
4496 This ioctl is used to get the values of the CMMA bits on the s390
4497 architecture. It is meant to be used in two scenarios:
4499 - During live migration to save the CMMA values. Live migration needs
4500 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4501 - To non-destructively peek at the CMMA values, with the flag
4502 KVM_S390_CMMA_PEEK set.
4504 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4505 values are written to a buffer whose location is indicated via the "values"
4506 member in the kvm_s390_cmma_log struct. The values in the input struct are
4507 also updated as needed.
4509 Each CMMA value takes up one byte.
4513 struct kvm_s390_cmma_log {
4524 start_gfn is the number of the first guest frame whose CMMA values are
4527 count is the length of the buffer in bytes,
4529 values points to the buffer where the result will be written to.
4531 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4532 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4535 The result is written in the buffer pointed to by the field values, and
4536 the values of the input parameter are updated as follows.
4538 Depending on the flags, different actions are performed. The only
4539 supported flag so far is KVM_S390_CMMA_PEEK.
4541 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4542 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4543 It is not necessarily the same as the one passed as input, as clean pages
4546 count will indicate the number of bytes actually written in the buffer.
4547 It can (and very often will) be smaller than the input value, since the
4548 buffer is only filled until 16 bytes of clean values are found (which
4549 are then not copied in the buffer). Since a CMMA migration block needs
4550 the base address and the length, for a total of 16 bytes, we will send
4551 back some clean data if there is some dirty data afterwards, as long as
4552 the size of the clean data does not exceed the size of the header. This
4553 allows to minimize the amount of data to be saved or transferred over
4554 the network at the expense of more roundtrips to userspace. The next
4555 invocation of the ioctl will skip over all the clean values, saving
4556 potentially more than just the 16 bytes we found.
4558 If KVM_S390_CMMA_PEEK is set:
4559 the existing storage attributes are read even when not in migration
4560 mode, and no other action is performed;
4562 the output start_gfn will be equal to the input start_gfn,
4564 the output count will be equal to the input count, except if the end of
4565 memory has been reached.
4568 the field "remaining" will indicate the total number of dirty CMMA values
4569 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4574 values points to the userspace buffer where the result will be stored.
4576 4.108 KVM_S390_SET_CMMA_BITS
4577 ----------------------------
4579 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4580 :Architectures: s390
4582 :Parameters: struct kvm_s390_cmma_log (in)
4583 :Returns: 0 on success, a negative value on error
4585 This ioctl is used to set the values of the CMMA bits on the s390
4586 architecture. It is meant to be used during live migration to restore
4587 the CMMA values, but there are no restrictions on its use.
4588 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4589 Each CMMA value takes up one byte.
4593 struct kvm_s390_cmma_log {
4604 start_gfn indicates the starting guest frame number,
4606 count indicates how many values are to be considered in the buffer,
4608 flags is not used and must be 0.
4610 mask indicates which PGSTE bits are to be considered.
4612 remaining is not used.
4614 values points to the buffer in userspace where to store the values.
4616 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4617 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4618 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4619 if the flags field was not 0, with -EFAULT if the userspace address is
4620 invalid, if invalid pages are written to (e.g. after the end of memory)
4621 or if no page table is present for the addresses (e.g. when using
4624 4.109 KVM_PPC_GET_CPU_CHAR
4625 --------------------------
4627 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4628 :Architectures: powerpc
4630 :Parameters: struct kvm_ppc_cpu_char (out)
4631 :Returns: 0 on successful completion,
4632 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4634 This ioctl gives userspace information about certain characteristics
4635 of the CPU relating to speculative execution of instructions and
4636 possible information leakage resulting from speculative execution (see
4637 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4638 returned in struct kvm_ppc_cpu_char, which looks like this::
4640 struct kvm_ppc_cpu_char {
4641 __u64 character; /* characteristics of the CPU */
4642 __u64 behaviour; /* recommended software behaviour */
4643 __u64 character_mask; /* valid bits in character */
4644 __u64 behaviour_mask; /* valid bits in behaviour */
4647 For extensibility, the character_mask and behaviour_mask fields
4648 indicate which bits of character and behaviour have been filled in by
4649 the kernel. If the set of defined bits is extended in future then
4650 userspace will be able to tell whether it is running on a kernel that
4651 knows about the new bits.
4653 The character field describes attributes of the CPU which can help
4654 with preventing inadvertent information disclosure - specifically,
4655 whether there is an instruction to flash-invalidate the L1 data cache
4656 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4657 to a mode where entries can only be used by the thread that created
4658 them, whether the bcctr[l] instruction prevents speculation, and
4659 whether a speculation barrier instruction (ori 31,31,0) is provided.
4661 The behaviour field describes actions that software should take to
4662 prevent inadvertent information disclosure, and thus describes which
4663 vulnerabilities the hardware is subject to; specifically whether the
4664 L1 data cache should be flushed when returning to user mode from the
4665 kernel, and whether a speculation barrier should be placed between an
4666 array bounds check and the array access.
4668 These fields use the same bit definitions as the new
4669 H_GET_CPU_CHARACTERISTICS hypercall.
4671 4.110 KVM_MEMORY_ENCRYPT_OP
4672 ---------------------------
4677 :Parameters: an opaque platform specific structure (in/out)
4678 :Returns: 0 on success; -1 on error
4680 If the platform supports creating encrypted VMs then this ioctl can be used
4681 for issuing platform-specific memory encryption commands to manage those
4684 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4685 (SEV) commands on AMD Processors. The SEV commands are defined in
4686 Documentation/virt/kvm/x86/amd-memory-encryption.rst.
4688 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4689 -----------------------------------
4694 :Parameters: struct kvm_enc_region (in)
4695 :Returns: 0 on success; -1 on error
4697 This ioctl can be used to register a guest memory region which may
4698 contain encrypted data (e.g. guest RAM, SMRAM etc).
4700 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4701 memory region may contain encrypted data. The SEV memory encryption
4702 engine uses a tweak such that two identical plaintext pages, each at
4703 different locations will have differing ciphertexts. So swapping or
4704 moving ciphertext of those pages will not result in plaintext being
4705 swapped. So relocating (or migrating) physical backing pages for the SEV
4706 guest will require some additional steps.
4708 Note: The current SEV key management spec does not provide commands to
4709 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4710 memory region registered with the ioctl.
4712 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4713 -------------------------------------
4718 :Parameters: struct kvm_enc_region (in)
4719 :Returns: 0 on success; -1 on error
4721 This ioctl can be used to unregister the guest memory region registered
4722 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4724 4.113 KVM_HYPERV_EVENTFD
4725 ------------------------
4727 :Capability: KVM_CAP_HYPERV_EVENTFD
4730 :Parameters: struct kvm_hyperv_eventfd (in)
4732 This ioctl (un)registers an eventfd to receive notifications from the guest on
4733 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4734 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4735 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4739 struct kvm_hyperv_eventfd {
4746 The conn_id field should fit within 24 bits::
4748 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4750 The acceptable values for the flags field are::
4752 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4754 :Returns: 0 on success,
4755 -EINVAL if conn_id or flags is outside the allowed range,
4756 -ENOENT on deassign if the conn_id isn't registered,
4757 -EEXIST on assign if the conn_id is already registered
4759 4.114 KVM_GET_NESTED_STATE
4760 --------------------------
4762 :Capability: KVM_CAP_NESTED_STATE
4765 :Parameters: struct kvm_nested_state (in/out)
4766 :Returns: 0 on success, -1 on error
4770 ===== =============================================================
4771 E2BIG the total state size exceeds the value of 'size' specified by
4772 the user; the size required will be written into size.
4773 ===== =============================================================
4777 struct kvm_nested_state {
4783 struct kvm_vmx_nested_state_hdr vmx;
4784 struct kvm_svm_nested_state_hdr svm;
4786 /* Pad the header to 128 bytes. */
4791 struct kvm_vmx_nested_state_data vmx[0];
4792 struct kvm_svm_nested_state_data svm[0];
4796 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4797 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4798 #define KVM_STATE_NESTED_EVMCS 0x00000004
4800 #define KVM_STATE_NESTED_FORMAT_VMX 0
4801 #define KVM_STATE_NESTED_FORMAT_SVM 1
4803 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4805 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4806 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4808 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4810 struct kvm_vmx_nested_state_hdr {
4819 __u64 preemption_timer_deadline;
4822 struct kvm_vmx_nested_state_data {
4823 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4824 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4827 This ioctl copies the vcpu's nested virtualization state from the kernel to
4830 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4831 to the KVM_CHECK_EXTENSION ioctl().
4833 4.115 KVM_SET_NESTED_STATE
4834 --------------------------
4836 :Capability: KVM_CAP_NESTED_STATE
4839 :Parameters: struct kvm_nested_state (in)
4840 :Returns: 0 on success, -1 on error
4842 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4843 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4845 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4846 -------------------------------------
4848 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4849 KVM_CAP_COALESCED_PIO (for coalesced pio)
4852 :Parameters: struct kvm_coalesced_mmio_zone
4853 :Returns: 0 on success, < 0 on error
4855 Coalesced I/O is a performance optimization that defers hardware
4856 register write emulation so that userspace exits are avoided. It is
4857 typically used to reduce the overhead of emulating frequently accessed
4860 When a hardware register is configured for coalesced I/O, write accesses
4861 do not exit to userspace and their value is recorded in a ring buffer
4862 that is shared between kernel and userspace.
4864 Coalesced I/O is used if one or more write accesses to a hardware
4865 register can be deferred until a read or a write to another hardware
4866 register on the same device. This last access will cause a vmexit and
4867 userspace will process accesses from the ring buffer before emulating
4868 it. That will avoid exiting to userspace on repeated writes.
4870 Coalesced pio is based on coalesced mmio. There is little difference
4871 between coalesced mmio and pio except that coalesced pio records accesses
4874 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4875 ------------------------------------
4877 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4878 :Architectures: x86, arm64, mips
4880 :Parameters: struct kvm_clear_dirty_log (in)
4881 :Returns: 0 on success, -1 on error
4885 /* for KVM_CLEAR_DIRTY_LOG */
4886 struct kvm_clear_dirty_log {
4891 void __user *dirty_bitmap; /* one bit per page */
4896 The ioctl clears the dirty status of pages in a memory slot, according to
4897 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4898 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4899 memory slot, and num_pages is the size in bits of the input bitmap.
4900 first_page must be a multiple of 64; num_pages must also be a multiple of
4901 64 unless first_page + num_pages is the size of the memory slot. For each
4902 bit that is set in the input bitmap, the corresponding page is marked "clean"
4903 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4904 (for example via write-protection, or by clearing the dirty bit in
4905 a page table entry).
4907 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4908 the address space for which you want to clear the dirty status. See
4909 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4911 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4912 is enabled; for more information, see the description of the capability.
4913 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4914 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4916 4.118 KVM_GET_SUPPORTED_HV_CPUID
4917 --------------------------------
4919 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4921 :Type: system ioctl, vcpu ioctl
4922 :Parameters: struct kvm_cpuid2 (in/out)
4923 :Returns: 0 on success, -1 on error
4930 struct kvm_cpuid_entry2 entries[0];
4933 struct kvm_cpuid_entry2 {
4944 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4945 KVM. Userspace can use the information returned by this ioctl to construct
4946 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4947 Windows or Hyper-V guests).
4949 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4950 Functional Specification (TLFS). These leaves can't be obtained with
4951 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4952 leaves (0x40000000, 0x40000001).
4954 Currently, the following list of CPUID leaves are returned:
4956 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4957 - HYPERV_CPUID_INTERFACE
4958 - HYPERV_CPUID_VERSION
4959 - HYPERV_CPUID_FEATURES
4960 - HYPERV_CPUID_ENLIGHTMENT_INFO
4961 - HYPERV_CPUID_IMPLEMENT_LIMITS
4962 - HYPERV_CPUID_NESTED_FEATURES
4963 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4964 - HYPERV_CPUID_SYNDBG_INTERFACE
4965 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4967 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4968 with the 'nent' field indicating the number of entries in the variable-size
4969 array 'entries'. If the number of entries is too low to describe all Hyper-V
4970 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4971 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4972 number of valid entries in the 'entries' array, which is then filled.
4974 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4975 userspace should not expect to get any particular value there.
4977 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4978 system ioctl which exposes all supported feature bits unconditionally, vcpu
4979 version has the following quirks:
4981 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4982 feature bit are only exposed when Enlightened VMCS was previously enabled
4983 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4984 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4985 (presumes KVM_CREATE_IRQCHIP has already been called).
4987 4.119 KVM_ARM_VCPU_FINALIZE
4988 ---------------------------
4990 :Architectures: arm64
4992 :Parameters: int feature (in)
4993 :Returns: 0 on success, -1 on error
4997 ====== ==============================================================
4998 EPERM feature not enabled, needs configuration, or already finalized
4999 EINVAL feature unknown or not present
5000 ====== ==============================================================
5002 Recognised values for feature:
5004 ===== ===========================================
5005 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
5006 ===== ===========================================
5008 Finalizes the configuration of the specified vcpu feature.
5010 The vcpu must already have been initialised, enabling the affected feature, by
5011 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
5014 For affected vcpu features, this is a mandatory step that must be performed
5015 before the vcpu is fully usable.
5017 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
5018 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
5019 that should be performaned and how to do it are feature-dependent.
5021 Other calls that depend on a particular feature being finalized, such as
5022 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
5023 -EPERM unless the feature has already been finalized by means of a
5024 KVM_ARM_VCPU_FINALIZE call.
5026 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
5029 4.120 KVM_SET_PMU_EVENT_FILTER
5030 ------------------------------
5032 :Capability: KVM_CAP_PMU_EVENT_FILTER
5035 :Parameters: struct kvm_pmu_event_filter (in)
5036 :Returns: 0 on success, -1 on error
5040 ====== ============================================================
5041 EFAULT args[0] cannot be accessed
5042 EINVAL args[0] contains invalid data in the filter or filter events
5043 E2BIG nevents is too large
5044 EBUSY not enough memory to allocate the filter
5045 ====== ============================================================
5049 struct kvm_pmu_event_filter {
5052 __u32 fixed_counter_bitmap;
5058 This ioctl restricts the set of PMU events the guest can program by limiting
5059 which event select and unit mask combinations are permitted.
5061 The argument holds a list of filter events which will be allowed or denied.
5063 Filter events only control general purpose counters; fixed purpose counters
5064 are controlled by the fixed_counter_bitmap.
5066 Valid values for 'flags'::
5070 To use this mode, clear the 'flags' field.
5072 In this mode each event will contain an event select + unit mask.
5074 When the guest attempts to program the PMU the guest's event select +
5075 unit mask is compared against the filter events to determine whether the
5076 guest should have access.
5078 ``KVM_PMU_EVENT_FLAG_MASKED_EVENTS``
5079 :Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS
5081 In this mode each filter event will contain an event select, mask, match, and
5082 exclude value. To encode a masked event use::
5084 KVM_PMU_ENCODE_MASKED_ENTRY()
5086 An encoded event will follow this layout::
5090 7:0 event select (low bits)
5093 35:32 event select (high bits)
5098 When the guest attempts to program the PMU, these steps are followed in
5099 determining if the guest should have access:
5101 1. Match the event select from the guest against the filter events.
5102 2. If a match is found, match the guest's unit mask to the mask and match
5103 values of the included filter events.
5104 I.e. (unit mask & mask) == match && !exclude.
5105 3. If a match is found, match the guest's unit mask to the mask and match
5106 values of the excluded filter events.
5107 I.e. (unit mask & mask) == match && exclude.
5109 a. If an included match is found and an excluded match is not found, filter
5111 b. For everything else, do not filter the event.
5113 a. If the event is filtered and it's an allow list, allow the guest to
5115 b. If the event is filtered and it's a deny list, do not allow the guest to
5118 When setting a new pmu event filter, -EINVAL will be returned if any of the
5119 unused fields are set or if any of the high bits (35:32) in the event
5120 select are set when called on Intel.
5122 Valid values for 'action'::
5124 #define KVM_PMU_EVENT_ALLOW 0
5125 #define KVM_PMU_EVENT_DENY 1
5127 4.121 KVM_PPC_SVM_OFF
5128 ---------------------
5131 :Architectures: powerpc
5134 :Returns: 0 on successful completion,
5138 ====== ================================================================
5139 EINVAL if ultravisor failed to terminate the secure guest
5140 ENOMEM if hypervisor failed to allocate new radix page tables for guest
5141 ====== ================================================================
5143 This ioctl is used to turn off the secure mode of the guest or transition
5144 the guest from secure mode to normal mode. This is invoked when the guest
5145 is reset. This has no effect if called for a normal guest.
5147 This ioctl issues an ultravisor call to terminate the secure guest,
5148 unpins the VPA pages and releases all the device pages that are used to
5149 track the secure pages by hypervisor.
5151 4.122 KVM_S390_NORMAL_RESET
5152 ---------------------------
5154 :Capability: KVM_CAP_S390_VCPU_RESETS
5155 :Architectures: s390
5160 This ioctl resets VCPU registers and control structures according to
5161 the cpu reset definition in the POP (Principles Of Operation).
5163 4.123 KVM_S390_INITIAL_RESET
5164 ----------------------------
5167 :Architectures: s390
5172 This ioctl resets VCPU registers and control structures according to
5173 the initial cpu reset definition in the POP. However, the cpu is not
5174 put into ESA mode. This reset is a superset of the normal reset.
5176 4.124 KVM_S390_CLEAR_RESET
5177 --------------------------
5179 :Capability: KVM_CAP_S390_VCPU_RESETS
5180 :Architectures: s390
5185 This ioctl resets VCPU registers and control structures according to
5186 the clear cpu reset definition in the POP. However, the cpu is not put
5187 into ESA mode. This reset is a superset of the initial reset.
5190 4.125 KVM_S390_PV_COMMAND
5191 -------------------------
5193 :Capability: KVM_CAP_S390_PROTECTED
5194 :Architectures: s390
5196 :Parameters: struct kvm_pv_cmd
5197 :Returns: 0 on success, < 0 on error
5202 __u32 cmd; /* Command to be executed */
5203 __u16 rc; /* Ultravisor return code */
5204 __u16 rrc; /* Ultravisor return reason code */
5205 __u64 data; /* Data or address */
5206 __u32 flags; /* flags for future extensions. Must be 0 for now */
5210 **Ultravisor return codes**
5211 The Ultravisor return (reason) codes are provided by the kernel if a
5212 Ultravisor call has been executed to achieve the results expected by
5213 the command. Therefore they are independent of the IOCTL return
5214 code. If KVM changes `rc`, its value will always be greater than 0
5215 hence setting it to 0 before issuing a PV command is advised to be
5216 able to detect a change of `rc`.
5221 Allocate memory and register the VM with the Ultravisor, thereby
5222 donating memory to the Ultravisor that will become inaccessible to
5223 KVM. All existing CPUs are converted to protected ones. After this
5224 command has succeeded, any CPU added via hotplug will become
5225 protected during its creation as well.
5229 ===== =============================
5230 EINTR an unmasked signal is pending
5231 ===== =============================
5234 Deregister the VM from the Ultravisor and reclaim the memory that had
5235 been donated to the Ultravisor, making it usable by the kernel again.
5236 All registered VCPUs are converted back to non-protected ones. If a
5237 previous protected VM had been prepared for asynchronous teardown with
5238 KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with
5239 KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call
5240 together with the current protected VM.
5242 KVM_PV_VM_SET_SEC_PARMS
5243 Pass the image header from VM memory to the Ultravisor in
5244 preparation of image unpacking and verification.
5247 Unpack (protect and decrypt) a page of the encrypted boot image.
5250 Verify the integrity of the unpacked image. Only if this succeeds,
5251 KVM is allowed to start protected VCPUs.
5254 :Capability: KVM_CAP_S390_PROTECTED_DUMP
5256 Presents an API that provides Ultravisor related data to userspace
5257 via subcommands. len_max is the size of the user space buffer,
5258 len_written is KVM's indication of how much bytes of that buffer
5259 were actually written to. len_written can be used to determine the
5260 valid fields if more response fields are added in the future.
5264 enum pv_cmd_info_id {
5269 struct kvm_s390_pv_info_header {
5276 struct kvm_s390_pv_info {
5277 struct kvm_s390_pv_info_header header;
5278 struct kvm_s390_pv_info_dump dump;
5279 struct kvm_s390_pv_info_vm vm;
5285 This subcommand provides basic Ultravisor information for PV
5286 hosts. These values are likely also exported as files in the sysfs
5287 firmware UV query interface but they are more easily available to
5288 programs in this API.
5290 The installed calls and feature_indication members provide the
5291 installed UV calls and the UV's other feature indications.
5293 The max_* members provide information about the maximum number of PV
5294 vcpus, PV guests and PV guest memory size.
5298 struct kvm_s390_pv_info_vm {
5299 __u64 inst_calls_list[4];
5302 __u64 max_guest_addr;
5303 __u64 feature_indication;
5308 This subcommand provides information related to dumping PV guests.
5312 struct kvm_s390_pv_info_dump {
5313 __u64 dump_cpu_buffer_len;
5314 __u64 dump_config_mem_buffer_per_1m;
5315 __u64 dump_config_finalize_len;
5319 :Capability: KVM_CAP_S390_PROTECTED_DUMP
5321 Presents an API that provides calls which facilitate dumping a
5326 struct kvm_s390_pv_dmp {
5330 __u64 gaddr; /* For dump storage state */
5336 Initializes the dump process of a protected VM. If this call does
5337 not succeed all other subcommands will fail with -EINVAL. This
5338 subcommand will return -EINVAL if a dump process has not yet been
5341 Not all PV vms can be dumped, the owner needs to set `dump
5342 allowed` PCF bit 34 in the SE header to allow dumping.
5344 KVM_PV_DUMP_CONFIG_STOR_STATE
5345 Stores `buff_len` bytes of tweak component values starting with
5346 the 1MB block specified by the absolute guest address
5347 (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len`
5348 aligned and at least >= the `conf_dump_storage_state_len` value
5349 provided by the dump uv_info data. buff_user might be written to
5350 even if an error rc is returned. For instance if we encounter a
5351 fault after writing the first page of data.
5353 KVM_PV_DUMP_COMPLETE
5354 If the subcommand succeeds it completes the dump process and lets
5355 KVM_PV_DUMP_INIT be called again.
5357 On success `conf_dump_finalize_len` bytes of completion data will be
5358 stored to the `buff_addr`. The completion data contains a key
5359 derivation seed, IV, tweak nonce and encryption keys as well as an
5360 authentication tag all of which are needed to decrypt the dump at a
5363 KVM_PV_ASYNC_CLEANUP_PREPARE
5364 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE
5366 Prepare the current protected VM for asynchronous teardown. Most
5367 resources used by the current protected VM will be set aside for a
5368 subsequent asynchronous teardown. The current protected VM will then
5369 resume execution immediately as non-protected. There can be at most
5370 one protected VM prepared for asynchronous teardown at any time. If
5371 a protected VM had already been prepared for teardown without
5372 subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will
5373 fail. In that case, the userspace process should issue a normal
5374 KVM_PV_DISABLE. The resources set aside with this call will need to
5375 be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM
5376 or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM
5377 terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon
5378 as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes.
5380 KVM_PV_ASYNC_CLEANUP_PERFORM
5381 :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE
5383 Tear down the protected VM previously prepared for teardown with
5384 KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside
5385 will be freed during the execution of this command. This PV command
5386 should ideally be issued by userspace from a separate thread. If a
5387 fatal signal is received (or the process terminates naturally), the
5388 command will terminate immediately without completing, and the normal
5389 KVM shutdown procedure will take care of cleaning up all remaining
5390 protected VMs, including the ones whose teardown was interrupted by
5391 process termination.
5393 4.126 KVM_XEN_HVM_SET_ATTR
5394 --------------------------
5396 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5399 :Parameters: struct kvm_xen_hvm_attr
5400 :Returns: 0 on success, < 0 on error
5404 struct kvm_xen_hvm_attr {
5410 __u8 runstate_update_flag;
5416 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */
5425 __u32 port; /* Zero for eventfd */
5438 KVM_XEN_ATTR_TYPE_LONG_MODE
5439 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
5440 determines the layout of the shared info pages exposed to the VM.
5442 KVM_XEN_ATTR_TYPE_SHARED_INFO
5443 Sets the guest physical frame number at which the Xen "shared info"
5444 page resides. Note that although Xen places vcpu_info for the first
5445 32 vCPUs in the shared_info page, KVM does not automatically do so
5446 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
5447 explicitly even when the vcpu_info for a given vCPU resides at the
5448 "default" location in the shared_info page. This is because KVM may
5449 not be aware of the Xen CPU id which is used as the index into the
5450 vcpu_info[] array, so may know the correct default location.
5452 Note that the shared info page may be constantly written to by KVM;
5453 it contains the event channel bitmap used to deliver interrupts to
5454 a Xen guest, amongst other things. It is exempt from dirty tracking
5455 mechanisms — KVM will not explicitly mark the page as dirty each
5456 time an event channel interrupt is delivered to the guest! Thus,
5457 userspace should always assume that the designated GFN is dirty if
5458 any vCPU has been running or any event channel interrupts can be
5459 routed to the guest.
5461 Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared info
5464 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
5465 Sets the exception vector used to deliver Xen event channel upcalls.
5466 This is the HVM-wide vector injected directly by the hypervisor
5467 (not through the local APIC), typically configured by a guest via
5468 HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest
5469 SHUTDOWN_soft_reset) by setting it to zero.
5471 KVM_XEN_ATTR_TYPE_EVTCHN
5472 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5473 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5474 an outbound port number for interception of EVTCHNOP_send requests
5475 from the guest. A given sending port number may be directed back to
5476 a specified vCPU (by APIC ID) / port / priority on the guest, or to
5477 trigger events on an eventfd. The vCPU and priority can be changed
5478 by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but but other
5479 fields cannot change for a given sending port. A port mapping is
5480 removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing
5481 KVM_XEN_EVTCHN_RESET in the flags field removes all interception of
5482 outbound event channels. The values of the flags field are mutually
5483 exclusive and cannot be combined as a bitmask.
5485 KVM_XEN_ATTR_TYPE_XEN_VERSION
5486 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5487 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5488 the 32-bit version code returned to the guest when it invokes the
5489 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV
5490 Xen guests will often use this to as a dummy hypercall to trigger
5491 event channel delivery, so responding within the kernel without
5492 exiting to userspace is beneficial.
5494 KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG
5495 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5496 support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the
5497 XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read
5498 other vCPUs' vcpu_runstate_info. Xen guests enable this feature via
5499 the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist
5502 4.127 KVM_XEN_HVM_GET_ATTR
5503 --------------------------
5505 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5508 :Parameters: struct kvm_xen_hvm_attr
5509 :Returns: 0 on success, < 0 on error
5511 Allows Xen VM attributes to be read. For the structure and types,
5512 see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN
5513 attribute cannot be read.
5515 4.128 KVM_XEN_VCPU_SET_ATTR
5516 ---------------------------
5518 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5521 :Parameters: struct kvm_xen_vcpu_attr
5522 :Returns: 0 on success, < 0 on error
5526 struct kvm_xen_vcpu_attr {
5534 __u64 state_entry_time;
5536 __u64 time_runnable;
5552 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
5553 Sets the guest physical address of the vcpu_info for a given vCPU.
5554 As with the shared_info page for the VM, the corresponding page may be
5555 dirtied at any time if event channel interrupt delivery is enabled, so
5556 userspace should always assume that the page is dirty without relying
5557 on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable
5560 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
5561 Sets the guest physical address of an additional pvclock structure
5562 for a given vCPU. This is typically used for guest vsyscall support.
5563 Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure.
5565 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5566 Sets the guest physical address of the vcpu_runstate_info for a given
5567 vCPU. This is how a Xen guest tracks CPU state such as steal time.
5568 Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area.
5570 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5571 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5572 the given vCPU from the .u.runstate.state member of the structure.
5573 KVM automatically accounts running and runnable time but blocked
5574 and offline states are only entered explicitly.
5576 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5577 Sets all fields of the vCPU runstate data from the .u.runstate member
5578 of the structure, including the current runstate. The state_entry_time
5579 must equal the sum of the other four times.
5581 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5582 This *adds* the contents of the .u.runstate members of the structure
5583 to the corresponding members of the given vCPU's runstate data, thus
5584 permitting atomic adjustments to the runstate times. The adjustment
5585 to the state_entry_time must equal the sum of the adjustments to the
5586 other four times. The state field must be set to -1, or to a valid
5587 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5588 or RUNSTATE_offline) to set the current accounted state as of the
5589 adjusted state_entry_time.
5591 KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID
5592 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5593 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen
5594 vCPU ID of the given vCPU, to allow timer-related VCPU operations to
5595 be intercepted by KVM.
5597 KVM_XEN_VCPU_ATTR_TYPE_TIMER
5598 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5599 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5600 event channel port/priority for the VIRQ_TIMER of the vCPU, as well
5601 as allowing a pending timer to be saved/restored. Setting the timer
5602 port to zero disables kernel handling of the singleshot timer.
5604 KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR
5605 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5606 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5607 per-vCPU local APIC upcall vector, configured by a Xen guest with
5608 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically
5609 used by Windows guests, and is distinct from the HVM-wide upcall
5610 vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by
5611 setting the vector to zero.
5614 4.129 KVM_XEN_VCPU_GET_ATTR
5615 ---------------------------
5617 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5620 :Parameters: struct kvm_xen_vcpu_attr
5621 :Returns: 0 on success, < 0 on error
5623 Allows Xen vCPU attributes to be read. For the structure and types,
5624 see KVM_XEN_VCPU_SET_ATTR above.
5626 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5627 with the KVM_XEN_VCPU_GET_ATTR ioctl.
5629 4.130 KVM_ARM_MTE_COPY_TAGS
5630 ---------------------------
5632 :Capability: KVM_CAP_ARM_MTE
5633 :Architectures: arm64
5635 :Parameters: struct kvm_arm_copy_mte_tags
5636 :Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5637 arguments, -EFAULT if memory cannot be accessed).
5641 struct kvm_arm_copy_mte_tags {
5649 Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5650 ``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned.
5651 ``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr``
5652 field must point to a buffer which the tags will be copied to or from.
5654 ``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5655 ``KVM_ARM_TAGS_FROM_GUEST``.
5657 The size of the buffer to store the tags is ``(length / 16)`` bytes
5658 (granules in MTE are 16 bytes long). Each byte contains a single tag
5659 value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5660 ``PTRACE_POKEMTETAGS``.
5662 If an error occurs before any data is copied then a negative error code is
5663 returned. If some tags have been copied before an error occurs then the number
5664 of bytes successfully copied is returned. If the call completes successfully
5665 then ``length`` is returned.
5667 4.131 KVM_GET_SREGS2
5668 --------------------
5670 :Capability: KVM_CAP_SREGS2
5673 :Parameters: struct kvm_sregs2 (out)
5674 :Returns: 0 on success, -1 on error
5676 Reads special registers from the vcpu.
5677 This ioctl (when supported) replaces the KVM_GET_SREGS.
5682 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5683 struct kvm_segment cs, ds, es, fs, gs, ss;
5684 struct kvm_segment tr, ldt;
5685 struct kvm_dtable gdt, idt;
5686 __u64 cr0, cr2, cr3, cr4, cr8;
5693 flags values for ``kvm_sregs2``:
5695 ``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5697 Indicates that the struct contains valid PDPTR values.
5700 4.132 KVM_SET_SREGS2
5701 --------------------
5703 :Capability: KVM_CAP_SREGS2
5706 :Parameters: struct kvm_sregs2 (in)
5707 :Returns: 0 on success, -1 on error
5709 Writes special registers into the vcpu.
5710 See KVM_GET_SREGS2 for the data structures.
5711 This ioctl (when supported) replaces the KVM_SET_SREGS.
5713 4.133 KVM_GET_STATS_FD
5714 ----------------------
5716 :Capability: KVM_CAP_STATS_BINARY_FD
5718 :Type: vm ioctl, vcpu ioctl
5720 :Returns: statistics file descriptor on success, < 0 on error
5724 ====== ======================================================
5725 ENOMEM if the fd could not be created due to lack of memory
5726 EMFILE if the number of opened files exceeds the limit
5727 ====== ======================================================
5729 The returned file descriptor can be used to read VM/vCPU statistics data in
5730 binary format. The data in the file descriptor consists of four blocks
5731 organized as follows:
5743 Apart from the header starting at offset 0, please be aware that it is
5744 not guaranteed that the four blocks are adjacent or in the above order;
5745 the offsets of the id, descriptors and data blocks are found in the
5746 header. However, all four blocks are aligned to 64 bit offsets in the
5747 file and they do not overlap.
5749 All blocks except the data block are immutable. Userspace can read them
5750 only one time after retrieving the file descriptor, and then use ``pread`` or
5751 ``lseek`` to read the statistics repeatedly.
5753 All data is in system endianness.
5755 The format of the header is as follows::
5757 struct kvm_stats_header {
5766 The ``flags`` field is not used at the moment. It is always read as 0.
5768 The ``name_size`` field is the size (in byte) of the statistics name string
5769 (including trailing '\0') which is contained in the "id string" block and
5770 appended at the end of every descriptor.
5772 The ``num_desc`` field is the number of descriptors that are included in the
5773 descriptor block. (The actual number of values in the data block may be
5774 larger, since each descriptor may comprise more than one value).
5776 The ``id_offset`` field is the offset of the id string from the start of the
5777 file indicated by the file descriptor. It is a multiple of 8.
5779 The ``desc_offset`` field is the offset of the Descriptors block from the start
5780 of the file indicated by the file descriptor. It is a multiple of 8.
5782 The ``data_offset`` field is the offset of the Stats Data block from the start
5783 of the file indicated by the file descriptor. It is a multiple of 8.
5785 The id string block contains a string which identifies the file descriptor on
5786 which KVM_GET_STATS_FD was invoked. The size of the block, including the
5787 trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5789 The descriptors block is only needed to be read once for the lifetime of the
5790 file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5791 by a string of size ``name_size``.
5794 #define KVM_STATS_TYPE_SHIFT 0
5795 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT)
5796 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT)
5797 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT)
5798 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT)
5799 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT)
5800 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT)
5801 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST
5803 #define KVM_STATS_UNIT_SHIFT 4
5804 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT)
5805 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT)
5806 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT)
5807 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT)
5808 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT)
5809 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT)
5810 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN
5812 #define KVM_STATS_BASE_SHIFT 8
5813 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT)
5814 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT)
5815 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT)
5816 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2
5818 struct kvm_stats_desc {
5827 The ``flags`` field contains the type and unit of the statistics data described
5828 by this descriptor. Its endianness is CPU native.
5829 The following flags are supported:
5831 Bits 0-3 of ``flags`` encode the type:
5833 * ``KVM_STATS_TYPE_CUMULATIVE``
5834 The statistics reports a cumulative count. The value of data can only be increased.
5835 Most of the counters used in KVM are of this type.
5836 The corresponding ``size`` field for this type is always 1.
5837 All cumulative statistics data are read/write.
5838 * ``KVM_STATS_TYPE_INSTANT``
5839 The statistics reports an instantaneous value. Its value can be increased or
5840 decreased. This type is usually used as a measurement of some resources,
5841 like the number of dirty pages, the number of large pages, etc.
5842 All instant statistics are read only.
5843 The corresponding ``size`` field for this type is always 1.
5844 * ``KVM_STATS_TYPE_PEAK``
5845 The statistics data reports a peak value, for example the maximum number
5846 of items in a hash table bucket, the longest time waited and so on.
5847 The value of data can only be increased.
5848 The corresponding ``size`` field for this type is always 1.
5849 * ``KVM_STATS_TYPE_LINEAR_HIST``
5850 The statistic is reported as a linear histogram. The number of
5851 buckets is specified by the ``size`` field. The size of buckets is specified
5852 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``)
5853 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last
5854 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity
5856 * ``KVM_STATS_TYPE_LOG_HIST``
5857 The statistic is reported as a logarithmic histogram. The number of
5858 buckets is specified by the ``size`` field. The range of the first bucket is
5859 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF).
5860 Otherwise, The Nth bucket (1 < N < ``size``) covers
5861 [pow(2, N-2), pow(2, N-1)).
5863 Bits 4-7 of ``flags`` encode the unit:
5865 * ``KVM_STATS_UNIT_NONE``
5866 There is no unit for the value of statistics data. This usually means that
5867 the value is a simple counter of an event.
5868 * ``KVM_STATS_UNIT_BYTES``
5869 It indicates that the statistics data is used to measure memory size, in the
5870 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
5871 determined by the ``exponent`` field in the descriptor.
5872 * ``KVM_STATS_UNIT_SECONDS``
5873 It indicates that the statistics data is used to measure time or latency.
5874 * ``KVM_STATS_UNIT_CYCLES``
5875 It indicates that the statistics data is used to measure CPU clock cycles.
5876 * ``KVM_STATS_UNIT_BOOLEAN``
5877 It indicates that the statistic will always be either 0 or 1. Boolean
5878 statistics of "peak" type will never go back from 1 to 0. Boolean
5879 statistics can be linear histograms (with two buckets) but not logarithmic
5882 Note that, in the case of histograms, the unit applies to the bucket
5883 ranges, while the bucket value indicates how many samples fell in the
5886 Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
5889 * ``KVM_STATS_BASE_POW10``
5890 The scale is based on power of 10. It is used for measurement of time and
5891 CPU clock cycles. For example, an exponent of -9 can be used with
5892 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
5893 * ``KVM_STATS_BASE_POW2``
5894 The scale is based on power of 2. It is used for measurement of memory size.
5895 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
5896 express that the unit is MiB.
5898 The ``size`` field is the number of values of this statistics data. Its
5899 value is usually 1 for most of simple statistics. 1 means it contains an
5900 unsigned 64bit data.
5902 The ``offset`` field is the offset from the start of Data Block to the start of
5903 the corresponding statistics data.
5905 The ``bucket_size`` field is used as a parameter for histogram statistics data.
5906 It is only used by linear histogram statistics data, specifying the size of a
5907 bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``.
5909 The ``name`` field is the name string of the statistics data. The name string
5910 starts at the end of ``struct kvm_stats_desc``. The maximum length including
5911 the trailing ``'\0'``, is indicated by ``name_size`` in the header.
5913 The Stats Data block contains an array of 64-bit values in the same order
5914 as the descriptors in Descriptors block.
5916 4.134 KVM_GET_XSAVE2
5917 --------------------
5919 :Capability: KVM_CAP_XSAVE2
5922 :Parameters: struct kvm_xsave (out)
5923 :Returns: 0 on success, -1 on error
5933 This ioctl would copy current vcpu's xsave struct to the userspace. It
5934 copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2)
5935 when invoked on the vm file descriptor. The size value returned by
5936 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
5937 Currently, it is only greater than 4096 if a dynamic feature has been
5938 enabled with ``arch_prctl()``, but this may change in the future.
5940 The offsets of the state save areas in struct kvm_xsave follow the contents
5941 of CPUID leaf 0xD on the host.
5943 4.135 KVM_XEN_HVM_EVTCHN_SEND
5944 -----------------------------
5946 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND
5949 :Parameters: struct kvm_irq_routing_xen_evtchn
5950 :Returns: 0 on success, < 0 on error
5955 struct kvm_irq_routing_xen_evtchn {
5961 This ioctl injects an event channel interrupt directly to the guest vCPU.
5963 4.136 KVM_S390_PV_CPU_COMMAND
5964 -----------------------------
5966 :Capability: KVM_CAP_S390_PROTECTED_DUMP
5967 :Architectures: s390
5970 :Returns: 0 on success, < 0 on error
5972 This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests
5973 for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares
5979 Presents an API that provides calls which facilitate dumping a vcpu
5985 Provides encrypted dump data like register values.
5986 The length of the returned data is provided by uv_info.guest_cpu_stor_len.
5988 4.137 KVM_S390_ZPCI_OP
5989 ----------------------
5991 :Capability: KVM_CAP_S390_ZPCI_OP
5992 :Architectures: s390
5994 :Parameters: struct kvm_s390_zpci_op (in)
5995 :Returns: 0 on success, <0 on error
5997 Used to manage hardware-assisted virtualization features for zPCI devices.
5999 Parameters are specified via the following structure::
6001 struct kvm_s390_zpci_op {
6003 __u32 fh; /* target device */
6004 __u8 op; /* operation to perform */
6007 /* for KVM_S390_ZPCIOP_REG_AEN */
6009 __u64 ibv; /* Guest addr of interrupt bit vector */
6010 __u64 sb; /* Guest addr of summary bit */
6012 __u32 noi; /* Number of interrupts */
6013 __u8 isc; /* Guest interrupt subclass */
6014 __u8 sbo; /* Offset of guest summary bit vector */
6021 The type of operation is specified in the "op" field.
6022 KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event
6023 notification interpretation, which will allow firmware delivery of adapter
6024 events directly to the vm, with KVM providing a backup delivery mechanism;
6025 KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of
6026 adapter event notifications.
6028 The target zPCI function must also be specified via the "fh" field. For the
6029 KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware
6030 delivery must be provided via the "reg_aen" struct.
6032 The "pad" and "reserved" fields may be used for future extensions and should be
6033 set to 0s by userspace.
6035 4.138 KVM_ARM_SET_COUNTER_OFFSET
6036 --------------------------------
6038 :Capability: KVM_CAP_COUNTER_OFFSET
6039 :Architectures: arm64
6041 :Parameters: struct kvm_arm_counter_offset (in)
6042 :Returns: 0 on success, < 0 on error
6044 This capability indicates that userspace is able to apply a single VM-wide
6045 offset to both the virtual and physical counters as viewed by the guest
6046 using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure:
6050 struct kvm_arm_counter_offset {
6051 __u64 counter_offset;
6055 The offset describes a number of counter cycles that are subtracted from
6056 both virtual and physical counter views (similar to the effects of the
6057 CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset
6058 always applies to all vcpus (already created or created after this ioctl)
6061 It is userspace's responsibility to compute the offset based, for example,
6062 on previous values of the guest counters.
6064 Any value other than 0 for the "reserved" field may result in an error
6065 (-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu
6066 ioctl is issued concurrently.
6068 Note that using this ioctl results in KVM ignoring subsequent userspace
6069 writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG
6070 interface. No error will be returned, but the resulting offset will not be
6073 5. The kvm_run structure
6074 ========================
6076 Application code obtains a pointer to the kvm_run structure by
6077 mmap()ing a vcpu fd. From that point, application code can control
6078 execution by changing fields in kvm_run prior to calling the KVM_RUN
6079 ioctl, and obtain information about the reason KVM_RUN returned by
6080 looking up structure members.
6086 __u8 request_interrupt_window;
6088 Request that KVM_RUN return when it becomes possible to inject external
6089 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
6093 __u8 immediate_exit;
6095 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
6096 exits immediately, returning -EINTR. In the common scenario where a
6097 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
6098 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
6099 Rather than blocking the signal outside KVM_RUN, userspace can set up
6100 a signal handler that sets run->immediate_exit to a non-zero value.
6102 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
6111 When KVM_RUN has returned successfully (return value 0), this informs
6112 application code why KVM_RUN has returned. Allowable values for this
6113 field are detailed below.
6117 __u8 ready_for_interrupt_injection;
6119 If request_interrupt_window has been specified, this field indicates
6120 an interrupt can be injected now with KVM_INTERRUPT.
6126 The value of the current interrupt flag. Only valid if in-kernel
6127 local APIC is not used.
6133 More architecture-specific flags detailing state of the VCPU that may
6134 affect the device's behavior. Current defined flags::
6136 /* x86, set if the VCPU is in system management mode */
6137 #define KVM_RUN_X86_SMM (1 << 0)
6138 /* x86, set if bus lock detected in VM */
6139 #define KVM_RUN_BUS_LOCK (1 << 1)
6140 /* arm64, set for KVM_EXIT_DEBUG */
6141 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0)
6145 /* in (pre_kvm_run), out (post_kvm_run) */
6148 The value of the cr8 register. Only valid if in-kernel local APIC is
6149 not used. Both input and output.
6155 The value of the APIC BASE msr. Only valid if in-kernel local
6156 APIC is not used. Both input and output.
6161 /* KVM_EXIT_UNKNOWN */
6163 __u64 hardware_exit_reason;
6166 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
6167 reasons. Further architecture-specific information is available in
6168 hardware_exit_reason.
6172 /* KVM_EXIT_FAIL_ENTRY */
6174 __u64 hardware_entry_failure_reason;
6175 __u32 cpu; /* if KVM_LAST_CPU */
6178 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
6179 to unknown reasons. Further architecture-specific information is
6180 available in hardware_entry_failure_reason.
6184 /* KVM_EXIT_EXCEPTION */
6196 #define KVM_EXIT_IO_IN 0
6197 #define KVM_EXIT_IO_OUT 1
6199 __u8 size; /* bytes */
6202 __u64 data_offset; /* relative to kvm_run start */
6205 If exit_reason is KVM_EXIT_IO, then the vcpu has
6206 executed a port I/O instruction which could not be satisfied by kvm.
6207 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
6208 where kvm expects application code to place the data for the next
6209 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
6213 /* KVM_EXIT_DEBUG */
6215 struct kvm_debug_exit_arch arch;
6218 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
6219 for which architecture specific information is returned.
6231 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
6232 executed a memory-mapped I/O instruction which could not be satisfied
6233 by kvm. The 'data' member contains the written data if 'is_write' is
6234 true, and should be filled by application code otherwise.
6236 The 'data' member contains, in its first 'len' bytes, the value as it would
6237 appear if the VCPU performed a load or store of the appropriate width directly
6242 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
6243 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
6244 operations are complete (and guest state is consistent) only after userspace
6245 has re-entered the kernel with KVM_RUN. The kernel side will first finish
6246 incomplete operations and then check for pending signals.
6248 The pending state of the operation is not preserved in state which is
6249 visible to userspace, thus userspace should ensure that the operation is
6250 completed before performing a live migration. Userspace can re-enter the
6251 guest with an unmasked signal pending or with the immediate_exit field set
6252 to complete pending operations without allowing any further instructions
6257 /* KVM_EXIT_HYPERCALL */
6266 It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or
6267 ``KVM_EXIT_MMIO`` (all except s390) to implement functionality that
6268 requires a guest to interact with host userspace.
6270 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
6275 SMCCC exits can be enabled depending on the configuration of the SMCCC
6276 filter. See the Documentation/virt/kvm/devices/vm.rst
6277 ``KVM_ARM_SMCCC_FILTER`` for more details.
6279 ``nr`` contains the function ID of the guest's SMCCC call. Userspace is
6280 expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call
6281 parameters from the vCPU's GPRs.
6283 Definition of ``flags``:
6284 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC
6285 conduit to initiate the SMCCC call. If this bit is 0 then the guest
6286 used the HVC conduit for the SMCCC call.
6288 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit
6289 instruction to initiate the SMCCC call. If this bit is 0 then the
6290 guest used a 32bit instruction. An AArch64 guest always has this
6293 At the point of exit, PC points to the instruction immediately following
6294 the trapping instruction.
6298 /* KVM_EXIT_TPR_ACCESS */
6305 To be documented (KVM_TPR_ACCESS_REPORTING).
6309 /* KVM_EXIT_S390_SIEIC */
6312 __u64 mask; /* psw upper half */
6313 __u64 addr; /* psw lower half */
6322 /* KVM_EXIT_S390_RESET */
6323 #define KVM_S390_RESET_POR 1
6324 #define KVM_S390_RESET_CLEAR 2
6325 #define KVM_S390_RESET_SUBSYSTEM 4
6326 #define KVM_S390_RESET_CPU_INIT 8
6327 #define KVM_S390_RESET_IPL 16
6328 __u64 s390_reset_flags;
6334 /* KVM_EXIT_S390_UCONTROL */
6336 __u64 trans_exc_code;
6340 s390 specific. A page fault has occurred for a user controlled virtual
6341 machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be
6342 resolved by the kernel.
6343 The program code and the translation exception code that were placed
6344 in the cpu's lowcore are presented here as defined by the z Architecture
6345 Principles of Operation Book in the Chapter for Dynamic Address Translation
6357 Deprecated - was used for 440 KVM.
6366 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
6367 hypercalls and exit with this exit struct that contains all the guest gprs.
6369 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
6370 Userspace can now handle the hypercall and when it's done modify the gprs as
6371 necessary. Upon guest entry all guest GPRs will then be replaced by the values
6376 /* KVM_EXIT_PAPR_HCALL */
6383 This is used on 64-bit PowerPC when emulating a pSeries partition,
6384 e.g. with the 'pseries' machine type in qemu. It occurs when the
6385 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
6386 contains the hypercall number (from the guest R3), and 'args' contains
6387 the arguments (from the guest R4 - R12). Userspace should put the
6388 return code in 'ret' and any extra returned values in args[].
6389 The possible hypercalls are defined in the Power Architecture Platform
6390 Requirements (PAPR) document available from www.power.org (free
6391 developer registration required to access it).
6395 /* KVM_EXIT_S390_TSCH */
6397 __u16 subchannel_id;
6398 __u16 subchannel_nr;
6405 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
6406 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
6407 interrupt for the target subchannel has been dequeued and subchannel_id,
6408 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
6409 interrupt. ipb is needed for instruction parameter decoding.
6418 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
6419 interrupt acknowledge path to the core. When the core successfully
6420 delivers an interrupt, it automatically populates the EPR register with
6421 the interrupt vector number and acknowledges the interrupt inside
6422 the interrupt controller.
6424 In case the interrupt controller lives in user space, we need to do
6425 the interrupt acknowledge cycle through it to fetch the next to be
6426 delivered interrupt vector using this exit.
6428 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
6429 external interrupt has just been delivered into the guest. User space
6430 should put the acknowledged interrupt vector into the 'epr' field.
6434 /* KVM_EXIT_SYSTEM_EVENT */
6436 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
6437 #define KVM_SYSTEM_EVENT_RESET 2
6438 #define KVM_SYSTEM_EVENT_CRASH 3
6439 #define KVM_SYSTEM_EVENT_WAKEUP 4
6440 #define KVM_SYSTEM_EVENT_SUSPEND 5
6441 #define KVM_SYSTEM_EVENT_SEV_TERM 6
6447 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
6448 a system-level event using some architecture specific mechanism (hypercall
6449 or some special instruction). In case of ARM64, this is triggered using
6450 HVC instruction based PSCI call from the vcpu.
6452 The 'type' field describes the system-level event type.
6453 Valid values for 'type' are:
6455 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
6456 VM. Userspace is not obliged to honour this, and if it does honour
6457 this does not need to destroy the VM synchronously (ie it may call
6458 KVM_RUN again before shutdown finally occurs).
6459 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
6460 As with SHUTDOWN, userspace can choose to ignore the request, or
6461 to schedule the reset to occur in the future and may call KVM_RUN again.
6462 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
6463 has requested a crash condition maintenance. Userspace can choose
6464 to ignore the request, or to gather VM memory core dump and/or
6465 reset/shutdown of the VM.
6466 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination.
6467 The guest physical address of the guest's GHCB is stored in `data[0]`.
6468 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and
6469 KVM has recognized a wakeup event. Userspace may honor this event by
6470 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again.
6471 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of
6474 If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain
6475 architecture specific information for the system-level event. Only
6476 the first `ndata` items (possibly zero) of the data array are valid.
6478 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if
6479 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI
6482 - for RISC-V, data[0] is set to the value of the second argument of the
6483 ``sbi_system_reset`` call.
6485 Previous versions of Linux defined a `flags` member in this struct. The
6486 field is now aliased to `data[0]`. Userspace can assume that it is only
6487 written if ndata is greater than 0.
6492 KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the
6493 KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI
6494 SYSTEM_SUSPEND function, KVM will exit to userspace with this event
6497 It is the sole responsibility of userspace to implement the PSCI
6498 SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND".
6499 KVM does not change the vCPU's state before exiting to userspace, so
6500 the call parameters are left in-place in the vCPU registers.
6502 Userspace is _required_ to take action for such an exit. It must
6505 - Honor the guest request to suspend the VM. Userspace can request
6506 in-kernel emulation of suspension by setting the calling vCPU's
6507 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's
6508 state according to the parameters passed to the PSCI function when
6509 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use"
6510 for details on the function parameters.
6512 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2
6513 "Caller responsibilities" for possible return values.
6517 /* KVM_EXIT_IOAPIC_EOI */
6522 Indicates that the VCPU's in-kernel local APIC received an EOI for a
6523 level-triggered IOAPIC interrupt. This exit only triggers when the
6524 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
6525 the userspace IOAPIC should process the EOI and retrigger the interrupt if
6526 it is still asserted. Vector is the LAPIC interrupt vector for which the
6531 struct kvm_hyperv_exit {
6532 #define KVM_EXIT_HYPERV_SYNIC 1
6533 #define KVM_EXIT_HYPERV_HCALL 2
6534 #define KVM_EXIT_HYPERV_SYNDBG 3
6561 /* KVM_EXIT_HYPERV */
6562 struct kvm_hyperv_exit hyperv;
6564 Indicates that the VCPU exits into userspace to process some tasks
6565 related to Hyper-V emulation.
6567 Valid values for 'type' are:
6569 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
6571 Hyper-V SynIC state change. Notification is used to remap SynIC
6572 event/message pages and to enable/disable SynIC messages/events processing
6575 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
6577 Hyper-V Synthetic debugger state change. Notification is used to either update
6578 the pending_page location or to send a control command (send the buffer located
6579 in send_page or recv a buffer to recv_page).
6583 /* KVM_EXIT_ARM_NISV */
6589 Used on arm64 systems. If a guest accesses memory not in a memslot,
6590 KVM will typically return to userspace and ask it to do MMIO emulation on its
6591 behalf. However, for certain classes of instructions, no instruction decode
6592 (direction, length of memory access) is provided, and fetching and decoding
6593 the instruction from the VM is overly complicated to live in the kernel.
6595 Historically, when this situation occurred, KVM would print a warning and kill
6596 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
6597 trying to do I/O, which just couldn't be emulated, and the warning message was
6598 phrased accordingly. However, what happened more often was that a guest bug
6599 caused access outside the guest memory areas which should lead to a more
6600 meaningful warning message and an external abort in the guest, if the access
6601 did not fall within an I/O window.
6603 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
6604 this capability at VM creation. Once this is done, these types of errors will
6605 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
6606 the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field.
6607 Userspace can either fix up the access if it's actually an I/O access by
6608 decoding the instruction from guest memory (if it's very brave) and continue
6609 executing the guest, or it can decide to suspend, dump, or restart the guest.
6611 Note that KVM does not skip the faulting instruction as it does for
6612 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
6613 if it decides to decode and emulate the instruction.
6617 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
6619 __u8 error; /* user -> kernel */
6621 __u32 reason; /* kernel -> user */
6622 __u32 index; /* kernel -> user */
6623 __u64 data; /* kernel <-> user */
6626 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
6627 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
6628 may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
6631 The "reason" field specifies why the MSR interception occurred. Userspace will
6632 only receive MSR exits when a particular reason was requested during through
6633 ENABLE_CAP. Currently valid exit reasons are:
6635 ============================ ========================================
6636 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM
6637 KVM_MSR_EXIT_REASON_INVAL access to invalid MSRs or reserved bits
6638 KVM_MSR_EXIT_REASON_FILTER access blocked by KVM_X86_SET_MSR_FILTER
6639 ============================ ========================================
6641 For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest
6642 wants to read. To respond to this request with a successful read, userspace
6643 writes the respective data into the "data" field and must continue guest
6644 execution to ensure the read data is transferred into guest register state.
6646 If the RDMSR request was unsuccessful, userspace indicates that with a "1" in
6647 the "error" field. This will inject a #GP into the guest when the VCPU is
6650 For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest
6651 wants to write. Once finished processing the event, userspace must continue
6652 vCPU execution. If the MSR write was unsuccessful, userspace also sets the
6653 "error" field to "1".
6655 See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering.
6660 struct kvm_xen_exit {
6661 #define KVM_EXIT_XEN_HCALL 1
6674 struct kvm_hyperv_exit xen;
6676 Indicates that the VCPU exits into userspace to process some tasks
6677 related to Xen emulation.
6679 Valid values for 'type' are:
6681 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
6682 Userspace is expected to place the hypercall result into the appropriate
6683 field before invoking KVM_RUN again.
6687 /* KVM_EXIT_RISCV_SBI */
6689 unsigned long extension_id;
6690 unsigned long function_id;
6691 unsigned long args[6];
6692 unsigned long ret[2];
6695 If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has
6696 done a SBI call which is not handled by KVM RISC-V kernel module. The details
6697 of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The
6698 'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the
6699 'function_id' field represents function ID of given SBI extension. The 'args'
6700 array field of 'riscv_sbi' represents parameters for the SBI call and 'ret'
6701 array field represents return values. The userspace should update the return
6702 values of SBI call before resuming the VCPU. For more details on RISC-V SBI
6703 spec refer, https://github.com/riscv/riscv-sbi-doc.
6707 /* KVM_EXIT_NOTIFY */
6709 #define KVM_NOTIFY_CONTEXT_INVALID (1 << 0)
6713 Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is
6714 enabled, a VM exit generated if no event window occurs in VM non-root mode
6715 for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when
6716 enabling the cap, it would exit to userspace with the exit reason
6717 KVM_EXIT_NOTIFY for further handling. The "flags" field contains more
6720 The valid value for 'flags' is:
6722 - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid
6723 in VMCS. It would run into unknown result if resume the target VM.
6727 /* Fix the size of the union. */
6732 * shared registers between kvm and userspace.
6733 * kvm_valid_regs specifies the register classes set by the host
6734 * kvm_dirty_regs specified the register classes dirtied by userspace
6735 * struct kvm_sync_regs is architecture specific, as well as the
6736 * bits for kvm_valid_regs and kvm_dirty_regs
6738 __u64 kvm_valid_regs;
6739 __u64 kvm_dirty_regs;
6741 struct kvm_sync_regs regs;
6742 char padding[SYNC_REGS_SIZE_BYTES];
6745 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
6746 certain guest registers without having to call SET/GET_*REGS. Thus we can
6747 avoid some system call overhead if userspace has to handle the exit.
6748 Userspace can query the validity of the structure by checking
6749 kvm_valid_regs for specific bits. These bits are architecture specific
6750 and usually define the validity of a groups of registers. (e.g. one bit
6751 for general purpose registers)
6753 Please note that the kernel is allowed to use the kvm_run structure as the
6754 primary storage for certain register types. Therefore, the kernel may use the
6755 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
6758 6. Capabilities that can be enabled on vCPUs
6759 ============================================
6761 There are certain capabilities that change the behavior of the virtual CPU or
6762 the virtual machine when enabled. To enable them, please see section 4.37.
6763 Below you can find a list of capabilities and what their effect on the vCPU or
6764 the virtual machine is when enabling them.
6766 The following information is provided along with the description:
6769 which instruction set architectures provide this ioctl.
6770 x86 includes both i386 and x86_64.
6773 whether this is a per-vcpu or per-vm capability.
6776 what parameters are accepted by the capability.
6779 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6780 are not detailed, but errors with specific meanings are.
6789 :Returns: 0 on success; -1 on error
6791 This capability enables interception of OSI hypercalls that otherwise would
6792 be treated as normal system calls to be injected into the guest. OSI hypercalls
6793 were invented by Mac-on-Linux to have a standardized communication mechanism
6794 between the guest and the host.
6796 When this capability is enabled, KVM_EXIT_OSI can occur.
6799 6.2 KVM_CAP_PPC_PAPR
6800 --------------------
6805 :Returns: 0 on success; -1 on error
6807 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
6808 done using the hypercall instruction "sc 1".
6810 It also sets the guest privilege level to "supervisor" mode. Usually the guest
6811 runs in "hypervisor" privilege mode with a few missing features.
6813 In addition to the above, it changes the semantics of SDR1. In this mode, the
6814 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
6815 HTAB invisible to the guest.
6817 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
6825 :Parameters: args[0] is the address of a struct kvm_config_tlb
6826 :Returns: 0 on success; -1 on error
6830 struct kvm_config_tlb {
6837 Configures the virtual CPU's TLB array, establishing a shared memory area
6838 between userspace and KVM. The "params" and "array" fields are userspace
6839 addresses of mmu-type-specific data structures. The "array_len" field is an
6840 safety mechanism, and should be set to the size in bytes of the memory that
6841 userspace has reserved for the array. It must be at least the size dictated
6842 by "mmu_type" and "params".
6844 While KVM_RUN is active, the shared region is under control of KVM. Its
6845 contents are undefined, and any modification by userspace results in
6846 boundedly undefined behavior.
6848 On return from KVM_RUN, the shared region will reflect the current state of
6849 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
6850 to tell KVM which entries have been changed, prior to calling KVM_RUN again
6853 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
6855 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
6856 - The "array" field points to an array of type "struct
6857 kvm_book3e_206_tlb_entry".
6858 - The array consists of all entries in the first TLB, followed by all
6859 entries in the second TLB.
6860 - Within a TLB, entries are ordered first by increasing set number. Within a
6861 set, entries are ordered by way (increasing ESEL).
6862 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
6863 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
6864 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
6865 hardware ignores this value for TLB0.
6867 6.4 KVM_CAP_S390_CSS_SUPPORT
6868 ----------------------------
6870 :Architectures: s390
6873 :Returns: 0 on success; -1 on error
6875 This capability enables support for handling of channel I/O instructions.
6877 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
6878 handled in-kernel, while the other I/O instructions are passed to userspace.
6880 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
6881 SUBCHANNEL intercepts.
6883 Note that even though this capability is enabled per-vcpu, the complete
6884 virtual machine is affected.
6891 :Parameters: args[0] defines whether the proxy facility is active
6892 :Returns: 0 on success; -1 on error
6894 This capability enables or disables the delivery of interrupts through the
6895 external proxy facility.
6897 When enabled (args[0] != 0), every time the guest gets an external interrupt
6898 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
6899 to receive the topmost interrupt vector.
6901 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
6903 When this capability is enabled, KVM_EXIT_EPR can occur.
6905 6.6 KVM_CAP_IRQ_MPIC
6906 --------------------
6909 :Parameters: args[0] is the MPIC device fd;
6910 args[1] is the MPIC CPU number for this vcpu
6912 This capability connects the vcpu to an in-kernel MPIC device.
6914 6.7 KVM_CAP_IRQ_XICS
6915 --------------------
6919 :Parameters: args[0] is the XICS device fd;
6920 args[1] is the XICS CPU number (server ID) for this vcpu
6922 This capability connects the vcpu to an in-kernel XICS device.
6924 6.8 KVM_CAP_S390_IRQCHIP
6925 ------------------------
6927 :Architectures: s390
6931 This capability enables the in-kernel irqchip for s390. Please refer to
6932 "4.24 KVM_CREATE_IRQCHIP" for details.
6934 6.9 KVM_CAP_MIPS_FPU
6935 --------------------
6937 :Architectures: mips
6939 :Parameters: args[0] is reserved for future use (should be 0).
6941 This capability allows the use of the host Floating Point Unit by the guest. It
6942 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
6943 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
6944 accessed (depending on the current guest FPU register mode), and the Status.FR,
6945 Config5.FRE bits are accessible via the KVM API and also from the guest,
6946 depending on them being supported by the FPU.
6948 6.10 KVM_CAP_MIPS_MSA
6949 ---------------------
6951 :Architectures: mips
6953 :Parameters: args[0] is reserved for future use (should be 0).
6955 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
6956 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
6957 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
6958 registers can be accessed, and the Config5.MSAEn bit is accessible via the
6959 KVM API and also from the guest.
6961 6.74 KVM_CAP_SYNC_REGS
6962 ----------------------
6964 :Architectures: s390, x86
6965 :Target: s390: always enabled, x86: vcpu
6967 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
6969 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
6971 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
6972 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
6973 without having to call SET/GET_*REGS". This reduces overhead by eliminating
6974 repeated ioctl calls for setting and/or getting register values. This is
6975 particularly important when userspace is making synchronous guest state
6976 modifications, e.g. when emulating and/or intercepting instructions in
6979 For s390 specifics, please refer to the source code.
6983 - the register sets to be copied out to kvm_run are selectable
6984 by userspace (rather that all sets being copied out for every exit).
6985 - vcpu_events are available in addition to regs and sregs.
6987 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
6988 function as an input bit-array field set by userspace to indicate the
6989 specific register sets to be copied out on the next exit.
6991 To indicate when userspace has modified values that should be copied into
6992 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
6993 This is done using the same bitflags as for the 'kvm_valid_regs' field.
6994 If the dirty bit is not set, then the register set values will not be copied
6995 into the vCPU even if they've been modified.
6997 Unused bitfields in the bitarrays must be set to zero.
7001 struct kvm_sync_regs {
7002 struct kvm_regs regs;
7003 struct kvm_sregs sregs;
7004 struct kvm_vcpu_events events;
7007 6.75 KVM_CAP_PPC_IRQ_XIVE
7008 -------------------------
7012 :Parameters: args[0] is the XIVE device fd;
7013 args[1] is the XIVE CPU number (server ID) for this vcpu
7015 This capability connects the vcpu to an in-kernel XIVE device.
7017 7. Capabilities that can be enabled on VMs
7018 ==========================================
7020 There are certain capabilities that change the behavior of the virtual
7021 machine when enabled. To enable them, please see section 4.37. Below
7022 you can find a list of capabilities and what their effect on the VM
7023 is when enabling them.
7025 The following information is provided along with the description:
7028 which instruction set architectures provide this ioctl.
7029 x86 includes both i386 and x86_64.
7032 what parameters are accepted by the capability.
7035 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
7036 are not detailed, but errors with specific meanings are.
7039 7.1 KVM_CAP_PPC_ENABLE_HCALL
7040 ----------------------------
7043 :Parameters: args[0] is the sPAPR hcall number;
7044 args[1] is 0 to disable, 1 to enable in-kernel handling
7046 This capability controls whether individual sPAPR hypercalls (hcalls)
7047 get handled by the kernel or not. Enabling or disabling in-kernel
7048 handling of an hcall is effective across the VM. On creation, an
7049 initial set of hcalls are enabled for in-kernel handling, which
7050 consists of those hcalls for which in-kernel handlers were implemented
7051 before this capability was implemented. If disabled, the kernel will
7052 not to attempt to handle the hcall, but will always exit to userspace
7053 to handle it. Note that it may not make sense to enable some and
7054 disable others of a group of related hcalls, but KVM does not prevent
7055 userspace from doing that.
7057 If the hcall number specified is not one that has an in-kernel
7058 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
7061 7.2 KVM_CAP_S390_USER_SIGP
7062 --------------------------
7064 :Architectures: s390
7067 This capability controls which SIGP orders will be handled completely in user
7068 space. With this capability enabled, all fast orders will be handled completely
7075 - CONDITIONAL EMERGENCY SIGNAL
7077 All other orders will be handled completely in user space.
7079 Only privileged operation exceptions will be checked for in the kernel (or even
7080 in the hardware prior to interception). If this capability is not enabled, the
7081 old way of handling SIGP orders is used (partially in kernel and user space).
7083 7.3 KVM_CAP_S390_VECTOR_REGISTERS
7084 ---------------------------------
7086 :Architectures: s390
7088 :Returns: 0 on success, negative value on error
7090 Allows use of the vector registers introduced with z13 processor, and
7091 provides for the synchronization between host and user space. Will
7092 return -EINVAL if the machine does not support vectors.
7094 7.4 KVM_CAP_S390_USER_STSI
7095 --------------------------
7097 :Architectures: s390
7100 This capability allows post-handlers for the STSI instruction. After
7101 initial handling in the kernel, KVM exits to user space with
7102 KVM_EXIT_S390_STSI to allow user space to insert further data.
7104 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
7116 @addr - guest address of STSI SYSIB
7120 @ar - access register number
7122 KVM handlers should exit to userspace with rc = -EREMOTE.
7124 7.5 KVM_CAP_SPLIT_IRQCHIP
7125 -------------------------
7128 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
7129 :Returns: 0 on success, -1 on error
7131 Create a local apic for each processor in the kernel. This can be used
7132 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
7133 IOAPIC and PIC (and also the PIT, even though this has to be enabled
7136 This capability also enables in kernel routing of interrupt requests;
7137 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
7138 used in the IRQ routing table. The first args[0] MSI routes are reserved
7139 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
7140 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
7142 Fails if VCPU has already been created, or if the irqchip is already in the
7143 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
7148 :Architectures: s390
7151 Allows use of runtime-instrumentation introduced with zEC12 processor.
7152 Will return -EINVAL if the machine does not support runtime-instrumentation.
7153 Will return -EBUSY if a VCPU has already been created.
7155 7.7 KVM_CAP_X2APIC_API
7156 ----------------------
7159 :Parameters: args[0] - features that should be enabled
7160 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
7162 Valid feature flags in args[0] are::
7164 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
7165 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
7167 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
7168 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
7169 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
7170 respective sections.
7172 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
7173 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
7174 as a broadcast even in x2APIC mode in order to support physical x2APIC
7175 without interrupt remapping. This is undesirable in logical mode,
7176 where 0xff represents CPUs 0-7 in cluster 0.
7178 7.8 KVM_CAP_S390_USER_INSTR0
7179 ----------------------------
7181 :Architectures: s390
7184 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
7185 be intercepted and forwarded to user space. User space can use this
7186 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
7187 not inject an operating exception for these instructions, user space has
7188 to take care of that.
7190 This capability can be enabled dynamically even if VCPUs were already
7191 created and are running.
7196 :Architectures: s390
7198 :Returns: 0 on success; -EINVAL if the machine does not support
7199 guarded storage; -EBUSY if a VCPU has already been created.
7201 Allows use of guarded storage for the KVM guest.
7203 7.10 KVM_CAP_S390_AIS
7204 ---------------------
7206 :Architectures: s390
7209 Allow use of adapter-interruption suppression.
7210 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
7212 7.11 KVM_CAP_PPC_SMT
7213 --------------------
7216 :Parameters: vsmt_mode, flags
7218 Enabling this capability on a VM provides userspace with a way to set
7219 the desired virtual SMT mode (i.e. the number of virtual CPUs per
7220 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
7221 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
7222 the number of threads per subcore for the host. Currently flags must
7223 be 0. A successful call to enable this capability will result in
7224 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
7225 subsequently queried for the VM. This capability is only supported by
7226 HV KVM, and can only be set before any VCPUs have been created.
7227 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
7228 modes are available.
7230 7.12 KVM_CAP_PPC_FWNMI
7231 ----------------------
7236 With this capability a machine check exception in the guest address
7237 space will cause KVM to exit the guest with NMI exit reason. This
7238 enables QEMU to build error log and branch to guest kernel registered
7239 machine check handling routine. Without this capability KVM will
7240 branch to guests' 0x200 interrupt vector.
7242 7.13 KVM_CAP_X86_DISABLE_EXITS
7243 ------------------------------
7246 :Parameters: args[0] defines which exits are disabled
7247 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
7249 Valid bits in args[0] are::
7251 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
7252 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
7253 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
7254 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
7256 Enabling this capability on a VM provides userspace with a way to no
7257 longer intercept some instructions for improved latency in some
7258 workloads, and is suggested when vCPUs are associated to dedicated
7259 physical CPUs. More bits can be added in the future; userspace can
7260 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
7263 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
7265 7.14 KVM_CAP_S390_HPAGE_1M
7266 --------------------------
7268 :Architectures: s390
7270 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
7271 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
7274 With this capability the KVM support for memory backing with 1m pages
7275 through hugetlbfs can be enabled for a VM. After the capability is
7276 enabled, cmma can't be enabled anymore and pfmfi and the storage key
7277 interpretation are disabled. If cmma has already been enabled or the
7278 hpage module parameter is not set to 1, -EINVAL is returned.
7280 While it is generally possible to create a huge page backed VM without
7281 this capability, the VM will not be able to run.
7283 7.15 KVM_CAP_MSR_PLATFORM_INFO
7284 ------------------------------
7287 :Parameters: args[0] whether feature should be enabled or not
7289 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
7290 a #GP would be raised when the guest tries to access. Currently, this
7291 capability does not enable write permissions of this MSR for the guest.
7293 7.16 KVM_CAP_PPC_NESTED_HV
7294 --------------------------
7298 :Returns: 0 on success, -EINVAL when the implementation doesn't support
7299 nested-HV virtualization.
7301 HV-KVM on POWER9 and later systems allows for "nested-HV"
7302 virtualization, which provides a way for a guest VM to run guests that
7303 can run using the CPU's supervisor mode (privileged non-hypervisor
7304 state). Enabling this capability on a VM depends on the CPU having
7305 the necessary functionality and on the facility being enabled with a
7306 kvm-hv module parameter.
7308 7.17 KVM_CAP_EXCEPTION_PAYLOAD
7309 ------------------------------
7312 :Parameters: args[0] whether feature should be enabled or not
7314 With this capability enabled, CR2 will not be modified prior to the
7315 emulated VM-exit when L1 intercepts a #PF exception that occurs in
7316 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
7317 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
7318 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
7319 #DB) exception for L2, exception.has_payload will be set and the
7320 faulting address (or the new DR6 bits*) will be reported in the
7321 exception_payload field. Similarly, when userspace injects a #PF (or
7322 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
7323 exception.has_payload and to put the faulting address - or the new DR6
7324 bits\ [#]_ - in the exception_payload field.
7326 This capability also enables exception.pending in struct
7327 kvm_vcpu_events, which allows userspace to distinguish between pending
7328 and injected exceptions.
7331 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
7334 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
7335 --------------------------------------
7337 :Architectures: x86, arm64, mips
7338 :Parameters: args[0] whether feature should be enabled or not
7342 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
7343 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
7345 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
7346 automatically clear and write-protect all pages that are returned as dirty.
7347 Rather, userspace will have to do this operation separately using
7348 KVM_CLEAR_DIRTY_LOG.
7350 At the cost of a slightly more complicated operation, this provides better
7351 scalability and responsiveness for two reasons. First,
7352 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
7353 than requiring to sync a full memslot; this ensures that KVM does not
7354 take spinlocks for an extended period of time. Second, in some cases a
7355 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
7356 userspace actually using the data in the page. Pages can be modified
7357 during this time, which is inefficient for both the guest and userspace:
7358 the guest will incur a higher penalty due to write protection faults,
7359 while userspace can see false reports of dirty pages. Manual reprotection
7360 helps reducing this time, improving guest performance and reducing the
7361 number of dirty log false positives.
7363 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
7364 will be initialized to 1 when created. This also improves performance because
7365 dirty logging can be enabled gradually in small chunks on the first call
7366 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
7367 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
7368 x86 and arm64 for now).
7370 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
7371 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
7372 it hard or impossible to use it correctly. The availability of
7373 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
7374 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
7376 7.19 KVM_CAP_PPC_SECURE_GUEST
7377 ------------------------------
7381 This capability indicates that KVM is running on a host that has
7382 ultravisor firmware and thus can support a secure guest. On such a
7383 system, a guest can ask the ultravisor to make it a secure guest,
7384 one whose memory is inaccessible to the host except for pages which
7385 are explicitly requested to be shared with the host. The ultravisor
7386 notifies KVM when a guest requests to become a secure guest, and KVM
7387 has the opportunity to veto the transition.
7389 If present, this capability can be enabled for a VM, meaning that KVM
7390 will allow the transition to secure guest mode. Otherwise KVM will
7391 veto the transition.
7393 7.20 KVM_CAP_HALT_POLL
7394 ----------------------
7398 :Parameters: args[0] is the maximum poll time in nanoseconds
7399 :Returns: 0 on success; -1 on error
7401 KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the
7402 maximum halt-polling time for all vCPUs in the target VM. This capability can
7403 be invoked at any time and any number of times to dynamically change the
7404 maximum halt-polling time.
7406 See Documentation/virt/kvm/halt-polling.rst for more information on halt
7409 7.21 KVM_CAP_X86_USER_SPACE_MSR
7410 -------------------------------
7414 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
7415 :Returns: 0 on success; -1 on error
7417 This capability allows userspace to intercept RDMSR and WRMSR instructions if
7418 access to an MSR is denied. By default, KVM injects #GP on denied accesses.
7420 When a guest requests to read or write an MSR, KVM may not implement all MSRs
7421 that are relevant to a respective system. It also does not differentiate by
7424 To allow more fine grained control over MSR handling, userspace may enable
7425 this capability. With it enabled, MSR accesses that match the mask specified in
7426 args[0] and would trigger a #GP inside the guest will instead trigger
7427 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications. Userspace
7428 can then implement model specific MSR handling and/or user notifications
7429 to inform a user that an MSR was not emulated/virtualized by KVM.
7431 The valid mask flags are:
7433 ============================ ===============================================
7434 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs
7435 KVM_MSR_EXIT_REASON_INVAL intercept accesses that are architecturally
7436 invalid according to the vCPU model and/or mode
7437 KVM_MSR_EXIT_REASON_FILTER intercept accesses that are denied by userspace
7438 via KVM_X86_SET_MSR_FILTER
7439 ============================ ===============================================
7441 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
7442 -------------------------------
7446 :Parameters: args[0] defines the policy used when bus locks detected in guest
7447 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
7449 Valid bits in args[0] are::
7451 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
7452 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
7454 Enabling this capability on a VM provides userspace with a way to select
7455 a policy to handle the bus locks detected in guest. Userspace can obtain
7456 the supported modes from the result of KVM_CHECK_EXTENSION and define it
7457 through the KVM_ENABLE_CAP.
7459 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
7460 currently and mutually exclusive with each other. More bits can be added in
7463 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
7464 so that no additional actions are needed. This is the default mode.
7466 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
7467 in VM. KVM just exits to userspace when handling them. Userspace can enforce
7468 its own throttling or other policy based mitigations.
7470 This capability is aimed to address the thread that VM can exploit bus locks to
7471 degree the performance of the whole system. Once the userspace enable this
7472 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
7473 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
7474 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
7475 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
7476 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
7478 7.23 KVM_CAP_PPC_DAWR1
7479 ----------------------
7483 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
7485 This capability can be used to check / enable 2nd DAWR feature provided
7486 by POWER10 processor.
7489 7.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
7490 -------------------------------------
7492 Architectures: x86 SEV enabled
7494 Parameters: args[0] is the fd of the source vm
7495 Returns: 0 on success; ENOTTY on error
7497 This capability enables userspace to copy encryption context from the vm
7498 indicated by the fd to the vm this is called on.
7500 This is intended to support in-guest workloads scheduled by the host. This
7501 allows the in-guest workload to maintain its own NPTs and keeps the two vms
7502 from accidentally clobbering each other with interrupts and the like (separate
7505 7.25 KVM_CAP_SGX_ATTRIBUTE
7506 --------------------------
7510 :Parameters: args[0] is a file handle of a SGX attribute file in securityfs
7511 :Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
7512 attribute is not supported by KVM.
7514 KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
7515 more privileged enclave attributes. args[0] must hold a file handle to a valid
7516 SGX attribute file corresponding to an attribute that is supported/restricted
7517 by KVM (currently only PROVISIONKEY).
7519 The SGX subsystem restricts access to a subset of enclave attributes to provide
7520 additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
7521 is restricted to deter malware from using the PROVISIONKEY to obtain a stable
7522 system fingerprint. To prevent userspace from circumventing such restrictions
7523 by running an enclave in a VM, KVM prevents access to privileged attributes by
7526 See Documentation/arch/x86/sgx.rst for more details.
7528 7.26 KVM_CAP_PPC_RPT_INVALIDATE
7529 -------------------------------
7531 :Capability: KVM_CAP_PPC_RPT_INVALIDATE
7535 This capability indicates that the kernel is capable of handling
7536 H_RPT_INVALIDATE hcall.
7538 In order to enable the use of H_RPT_INVALIDATE in the guest,
7539 user space might have to advertise it for the guest. For example,
7540 IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
7541 present in the "ibm,hypertas-functions" device-tree property.
7543 This capability is enabled for hypervisors on platforms like POWER9
7544 that support radix MMU.
7546 7.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
7547 --------------------------------------
7550 :Parameters: args[0] whether the feature should be enabled or not
7552 When this capability is enabled, an emulation failure will result in an exit
7553 to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
7554 to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
7555 to 15 instruction bytes for any exit to userspace resulting from an emulation
7556 failure. When these exits to userspace occur use the emulation_failure struct
7557 instead of the internal struct. They both have the same layout, but the
7558 emulation_failure struct matches the content better. It also explicitly
7559 defines the 'flags' field which is used to describe the fields in the struct
7560 that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
7561 set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
7564 7.28 KVM_CAP_ARM_MTE
7565 --------------------
7567 :Architectures: arm64
7570 This capability indicates that KVM (and the hardware) supports exposing the
7571 Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
7572 VMM before creating any VCPUs to allow the guest access. Note that MTE is only
7573 available to a guest running in AArch64 mode and enabling this capability will
7574 cause attempts to create AArch32 VCPUs to fail.
7576 When enabled the guest is able to access tags associated with any memory given
7577 to the guest. KVM will ensure that the tags are maintained during swap or
7578 hibernation of the host; however the VMM needs to manually save/restore the
7579 tags as appropriate if the VM is migrated.
7581 When this capability is enabled all memory in memslots must be mapped as
7582 ``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``),
7583 attempts to create a memslot with an invalid mmap will result in an
7586 When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
7587 perform a bulk copy of tags to/from the guest.
7589 7.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM
7590 -------------------------------------
7592 Architectures: x86 SEV enabled
7594 Parameters: args[0] is the fd of the source vm
7595 Returns: 0 on success
7597 This capability enables userspace to migrate the encryption context from the VM
7598 indicated by the fd to the VM this is called on.
7600 This is intended to support intra-host migration of VMs between userspace VMMs,
7601 upgrading the VMM process without interrupting the guest.
7603 7.30 KVM_CAP_PPC_AIL_MODE_3
7604 -------------------------------
7606 :Capability: KVM_CAP_PPC_AIL_MODE_3
7610 This capability indicates that the kernel supports the mode 3 setting for the
7611 "Address Translation Mode on Interrupt" aka "Alternate Interrupt Location"
7612 resource that is controlled with the H_SET_MODE hypercall.
7614 This capability allows a guest kernel to use a better-performance mode for
7615 handling interrupts and system calls.
7617 7.31 KVM_CAP_DISABLE_QUIRKS2
7618 ----------------------------
7620 :Capability: KVM_CAP_DISABLE_QUIRKS2
7621 :Parameters: args[0] - set of KVM quirks to disable
7625 This capability, if enabled, will cause KVM to disable some behavior
7628 Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
7629 quirks that can be disabled in KVM.
7631 The argument to KVM_ENABLE_CAP for this capability is a bitmask of
7632 quirks to disable, and must be a subset of the bitmask returned by
7633 KVM_CHECK_EXTENSION.
7635 The valid bits in cap.args[0] are:
7637 =================================== ============================================
7638 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT
7639 LINT0 register is 0x700 (APIC_MODE_EXTINT).
7640 When this quirk is disabled, the reset value
7641 is 0x10000 (APIC_LVT_MASKED).
7643 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW.
7644 When this quirk is disabled, KVM does not
7645 change the value of CR0.CD and CR0.NW.
7647 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is
7648 available even when configured for x2APIC
7649 mode. When this quirk is disabled, KVM
7650 disables the MMIO LAPIC interface if the
7651 LAPIC is in x2APIC mode.
7653 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before
7654 exiting to userspace for an OUT instruction
7655 to port 0x7e. When this quirk is disabled,
7656 KVM does not pre-increment %rip before
7657 exiting to userspace.
7659 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets
7660 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if
7661 IA32_MISC_ENABLE[bit 18] (MWAIT) is set.
7662 Additionally, when this quirk is disabled,
7663 KVM clears CPUID.01H:ECX[bit 3] if
7664 IA32_MISC_ENABLE[bit 18] is cleared.
7666 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest
7667 VMMCALL/VMCALL instructions to match the
7668 vendor's hypercall instruction for the
7669 system. When this quirk is disabled, KVM
7670 will no longer rewrite invalid guest
7671 hypercall instructions. Executing the
7672 incorrect hypercall instruction will
7673 generate a #UD within the guest.
7675 KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if
7676 they are intercepted) as NOPs regardless of
7677 whether or not MONITOR/MWAIT are supported
7678 according to guest CPUID. When this quirk
7679 is disabled and KVM_X86_DISABLE_EXITS_MWAIT
7680 is not set (MONITOR/MWAIT are intercepted),
7681 KVM will inject a #UD on MONITOR/MWAIT if
7682 they're unsupported per guest CPUID. Note,
7683 KVM will modify MONITOR/MWAIT support in
7684 guest CPUID on writes to MISC_ENABLE if
7685 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is
7687 =================================== ============================================
7689 7.32 KVM_CAP_MAX_VCPU_ID
7690 ------------------------
7694 :Parameters: args[0] - maximum APIC ID value set for current VM
7695 :Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS
7696 supported in KVM or if it has been set.
7698 This capability allows userspace to specify maximum possible APIC ID
7699 assigned for current VM session prior to the creation of vCPUs, saving
7700 memory for data structures indexed by the APIC ID. Userspace is able
7701 to calculate the limit to APIC ID values from designated
7704 The value can be changed only until KVM_ENABLE_CAP is set to a nonzero
7705 value or until a vCPU is created. Upon creation of the first vCPU,
7706 if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM
7707 uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as
7708 the maximum APIC ID.
7710 7.33 KVM_CAP_X86_NOTIFY_VMEXIT
7711 ------------------------------
7715 :Parameters: args[0] is the value of notify window as well as some flags
7716 :Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify
7717 VM exit is unsupported.
7719 Bits 63:32 of args[0] are used for notify window.
7720 Bits 31:0 of args[0] are for some flags. Valid bits are::
7722 #define KVM_X86_NOTIFY_VMEXIT_ENABLED (1 << 0)
7723 #define KVM_X86_NOTIFY_VMEXIT_USER (1 << 1)
7725 This capability allows userspace to configure the notify VM exit on/off
7726 in per-VM scope during VM creation. Notify VM exit is disabled by default.
7727 When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will
7728 enable this feature with the notify window provided, which will generate
7729 a VM exit if no event window occurs in VM non-root mode for a specified of
7730 time (notify window).
7732 If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen,
7733 KVM would exit to userspace for handling.
7735 This capability is aimed to mitigate the threat that malicious VMs can
7736 cause CPU stuck (due to event windows don't open up) and make the CPU
7737 unavailable to host or other VMs.
7739 8. Other capabilities.
7740 ======================
7742 This section lists capabilities that give information about other
7743 features of the KVM implementation.
7745 8.1 KVM_CAP_PPC_HWRNG
7746 ---------------------
7750 This capability, if KVM_CHECK_EXTENSION indicates that it is
7751 available, means that the kernel has an implementation of the
7752 H_RANDOM hypercall backed by a hardware random-number generator.
7753 If present, the kernel H_RANDOM handler can be enabled for guest use
7754 with the KVM_CAP_PPC_ENABLE_HCALL capability.
7756 8.2 KVM_CAP_HYPERV_SYNIC
7757 ------------------------
7761 This capability, if KVM_CHECK_EXTENSION indicates that it is
7762 available, means that the kernel has an implementation of the
7763 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
7764 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
7766 In order to use SynIC, it has to be activated by setting this
7767 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
7768 will disable the use of APIC hardware virtualization even if supported
7769 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
7771 8.3 KVM_CAP_PPC_RADIX_MMU
7772 -------------------------
7776 This capability, if KVM_CHECK_EXTENSION indicates that it is
7777 available, means that the kernel can support guests using the
7778 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
7781 8.4 KVM_CAP_PPC_HASH_MMU_V3
7782 ---------------------------
7786 This capability, if KVM_CHECK_EXTENSION indicates that it is
7787 available, means that the kernel can support guests using the
7788 hashed page table MMU defined in Power ISA V3.00 (as implemented in
7789 the POWER9 processor), including in-memory segment tables.
7794 :Architectures: mips
7796 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7797 it is available, means that full hardware assisted virtualization capabilities
7798 of the hardware are available for use through KVM. An appropriate
7799 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
7802 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7803 available, it means that the VM is using full hardware assisted virtualization
7804 capabilities of the hardware. This is useful to check after creating a VM with
7805 KVM_VM_MIPS_DEFAULT.
7807 The value returned by KVM_CHECK_EXTENSION should be compared against known
7808 values (see below). All other values are reserved. This is to allow for the
7809 possibility of other hardware assisted virtualization implementations which
7810 may be incompatible with the MIPS VZ ASE.
7812 == ==========================================================================
7813 0 The trap & emulate implementation is in use to run guest code in user
7814 mode. Guest virtual memory segments are rearranged to fit the guest in the
7815 user mode address space.
7817 1 The MIPS VZ ASE is in use, providing full hardware assisted
7818 virtualization, including standard guest virtual memory segments.
7819 == ==========================================================================
7824 :Architectures: mips
7826 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7827 it is available, means that the trap & emulate implementation is available to
7828 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
7829 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
7830 to KVM_CREATE_VM to create a VM which utilises it.
7832 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7833 available, it means that the VM is using trap & emulate.
7835 8.7 KVM_CAP_MIPS_64BIT
7836 ----------------------
7838 :Architectures: mips
7840 This capability indicates the supported architecture type of the guest, i.e. the
7841 supported register and address width.
7843 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
7844 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
7845 be checked specifically against known values (see below). All other values are
7848 == ========================================================================
7849 0 MIPS32 or microMIPS32.
7850 Both registers and addresses are 32-bits wide.
7851 It will only be possible to run 32-bit guest code.
7853 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
7854 Registers are 64-bits wide, but addresses are 32-bits wide.
7855 64-bit guest code may run but cannot access MIPS64 memory segments.
7856 It will also be possible to run 32-bit guest code.
7858 2 MIPS64 or microMIPS64 with access to all address segments.
7859 Both registers and addresses are 64-bits wide.
7860 It will be possible to run 64-bit or 32-bit guest code.
7861 == ========================================================================
7863 8.9 KVM_CAP_ARM_USER_IRQ
7864 ------------------------
7866 :Architectures: arm64
7868 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
7869 that if userspace creates a VM without an in-kernel interrupt controller, it
7870 will be notified of changes to the output level of in-kernel emulated devices,
7871 which can generate virtual interrupts, presented to the VM.
7872 For such VMs, on every return to userspace, the kernel
7873 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
7874 output level of the device.
7876 Whenever kvm detects a change in the device output level, kvm guarantees at
7877 least one return to userspace before running the VM. This exit could either
7878 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
7879 userspace can always sample the device output level and re-compute the state of
7880 the userspace interrupt controller. Userspace should always check the state
7881 of run->s.regs.device_irq_level on every kvm exit.
7882 The value in run->s.regs.device_irq_level can represent both level and edge
7883 triggered interrupt signals, depending on the device. Edge triggered interrupt
7884 signals will exit to userspace with the bit in run->s.regs.device_irq_level
7885 set exactly once per edge signal.
7887 The field run->s.regs.device_irq_level is available independent of
7888 run->kvm_valid_regs or run->kvm_dirty_regs bits.
7890 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
7891 number larger than 0 indicating the version of this capability is implemented
7892 and thereby which bits in run->s.regs.device_irq_level can signal values.
7894 Currently the following bits are defined for the device_irq_level bitmap::
7896 KVM_CAP_ARM_USER_IRQ >= 1:
7898 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
7899 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
7900 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
7902 Future versions of kvm may implement additional events. These will get
7903 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
7906 8.10 KVM_CAP_PPC_SMT_POSSIBLE
7907 -----------------------------
7911 Querying this capability returns a bitmap indicating the possible
7912 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
7913 (counting from the right) is set, then a virtual SMT mode of 2^N is
7916 8.11 KVM_CAP_HYPERV_SYNIC2
7917 --------------------------
7921 This capability enables a newer version of Hyper-V Synthetic interrupt
7922 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
7923 doesn't clear SynIC message and event flags pages when they are enabled by
7924 writing to the respective MSRs.
7926 8.12 KVM_CAP_HYPERV_VP_INDEX
7927 ----------------------------
7931 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
7932 value is used to denote the target vcpu for a SynIC interrupt. For
7933 compatibility, KVM initializes this msr to KVM's internal vcpu index. When this
7934 capability is absent, userspace can still query this msr's value.
7936 8.13 KVM_CAP_S390_AIS_MIGRATION
7937 -------------------------------
7939 :Architectures: s390
7942 This capability indicates if the flic device will be able to get/set the
7943 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
7944 to discover this without having to create a flic device.
7946 8.14 KVM_CAP_S390_PSW
7947 ---------------------
7949 :Architectures: s390
7951 This capability indicates that the PSW is exposed via the kvm_run structure.
7953 8.15 KVM_CAP_S390_GMAP
7954 ----------------------
7956 :Architectures: s390
7958 This capability indicates that the user space memory used as guest mapping can
7959 be anywhere in the user memory address space, as long as the memory slots are
7960 aligned and sized to a segment (1MB) boundary.
7962 8.16 KVM_CAP_S390_COW
7963 ---------------------
7965 :Architectures: s390
7967 This capability indicates that the user space memory used as guest mapping can
7968 use copy-on-write semantics as well as dirty pages tracking via read-only page
7971 8.17 KVM_CAP_S390_BPB
7972 ---------------------
7974 :Architectures: s390
7976 This capability indicates that kvm will implement the interfaces to handle
7977 reset, migration and nested KVM for branch prediction blocking. The stfle
7978 facility 82 should not be provided to the guest without this capability.
7980 8.18 KVM_CAP_HYPERV_TLBFLUSH
7981 ----------------------------
7985 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
7987 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
7988 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
7990 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
7991 ----------------------------------
7993 :Architectures: arm64
7995 This capability indicates that userspace can specify (via the
7996 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
7997 takes a virtual SError interrupt exception.
7998 If KVM advertises this capability, userspace can only specify the ISS field for
7999 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
8000 CPU when the exception is taken. If this virtual SError is taken to EL1 using
8001 AArch64, this value will be reported in the ISS field of ESR_ELx.
8003 See KVM_CAP_VCPU_EVENTS for more details.
8005 8.20 KVM_CAP_HYPERV_SEND_IPI
8006 ----------------------------
8010 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
8012 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
8014 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
8015 -----------------------------------
8019 This capability indicates that KVM running on top of Hyper-V hypervisor
8020 enables Direct TLB flush for its guests meaning that TLB flush
8021 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
8022 Due to the different ABI for hypercall parameters between Hyper-V and
8023 KVM, enabling this capability effectively disables all hypercall
8024 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
8025 flush hypercalls by Hyper-V) so userspace should disable KVM identification
8026 in CPUID and only exposes Hyper-V identification. In this case, guest
8027 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
8029 8.22 KVM_CAP_S390_VCPU_RESETS
8030 -----------------------------
8032 :Architectures: s390
8034 This capability indicates that the KVM_S390_NORMAL_RESET and
8035 KVM_S390_CLEAR_RESET ioctls are available.
8037 8.23 KVM_CAP_S390_PROTECTED
8038 ---------------------------
8040 :Architectures: s390
8042 This capability indicates that the Ultravisor has been initialized and
8043 KVM can therefore start protected VMs.
8044 This capability governs the KVM_S390_PV_COMMAND ioctl and the
8045 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
8046 guests when the state change is invalid.
8048 8.24 KVM_CAP_STEAL_TIME
8049 -----------------------
8051 :Architectures: arm64, x86
8053 This capability indicates that KVM supports steal time accounting.
8054 When steal time accounting is supported it may be enabled with
8055 architecture-specific interfaces. This capability and the architecture-
8056 specific interfaces must be consistent, i.e. if one says the feature
8057 is supported, than the other should as well and vice versa. For arm64
8058 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
8059 For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME".
8061 8.25 KVM_CAP_S390_DIAG318
8062 -------------------------
8064 :Architectures: s390
8066 This capability enables a guest to set information about its control program
8067 (i.e. guest kernel type and version). The information is helpful during
8068 system/firmware service events, providing additional data about the guest
8069 environments running on the machine.
8071 The information is associated with the DIAGNOSE 0x318 instruction, which sets
8072 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
8073 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
8074 environment the control program is running in (e.g. Linux, z/VM...), and the
8075 CPVC is used for information specific to OS (e.g. Linux version, Linux
8078 If this capability is available, then the CPNC and CPVC can be synchronized
8079 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
8081 8.26 KVM_CAP_X86_USER_SPACE_MSR
8082 -------------------------------
8086 This capability indicates that KVM supports deflection of MSR reads and
8087 writes to user space. It can be enabled on a VM level. If enabled, MSR
8088 accesses that would usually trigger a #GP by KVM into the guest will
8089 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
8090 KVM_EXIT_X86_WRMSR exit notifications.
8092 8.27 KVM_CAP_X86_MSR_FILTER
8093 ---------------------------
8097 This capability indicates that KVM supports that accesses to user defined MSRs
8098 may be rejected. With this capability exposed, KVM exports new VM ioctl
8099 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
8100 ranges that KVM should deny access to.
8102 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
8103 trap and emulate MSRs that are outside of the scope of KVM as well as
8104 limit the attack surface on KVM's MSR emulation code.
8106 8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
8107 -------------------------------------
8111 When enabled, KVM will disable paravirtual features provided to the
8112 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
8113 (0x40000001). Otherwise, a guest may use the paravirtual features
8114 regardless of what has actually been exposed through the CPUID leaf.
8116 8.29 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8117 ----------------------------------------------------------
8119 :Architectures: x86, arm64
8120 :Parameters: args[0] - size of the dirty log ring
8122 KVM is capable of tracking dirty memory using ring buffers that are
8123 mmapped into userspace; there is one dirty ring per vcpu.
8125 The dirty ring is available to userspace as an array of
8126 ``struct kvm_dirty_gfn``. Each dirty entry is defined as::
8128 struct kvm_dirty_gfn {
8130 __u32 slot; /* as_id | slot_id */
8134 The following values are defined for the flags field to define the
8135 current state of the entry::
8137 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
8138 #define KVM_DIRTY_GFN_F_RESET BIT(1)
8139 #define KVM_DIRTY_GFN_F_MASK 0x3
8141 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
8142 ioctl to enable this capability for the new guest and set the size of
8143 the rings. Enabling the capability is only allowed before creating any
8144 vCPU, and the size of the ring must be a power of two. The larger the
8145 ring buffer, the less likely the ring is full and the VM is forced to
8146 exit to userspace. The optimal size depends on the workload, but it is
8147 recommended that it be at least 64 KiB (4096 entries).
8149 Just like for dirty page bitmaps, the buffer tracks writes to
8150 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
8151 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
8152 with the flag set, userspace can start harvesting dirty pages from the
8155 An entry in the ring buffer can be unused (flag bits ``00``),
8156 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
8157 state machine for the entry is as follows::
8159 dirtied harvested reset
8160 00 -----------> 01 -------------> 1X -------+
8163 +------------------------------------------+
8165 To harvest the dirty pages, userspace accesses the mmapped ring buffer
8166 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
8167 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
8168 The userspace should harvest this GFN and mark the flags from state
8169 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
8170 to show that this GFN is harvested and waiting for a reset), and move
8171 on to the next GFN. The userspace should continue to do this until the
8172 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
8173 all the dirty GFNs that were available.
8175 Note that on weakly ordered architectures, userspace accesses to the
8176 ring buffer (and more specifically the 'flags' field) must be ordered,
8177 using load-acquire/store-release accessors when available, or any
8178 other memory barrier that will ensure this ordering.
8180 It's not necessary for userspace to harvest the all dirty GFNs at once.
8181 However it must collect the dirty GFNs in sequence, i.e., the userspace
8182 program cannot skip one dirty GFN to collect the one next to it.
8184 After processing one or more entries in the ring buffer, userspace
8185 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
8186 it, so that the kernel will reprotect those collected GFNs.
8187 Therefore, the ioctl must be called *before* reading the content of
8190 The dirty ring can get full. When it happens, the KVM_RUN of the
8191 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
8193 The dirty ring interface has a major difference comparing to the
8194 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
8195 userspace, it's still possible that the kernel has not yet flushed the
8196 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
8197 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
8198 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
8199 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
8201 NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that
8202 should be exposed by weakly ordered architecture, in order to indicate
8203 the additional memory ordering requirements imposed on userspace when
8204 reading the state of an entry and mutating it from DIRTY to HARVESTED.
8205 Architecture with TSO-like ordering (such as x86) are allowed to
8206 expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8209 After enabling the dirty rings, the userspace needs to detect the
8210 capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the
8211 ring structures can be backed by per-slot bitmaps. With this capability
8212 advertised, it means the architecture can dirty guest pages without
8213 vcpu/ring context, so that some of the dirty information will still be
8214 maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP
8215 can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8216 hasn't been enabled, or any memslot has been existing.
8218 Note that the bitmap here is only a backup of the ring structure. The
8219 use of the ring and bitmap combination is only beneficial if there is
8220 only a very small amount of memory that is dirtied out of vcpu/ring
8221 context. Otherwise, the stand-alone per-slot bitmap mechanism needs to
8224 To collect dirty bits in the backup bitmap, userspace can use the same
8225 KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all
8226 the generation of the dirty bits is done in a single pass. Collecting
8227 the dirty bitmap should be the very last thing that the VMM does before
8228 considering the state as complete. VMM needs to ensure that the dirty
8229 state is final and avoid missing dirty pages from another ioctl ordered
8230 after the bitmap collection.
8232 NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its
8233 tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on
8234 KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through
8235 command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device
8236 "kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save
8237 vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES}
8238 command on KVM device "kvm-arm-vgic-v3".
8240 8.30 KVM_CAP_XEN_HVM
8241 --------------------
8245 This capability indicates the features that Xen supports for hosting Xen
8246 PVHVM guests. Valid flags are::
8248 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
8249 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
8250 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
8251 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3)
8252 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4)
8253 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5)
8254 #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG (1 << 6)
8256 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
8257 ioctl is available, for the guest to set its hypercall page.
8259 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
8260 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
8261 contents, to request that KVM generate hypercall page content automatically
8262 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
8264 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
8265 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
8266 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
8267 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
8270 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
8271 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
8272 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
8274 The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries
8275 of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority
8276 field set to indicate 2 level event channel delivery.
8278 The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports
8279 injecting event channel events directly into the guest with the
8280 KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the
8281 KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the
8282 KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes.
8283 related to event channel delivery, timers, and the XENVER_version
8286 The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports
8287 the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR
8288 and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the
8289 XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during
8290 updates of the runstate information. Note that versions of KVM which support
8291 the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will
8292 always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure,
8293 which is perhaps counterintuitive. When this flag is advertised, KVM will
8294 behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless
8295 specifically enabled (by the guest making the hypercall, causing the VMM
8296 to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute).
8298 8.31 KVM_CAP_PPC_MULTITCE
8299 -------------------------
8301 :Capability: KVM_CAP_PPC_MULTITCE
8305 This capability means the kernel is capable of handling hypercalls
8306 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
8307 space. This significantly accelerates DMA operations for PPC KVM guests.
8308 User space should expect that its handlers for these hypercalls
8309 are not going to be called if user space previously registered LIOBN
8310 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
8312 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
8313 user space might have to advertise it for the guest. For example,
8314 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
8315 present in the "ibm,hypertas-functions" device-tree property.
8317 The hypercalls mentioned above may or may not be processed successfully
8318 in the kernel based fast path. If they can not be handled by the kernel,
8319 they will get passed on to user space. So user space still has to have
8320 an implementation for these despite the in kernel acceleration.
8322 This capability is always enabled.
8324 8.32 KVM_CAP_PTP_KVM
8325 --------------------
8327 :Architectures: arm64
8329 This capability indicates that the KVM virtual PTP service is
8330 supported in the host. A VMM can check whether the service is
8331 available to the guest on migration.
8333 8.33 KVM_CAP_HYPERV_ENFORCE_CPUID
8334 ---------------------------------
8338 When enabled, KVM will disable emulated Hyper-V features provided to the
8339 guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
8340 currently implemented Hyper-V features are provided unconditionally when
8341 Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
8344 8.34 KVM_CAP_EXIT_HYPERCALL
8345 ---------------------------
8347 :Capability: KVM_CAP_EXIT_HYPERCALL
8351 This capability, if enabled, will cause KVM to exit to userspace
8352 with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
8354 Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
8355 of hypercalls that can be configured to exit to userspace.
8356 Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
8358 The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
8359 of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace
8360 the hypercalls whose corresponding bit is in the argument, and return
8361 ENOSYS for the others.
8363 8.35 KVM_CAP_PMU_CAPABILITY
8364 ---------------------------
8366 :Capability: KVM_CAP_PMU_CAPABILITY
8369 :Parameters: arg[0] is bitmask of PMU virtualization capabilities.
8370 :Returns: 0 on success, -EINVAL when arg[0] contains invalid bits
8372 This capability alters PMU virtualization in KVM.
8374 Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
8375 PMU virtualization capabilities that can be adjusted on a VM.
8377 The argument to KVM_ENABLE_CAP is also a bitmask and selects specific
8378 PMU virtualization capabilities to be applied to the VM. This can
8379 only be invoked on a VM prior to the creation of VCPUs.
8381 At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting
8382 this capability will disable PMU virtualization for that VM. Usermode
8383 should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
8385 8.36 KVM_CAP_ARM_SYSTEM_SUSPEND
8386 -------------------------------
8388 :Capability: KVM_CAP_ARM_SYSTEM_SUSPEND
8389 :Architectures: arm64
8392 When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of
8393 type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request.
8395 8.37 KVM_CAP_S390_PROTECTED_DUMP
8396 --------------------------------
8398 :Capability: KVM_CAP_S390_PROTECTED_DUMP
8399 :Architectures: s390
8402 This capability indicates that KVM and the Ultravisor support dumping
8403 PV guests. The `KVM_PV_DUMP` command is available for the
8404 `KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides
8405 dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is
8406 available and supports the `KVM_PV_DUMP_CPU` subcommand.
8408 8.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
8409 -------------------------------------
8411 :Capability: KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
8414 :Parameters: arg[0] must be 0.
8415 :Returns: 0 on success, -EPERM if the userspace process does not
8416 have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been
8419 This capability disables the NX huge pages mitigation for iTLB MULTIHIT.
8421 The capability has no effect if the nx_huge_pages module parameter is not set.
8423 This capability may only be set before any vCPUs are created.
8425 8.39 KVM_CAP_S390_CPU_TOPOLOGY
8426 ------------------------------
8428 :Capability: KVM_CAP_S390_CPU_TOPOLOGY
8429 :Architectures: s390
8432 This capability indicates that KVM will provide the S390 CPU Topology
8433 facility which consist of the interpretation of the PTF instruction for
8434 the function code 2 along with interception and forwarding of both the
8435 PTF instruction with function codes 0 or 1 and the STSI(15,1,x)
8436 instruction to the userland hypervisor.
8438 The stfle facility 11, CPU Topology facility, should not be indicated
8439 to the guest without this capability.
8441 When this capability is present, KVM provides a new attribute group
8442 on vm fd, KVM_S390_VM_CPU_TOPOLOGY.
8443 This new attribute allows to get, set or clear the Modified Change
8444 Topology Report (MTCR) bit of the SCA through the kvm_device_attr
8447 When getting the Modified Change Topology Report value, the attr->addr
8448 must point to a byte where the value will be stored or retrieved from.
8450 8.40 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE
8451 ---------------------------------------
8453 :Capability: KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE
8454 :Architectures: arm64
8456 :Parameters: arg[0] is the new split chunk size.
8457 :Returns: 0 on success, -EINVAL if any memslot was already created.
8459 This capability sets the chunk size used in Eager Page Splitting.
8461 Eager Page Splitting improves the performance of dirty-logging (used
8462 in live migrations) when guest memory is backed by huge-pages. It
8463 avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing
8464 it eagerly when enabling dirty logging (with the
8465 KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using
8466 KVM_CLEAR_DIRTY_LOG.
8468 The chunk size specifies how many pages to break at a time, using a
8469 single allocation for each chunk. Bigger the chunk size, more pages
8470 need to be allocated ahead of time.
8472 The chunk size needs to be a valid block size. The list of acceptable
8473 block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a
8474 64-bit bitmap (each bit describing a block size). The default value is
8475 0, to disable the eager page splitting.
8477 9. Known KVM API problems
8478 =========================
8480 In some cases, KVM's API has some inconsistencies or common pitfalls
8481 that userspace need to be aware of. This section details some of
8484 Most of them are architecture specific, so the section is split by
8490 ``KVM_GET_SUPPORTED_CPUID`` issues
8491 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8493 In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible
8494 to take its result and pass it directly to ``KVM_SET_CPUID2``. This section
8495 documents some cases in which that requires some care.
8500 CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``,
8501 but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or
8502 ``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of
8505 The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature.
8507 CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``.
8508 It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel
8509 has enabled in-kernel emulation of the local APIC.
8514 Several CPUID values include topology information for the host CPU:
8515 0x0b and 0x1f for Intel systems, 0x8000001e for AMD systems. Different
8516 versions of KVM return different values for this information and userspace
8517 should not rely on it. Currently they return all zeroes.
8519 If userspace wishes to set up a guest topology, it should be careful that
8520 the values of these three leaves differ for each CPU. In particular,
8521 the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX
8522 for 0x8000001e; the latter also encodes the core id and node id in bits
8523 7:0 of EBX and ECX respectively.
8525 Obsolete ioctls and capabilities
8526 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8528 KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually
8529 available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if
8532 Ordering of KVM_GET_*/KVM_SET_* ioctls
8533 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^