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.
275 4.6 KVM_SET_MEMORY_REGION
276 -------------------------
281 :Parameters: struct kvm_memory_region (in)
282 :Returns: 0 on success, -1 on error
284 This ioctl is obsolete and has been removed.
293 :Parameters: vcpu id (apic id on x86)
294 :Returns: vcpu fd on success, -1 on error
296 This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
297 The vcpu id is an integer in the range [0, max_vcpu_id).
299 The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
300 the KVM_CHECK_EXTENSION ioctl() at run-time.
301 The maximum possible value for max_vcpus can be retrieved using the
302 KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
304 If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
306 If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
307 same as the value returned from KVM_CAP_NR_VCPUS.
309 The maximum possible value for max_vcpu_id can be retrieved using the
310 KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
312 If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
313 is the same as the value returned from KVM_CAP_MAX_VCPUS.
315 On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
316 threads in one or more virtual CPU cores. (This is because the
317 hardware requires all the hardware threads in a CPU core to be in the
318 same partition.) The KVM_CAP_PPC_SMT capability indicates the number
319 of vcpus per virtual core (vcore). The vcore id is obtained by
320 dividing the vcpu id by the number of vcpus per vcore. The vcpus in a
321 given vcore will always be in the same physical core as each other
322 (though that might be a different physical core from time to time).
323 Userspace can control the threading (SMT) mode of the guest by its
324 allocation of vcpu ids. For example, if userspace wants
325 single-threaded guest vcpus, it should make all vcpu ids be a multiple
326 of the number of vcpus per vcore.
328 For virtual cpus that have been created with S390 user controlled virtual
329 machines, the resulting vcpu fd can be memory mapped at page offset
330 KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
331 cpu's hardware control block.
334 4.8 KVM_GET_DIRTY_LOG (vm ioctl)
335 --------------------------------
340 :Parameters: struct kvm_dirty_log (in/out)
341 :Returns: 0 on success, -1 on error
345 /* for KVM_GET_DIRTY_LOG */
346 struct kvm_dirty_log {
350 void __user *dirty_bitmap; /* one bit per page */
355 Given a memory slot, return a bitmap containing any pages dirtied
356 since the last call to this ioctl. Bit 0 is the first page in the
357 memory slot. Ensure the entire structure is cleared to avoid padding
360 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
361 the address space for which you want to return the dirty bitmap. See
362 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
364 The bits in the dirty bitmap are cleared before the ioctl returns, unless
365 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled. For more information,
366 see the description of the capability.
368 Note that the Xen shared info page, if configured, shall always be assumed
369 to be dirty. KVM will not explicitly mark it such.
371 4.9 KVM_SET_MEMORY_ALIAS
372 ------------------------
377 :Parameters: struct kvm_memory_alias (in)
378 :Returns: 0 (success), -1 (error)
380 This ioctl is obsolete and has been removed.
390 :Returns: 0 on success, -1 on error
394 ======= ==============================================================
395 EINTR an unmasked signal is pending
396 ENOEXEC the vcpu hasn't been initialized or the guest tried to execute
397 instructions from device memory (arm64)
398 ENOSYS data abort outside memslots with no syndrome info and
399 KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
400 EPERM SVE feature set but not finalized (arm64)
401 ======= ==============================================================
403 This ioctl is used to run a guest virtual cpu. While there are no
404 explicit parameters, there is an implicit parameter block that can be
405 obtained by mmap()ing the vcpu fd at offset 0, with the size given by
406 KVM_GET_VCPU_MMAP_SIZE. The parameter block is formatted as a 'struct
407 kvm_run' (see below).
414 :Architectures: all except arm64
416 :Parameters: struct kvm_regs (out)
417 :Returns: 0 on success, -1 on error
419 Reads the general purpose registers from the vcpu.
425 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
426 __u64 rax, rbx, rcx, rdx;
427 __u64 rsi, rdi, rsp, rbp;
428 __u64 r8, r9, r10, r11;
429 __u64 r12, r13, r14, r15;
435 /* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
447 :Architectures: all except arm64
449 :Parameters: struct kvm_regs (in)
450 :Returns: 0 on success, -1 on error
452 Writes the general purpose registers into the vcpu.
454 See KVM_GET_REGS for the data structure.
461 :Architectures: x86, ppc
463 :Parameters: struct kvm_sregs (out)
464 :Returns: 0 on success, -1 on error
466 Reads special registers from the vcpu.
472 struct kvm_segment cs, ds, es, fs, gs, ss;
473 struct kvm_segment tr, ldt;
474 struct kvm_dtable gdt, idt;
475 __u64 cr0, cr2, cr3, cr4, cr8;
478 __u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
481 /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
483 interrupt_bitmap is a bitmap of pending external interrupts. At most
484 one bit may be set. This interrupt has been acknowledged by the APIC
485 but not yet injected into the cpu core.
492 :Architectures: x86, ppc
494 :Parameters: struct kvm_sregs (in)
495 :Returns: 0 on success, -1 on error
497 Writes special registers into the vcpu. See KVM_GET_SREGS for the
507 :Parameters: struct kvm_translation (in/out)
508 :Returns: 0 on success, -1 on error
510 Translates a virtual address according to the vcpu's current address
515 struct kvm_translation {
517 __u64 linear_address;
520 __u64 physical_address;
532 :Architectures: x86, ppc, mips, riscv
534 :Parameters: struct kvm_interrupt (in)
535 :Returns: 0 on success, negative on failure.
537 Queues a hardware interrupt vector to be injected.
541 /* for KVM_INTERRUPT */
542 struct kvm_interrupt {
552 ========= ===================================
554 -EEXIST if an interrupt is already enqueued
555 -EINVAL the irq number is invalid
556 -ENXIO if the PIC is in the kernel
557 -EFAULT if the pointer is invalid
558 ========= ===================================
560 Note 'irq' is an interrupt vector, not an interrupt pin or line. This
561 ioctl is useful if the in-kernel PIC is not used.
566 Queues an external interrupt to be injected. This ioctl is overleaded
567 with 3 different irq values:
571 This injects an edge type external interrupt into the guest once it's ready
572 to receive interrupts. When injected, the interrupt is done.
574 b) KVM_INTERRUPT_UNSET
576 This unsets any pending interrupt.
578 Only available with KVM_CAP_PPC_UNSET_IRQ.
580 c) KVM_INTERRUPT_SET_LEVEL
582 This injects a level type external interrupt into the guest context. The
583 interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
586 Only available with KVM_CAP_PPC_IRQ_LEVEL.
588 Note that any value for 'irq' other than the ones stated above is invalid
589 and incurs unexpected behavior.
591 This is an asynchronous vcpu ioctl and can be invoked from any thread.
596 Queues an external interrupt to be injected into the virtual CPU. A negative
597 interrupt number dequeues the interrupt.
599 This is an asynchronous vcpu ioctl and can be invoked from any thread.
604 Queues an external interrupt to be injected into the virutal CPU. This ioctl
605 is overloaded with 2 different irq values:
609 This sets external interrupt for a virtual CPU and it will receive
612 b) KVM_INTERRUPT_UNSET
614 This clears pending external interrupt for a virtual CPU.
616 This is an asynchronous vcpu ioctl and can be invoked from any thread.
626 :Returns: -1 on error
628 Support for this has been removed. Use KVM_SET_GUEST_DEBUG instead.
634 :Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
636 :Type: system ioctl, vcpu ioctl
637 :Parameters: struct kvm_msrs (in/out)
638 :Returns: number of msrs successfully returned;
641 When used as a system ioctl:
642 Reads the values of MSR-based features that are available for the VM. This
643 is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
644 The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
647 When used as a vcpu ioctl:
648 Reads model-specific registers from the vcpu. Supported msr indices can
649 be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
654 __u32 nmsrs; /* number of msrs in entries */
657 struct kvm_msr_entry entries[0];
660 struct kvm_msr_entry {
666 Application code should set the 'nmsrs' member (which indicates the
667 size of the entries array) and the 'index' member of each array entry.
668 kvm will fill in the 'data' member.
677 :Parameters: struct kvm_msrs (in)
678 :Returns: number of msrs successfully set (see below), -1 on error
680 Writes model-specific registers to the vcpu. See KVM_GET_MSRS for the
683 Application code should set the 'nmsrs' member (which indicates the
684 size of the entries array), and the 'index' and 'data' members of each
687 It tries to set the MSRs in array entries[] one by one. If setting an MSR
688 fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
689 by KVM, etc..., it stops processing the MSR list and returns the number of
690 MSRs that have been set successfully.
699 :Parameters: struct kvm_cpuid (in)
700 :Returns: 0 on success, -1 on error
702 Defines the vcpu responses to the cpuid instruction. Applications
703 should use the KVM_SET_CPUID2 ioctl if available.
706 - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID
707 configuration (if there is) is not corrupted. Userspace can get a copy
708 of the resulting CPUID configuration through KVM_GET_CPUID2 in case.
709 - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model
710 after running the guest, may cause guest instability.
711 - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc...
712 may cause guest instability.
716 struct kvm_cpuid_entry {
725 /* for KVM_SET_CPUID */
729 struct kvm_cpuid_entry entries[0];
733 4.21 KVM_SET_SIGNAL_MASK
734 ------------------------
739 :Parameters: struct kvm_signal_mask (in)
740 :Returns: 0 on success, -1 on error
742 Defines which signals are blocked during execution of KVM_RUN. This
743 signal mask temporarily overrides the threads signal mask. Any
744 unblocked signal received (except SIGKILL and SIGSTOP, which retain
745 their traditional behaviour) will cause KVM_RUN to return with -EINTR.
747 Note the signal will only be delivered if not blocked by the original
752 /* for KVM_SET_SIGNAL_MASK */
753 struct kvm_signal_mask {
765 :Parameters: struct kvm_fpu (out)
766 :Returns: 0 on success, -1 on error
768 Reads the floating point state from the vcpu.
772 /* for KVM_GET_FPU and KVM_SET_FPU */
777 __u8 ftwx; /* in fxsave format */
794 :Parameters: struct kvm_fpu (in)
795 :Returns: 0 on success, -1 on error
797 Writes the floating point state to the vcpu.
801 /* for KVM_GET_FPU and KVM_SET_FPU */
806 __u8 ftwx; /* in fxsave format */
817 4.24 KVM_CREATE_IRQCHIP
818 -----------------------
820 :Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
821 :Architectures: x86, arm64, s390
824 :Returns: 0 on success, -1 on error
826 Creates an interrupt controller model in the kernel.
827 On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
828 future vcpus to have a local APIC. IRQ routing for GSIs 0-15 is set to both
829 PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
830 On arm64, a GICv2 is created. Any other GIC versions require the usage of
831 KVM_CREATE_DEVICE, which also supports creating a GICv2. Using
832 KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
833 On s390, a dummy irq routing table is created.
835 Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
836 before KVM_CREATE_IRQCHIP can be used.
842 :Capability: KVM_CAP_IRQCHIP
843 :Architectures: x86, arm64
845 :Parameters: struct kvm_irq_level
846 :Returns: 0 on success, -1 on error
848 Sets the level of a GSI input to the interrupt controller model in the kernel.
849 On some architectures it is required that an interrupt controller model has
850 been previously created with KVM_CREATE_IRQCHIP. Note that edge-triggered
851 interrupts require the level to be set to 1 and then back to 0.
853 On real hardware, interrupt pins can be active-low or active-high. This
854 does not matter for the level field of struct kvm_irq_level: 1 always
855 means active (asserted), 0 means inactive (deasserted).
857 x86 allows the operating system to program the interrupt polarity
858 (active-low/active-high) for level-triggered interrupts, and KVM used
859 to consider the polarity. However, due to bitrot in the handling of
860 active-low interrupts, the above convention is now valid on x86 too.
861 This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED. Userspace
862 should not present interrupts to the guest as active-low unless this
863 capability is present (or unless it is not using the in-kernel irqchip,
867 arm64 can signal an interrupt either at the CPU level, or at the
868 in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
869 use PPIs designated for specific cpus. The irq field is interpreted
872 bits: | 31 ... 28 | 27 ... 24 | 23 ... 16 | 15 ... 0 |
873 field: | vcpu2_index | irq_type | vcpu_index | irq_id |
875 The irq_type field has the following values:
878 out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
880 in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
881 (the vcpu_index field is ignored)
883 in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
885 (The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
887 In both cases, level is used to assert/deassert the line.
889 When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
890 identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
893 Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions
894 injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
895 be used for a userspace interrupt controller.
899 struct kvm_irq_level {
902 __s32 status; /* not used for KVM_IRQ_LEVEL */
904 __u32 level; /* 0 or 1 */
911 :Capability: KVM_CAP_IRQCHIP
914 :Parameters: struct kvm_irqchip (in/out)
915 :Returns: 0 on success, -1 on error
917 Reads the state of a kernel interrupt controller created with
918 KVM_CREATE_IRQCHIP into a buffer provided by the caller.
923 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
926 char dummy[512]; /* reserving space */
927 struct kvm_pic_state pic;
928 struct kvm_ioapic_state ioapic;
936 :Capability: KVM_CAP_IRQCHIP
939 :Parameters: struct kvm_irqchip (in)
940 :Returns: 0 on success, -1 on error
942 Sets the state of a kernel interrupt controller created with
943 KVM_CREATE_IRQCHIP from a buffer provided by the caller.
948 __u32 chip_id; /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
951 char dummy[512]; /* reserving space */
952 struct kvm_pic_state pic;
953 struct kvm_ioapic_state ioapic;
958 4.28 KVM_XEN_HVM_CONFIG
959 -----------------------
961 :Capability: KVM_CAP_XEN_HVM
964 :Parameters: struct kvm_xen_hvm_config (in)
965 :Returns: 0 on success, -1 on error
967 Sets the MSR that the Xen HVM guest uses to initialize its hypercall
968 page, and provides the starting address and size of the hypercall
969 blobs in userspace. When the guest writes the MSR, kvm copies one
970 page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
975 struct kvm_xen_hvm_config {
985 If the KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag is returned from the
986 KVM_CAP_XEN_HVM check, it may be set in the flags field of this ioctl.
987 This requests KVM to generate the contents of the hypercall page
988 automatically; hypercalls will be intercepted and passed to userspace
989 through KVM_EXIT_XEN. In this case, all of the blob size and address
992 No other flags are currently valid in the struct kvm_xen_hvm_config.
997 :Capability: KVM_CAP_ADJUST_CLOCK
1000 :Parameters: struct kvm_clock_data (out)
1001 :Returns: 0 on success, -1 on error
1003 Gets the current timestamp of kvmclock as seen by the current guest. In
1004 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
1007 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
1008 set of bits that KVM can return in struct kvm_clock_data's flag member.
1010 The following flags are defined:
1012 KVM_CLOCK_TSC_STABLE
1013 If set, the returned value is the exact kvmclock
1014 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called.
1015 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant
1016 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try
1017 to make all VCPUs follow this clock, but the exact value read by each
1018 VCPU could differ, because the host TSC is not stable.
1021 If set, the `realtime` field in the kvm_clock_data
1022 structure is populated with the value of the host's real time
1023 clocksource at the instant when KVM_GET_CLOCK was called. If clear,
1024 the `realtime` field does not contain a value.
1027 If set, the `host_tsc` field in the kvm_clock_data
1028 structure is populated with the value of the host's timestamp counter (TSC)
1029 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field
1030 does not contain a value.
1034 struct kvm_clock_data {
1035 __u64 clock; /* kvmclock current value */
1047 :Capability: KVM_CAP_ADJUST_CLOCK
1050 :Parameters: struct kvm_clock_data (in)
1051 :Returns: 0 on success, -1 on error
1053 Sets the current timestamp of kvmclock to the value specified in its parameter.
1054 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1057 The following flags can be passed:
1060 If set, KVM will compare the value of the `realtime` field
1061 with the value of the host's real time clocksource at the instant when
1062 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final
1063 kvmclock value that will be provided to guests.
1065 Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored.
1069 struct kvm_clock_data {
1070 __u64 clock; /* kvmclock current value */
1079 4.31 KVM_GET_VCPU_EVENTS
1080 ------------------------
1082 :Capability: KVM_CAP_VCPU_EVENTS
1083 :Extended by: KVM_CAP_INTR_SHADOW
1084 :Architectures: x86, arm64
1086 :Parameters: struct kvm_vcpu_event (out)
1087 :Returns: 0 on success, -1 on error
1092 Gets currently pending exceptions, interrupts, and NMIs as well as related
1097 struct kvm_vcpu_events {
1101 __u8 has_error_code;
1122 __u8 smm_inside_nmi;
1126 __u8 exception_has_payload;
1127 __u64 exception_payload;
1130 The following bits are defined in the flags field:
1132 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1133 interrupt.shadow contains a valid state.
1135 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1138 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1139 exception_has_payload, exception_payload, and exception.pending
1140 fields contain a valid state. This bit will be set whenever
1141 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1146 If the guest accesses a device that is being emulated by the host kernel in
1147 such a way that a real device would generate a physical SError, KVM may make
1148 a virtual SError pending for that VCPU. This system error interrupt remains
1149 pending until the guest takes the exception by unmasking PSTATE.A.
1151 Running the VCPU may cause it to take a pending SError, or make an access that
1152 causes an SError to become pending. The event's description is only valid while
1153 the VPCU is not running.
1155 This API provides a way to read and write the pending 'event' state that is not
1156 visible to the guest. To save, restore or migrate a VCPU the struct representing
1157 the state can be read then written using this GET/SET API, along with the other
1158 guest-visible registers. It is not possible to 'cancel' an SError that has been
1161 A device being emulated in user-space may also wish to generate an SError. To do
1162 this the events structure can be populated by user-space. The current state
1163 should be read first, to ensure no existing SError is pending. If an existing
1164 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1165 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1166 Serviceability (RAS) Specification").
1168 SError exceptions always have an ESR value. Some CPUs have the ability to
1169 specify what the virtual SError's ESR value should be. These systems will
1170 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1171 always have a non-zero value when read, and the agent making an SError pending
1172 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1173 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1174 with exception.has_esr as zero, KVM will choose an ESR.
1176 Specifying exception.has_esr on a system that does not support it will return
1177 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1178 will return -EINVAL.
1180 It is not possible to read back a pending external abort (injected via
1181 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1182 directly to the virtual CPU).
1186 struct kvm_vcpu_events {
1188 __u8 serror_pending;
1189 __u8 serror_has_esr;
1190 __u8 ext_dabt_pending;
1191 /* Align it to 8 bytes */
1198 4.32 KVM_SET_VCPU_EVENTS
1199 ------------------------
1201 :Capability: KVM_CAP_VCPU_EVENTS
1202 :Extended by: KVM_CAP_INTR_SHADOW
1203 :Architectures: x86, arm64
1205 :Parameters: struct kvm_vcpu_event (in)
1206 :Returns: 0 on success, -1 on error
1211 Set pending exceptions, interrupts, and NMIs as well as related states of the
1214 See KVM_GET_VCPU_EVENTS for the data structure.
1216 Fields that may be modified asynchronously by running VCPUs can be excluded
1217 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1218 smi.pending. Keep the corresponding bits in the flags field cleared to
1219 suppress overwriting the current in-kernel state. The bits are:
1221 =============================== ==================================
1222 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1223 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1224 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1225 =============================== ==================================
1227 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1228 the flags field to signal that interrupt.shadow contains a valid state and
1229 shall be written into the VCPU.
1231 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1233 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1234 can be set in the flags field to signal that the
1235 exception_has_payload, exception_payload, and exception.pending fields
1236 contain a valid state and shall be written into the VCPU.
1241 User space may need to inject several types of events to the guest.
1243 Set the pending SError exception state for this VCPU. It is not possible to
1244 'cancel' an Serror that has been made pending.
1246 If the guest performed an access to I/O memory which could not be handled by
1247 userspace, for example because of missing instruction syndrome decode
1248 information or because there is no device mapped at the accessed IPA, then
1249 userspace can ask the kernel to inject an external abort using the address
1250 from the exiting fault on the VCPU. It is a programming error to set
1251 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1252 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1253 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1254 how userspace reports accesses for the above cases to guests, across different
1255 userspace implementations. Nevertheless, userspace can still emulate all Arm
1256 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1258 See KVM_GET_VCPU_EVENTS for the data structure.
1261 4.33 KVM_GET_DEBUGREGS
1262 ----------------------
1264 :Capability: KVM_CAP_DEBUGREGS
1267 :Parameters: struct kvm_debugregs (out)
1268 :Returns: 0 on success, -1 on error
1270 Reads debug registers from the vcpu.
1274 struct kvm_debugregs {
1283 4.34 KVM_SET_DEBUGREGS
1284 ----------------------
1286 :Capability: KVM_CAP_DEBUGREGS
1289 :Parameters: struct kvm_debugregs (in)
1290 :Returns: 0 on success, -1 on error
1292 Writes debug registers into the vcpu.
1294 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1295 yet and must be cleared on entry.
1298 4.35 KVM_SET_USER_MEMORY_REGION
1299 -------------------------------
1301 :Capability: KVM_CAP_USER_MEMORY
1304 :Parameters: struct kvm_userspace_memory_region (in)
1305 :Returns: 0 on success, -1 on error
1309 struct kvm_userspace_memory_region {
1312 __u64 guest_phys_addr;
1313 __u64 memory_size; /* bytes */
1314 __u64 userspace_addr; /* start of the userspace allocated memory */
1317 /* for kvm_memory_region::flags */
1318 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1319 #define KVM_MEM_READONLY (1UL << 1)
1321 This ioctl allows the user to create, modify or delete a guest physical
1322 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1323 should be less than the maximum number of user memory slots supported per
1324 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1325 Slots may not overlap in guest physical address space.
1327 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1328 specifies the address space which is being modified. They must be
1329 less than the value that KVM_CHECK_EXTENSION returns for the
1330 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1331 are unrelated; the restriction on overlapping slots only applies within
1334 Deleting a slot is done by passing zero for memory_size. When changing
1335 an existing slot, it may be moved in the guest physical memory space,
1336 or its flags may be modified, but it may not be resized.
1338 Memory for the region is taken starting at the address denoted by the
1339 field userspace_addr, which must point at user addressable memory for
1340 the entire memory slot size. Any object may back this memory, including
1341 anonymous memory, ordinary files, and hugetlbfs.
1343 On architectures that support a form of address tagging, userspace_addr must
1344 be an untagged address.
1346 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1347 be identical. This allows large pages in the guest to be backed by large
1350 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1351 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1352 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1353 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1354 to make a new slot read-only. In this case, writes to this memory will be
1355 posted to userspace as KVM_EXIT_MMIO exits.
1357 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1358 the memory region are automatically reflected into the guest. For example, an
1359 mmap() that affects the region will be made visible immediately. Another
1360 example is madvise(MADV_DROP).
1362 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1363 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1364 allocation and is deprecated.
1367 4.36 KVM_SET_TSS_ADDR
1368 ---------------------
1370 :Capability: KVM_CAP_SET_TSS_ADDR
1373 :Parameters: unsigned long tss_address (in)
1374 :Returns: 0 on success, -1 on error
1376 This ioctl defines the physical address of a three-page region in the guest
1377 physical address space. The region must be within the first 4GB of the
1378 guest physical address space and must not conflict with any memory slot
1379 or any mmio address. The guest may malfunction if it accesses this memory
1382 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1383 because of a quirk in the virtualization implementation (see the internals
1384 documentation when it pops into existence).
1390 :Capability: KVM_CAP_ENABLE_CAP
1391 :Architectures: mips, ppc, s390, x86
1393 :Parameters: struct kvm_enable_cap (in)
1394 :Returns: 0 on success; -1 on error
1396 :Capability: KVM_CAP_ENABLE_CAP_VM
1399 :Parameters: struct kvm_enable_cap (in)
1400 :Returns: 0 on success; -1 on error
1404 Not all extensions are enabled by default. Using this ioctl the application
1405 can enable an extension, making it available to the guest.
1407 On systems that do not support this ioctl, it always fails. On systems that
1408 do support it, it only works for extensions that are supported for enablement.
1410 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1415 struct kvm_enable_cap {
1419 The capability that is supposed to get enabled.
1425 A bitfield indicating future enhancements. Has to be 0 for now.
1431 Arguments for enabling a feature. If a feature needs initial values to
1432 function properly, this is the place to put them.
1439 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1440 for vm-wide capabilities.
1442 4.38 KVM_GET_MP_STATE
1443 ---------------------
1445 :Capability: KVM_CAP_MP_STATE
1446 :Architectures: x86, s390, arm64, riscv
1448 :Parameters: struct kvm_mp_state (out)
1449 :Returns: 0 on success; -1 on error
1453 struct kvm_mp_state {
1457 Returns the vcpu's current "multiprocessing state" (though also valid on
1458 uniprocessor guests).
1460 Possible values are:
1462 ========================== ===============================================
1463 KVM_MP_STATE_RUNNABLE the vcpu is currently running
1465 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1466 which has not yet received an INIT signal [x86]
1467 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1468 now ready for a SIPI [x86]
1469 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1470 is waiting for an interrupt [x86]
1471 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1472 accessible via KVM_GET_VCPU_EVENTS) [x86]
1473 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv]
1474 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1475 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1477 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1479 ========================== ===============================================
1481 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1482 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1483 these architectures.
1488 The only states that are valid are KVM_MP_STATE_STOPPED and
1489 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1491 4.39 KVM_SET_MP_STATE
1492 ---------------------
1494 :Capability: KVM_CAP_MP_STATE
1495 :Architectures: x86, s390, arm64, riscv
1497 :Parameters: struct kvm_mp_state (in)
1498 :Returns: 0 on success; -1 on error
1500 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1503 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1504 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1505 these architectures.
1510 The only states that are valid are KVM_MP_STATE_STOPPED and
1511 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1513 4.40 KVM_SET_IDENTITY_MAP_ADDR
1514 ------------------------------
1516 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1519 :Parameters: unsigned long identity (in)
1520 :Returns: 0 on success, -1 on error
1522 This ioctl defines the physical address of a one-page region in the guest
1523 physical address space. The region must be within the first 4GB of the
1524 guest physical address space and must not conflict with any memory slot
1525 or any mmio address. The guest may malfunction if it accesses this memory
1528 Setting the address to 0 will result in resetting the address to its default
1531 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1532 because of a quirk in the virtualization implementation (see the internals
1533 documentation when it pops into existence).
1535 Fails if any VCPU has already been created.
1537 4.41 KVM_SET_BOOT_CPU_ID
1538 ------------------------
1540 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1543 :Parameters: unsigned long vcpu_id
1544 :Returns: 0 on success, -1 on error
1546 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1547 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1548 is vcpu 0. This ioctl has to be called before vcpu creation,
1549 otherwise it will return EBUSY error.
1555 :Capability: KVM_CAP_XSAVE
1558 :Parameters: struct kvm_xsave (out)
1559 :Returns: 0 on success, -1 on error
1569 This ioctl would copy current vcpu's xsave struct to the userspace.
1575 :Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2
1578 :Parameters: struct kvm_xsave (in)
1579 :Returns: 0 on success, -1 on error
1589 This ioctl would copy userspace's xsave struct to the kernel. It copies
1590 as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2),
1591 when invoked on the vm file descriptor. The size value returned by
1592 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
1593 Currently, it is only greater than 4096 if a dynamic feature has been
1594 enabled with ``arch_prctl()``, but this may change in the future.
1596 The offsets of the state save areas in struct kvm_xsave follow the
1597 contents of CPUID leaf 0xD on the host.
1603 :Capability: KVM_CAP_XCRS
1606 :Parameters: struct kvm_xcrs (out)
1607 :Returns: 0 on success, -1 on error
1620 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1624 This ioctl would copy current vcpu's xcrs to the userspace.
1630 :Capability: KVM_CAP_XCRS
1633 :Parameters: struct kvm_xcrs (in)
1634 :Returns: 0 on success, -1 on error
1647 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1651 This ioctl would set vcpu's xcr to the value userspace specified.
1654 4.46 KVM_GET_SUPPORTED_CPUID
1655 ----------------------------
1657 :Capability: KVM_CAP_EXT_CPUID
1660 :Parameters: struct kvm_cpuid2 (in/out)
1661 :Returns: 0 on success, -1 on error
1668 struct kvm_cpuid_entry2 entries[0];
1671 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1672 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1673 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1675 struct kvm_cpuid_entry2 {
1686 This ioctl returns x86 cpuid features which are supported by both the
1687 hardware and kvm in its default configuration. Userspace can use the
1688 information returned by this ioctl to construct cpuid information (for
1689 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1690 userspace capabilities, and with user requirements (for example, the
1691 user may wish to constrain cpuid to emulate older hardware, or for
1692 feature consistency across a cluster).
1694 Dynamically-enabled feature bits need to be requested with
1695 ``arch_prctl()`` before calling this ioctl. Feature bits that have not
1696 been requested are excluded from the result.
1698 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1699 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1700 its default configuration. If userspace enables such capabilities, it
1701 is responsible for modifying the results of this ioctl appropriately.
1703 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1704 with the 'nent' field indicating the number of entries in the variable-size
1705 array 'entries'. If the number of entries is too low to describe the cpu
1706 capabilities, an error (E2BIG) is returned. If the number is too high,
1707 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1708 number is just right, the 'nent' field is adjusted to the number of valid
1709 entries in the 'entries' array, which is then filled.
1711 The entries returned are the host cpuid as returned by the cpuid instruction,
1712 with unknown or unsupported features masked out. Some features (for example,
1713 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1714 emulate them efficiently. The fields in each entry are defined as follows:
1717 the eax value used to obtain the entry
1720 the ecx value used to obtain the entry (for entries that are
1724 an OR of zero or more of the following:
1726 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1727 if the index field is valid
1730 the values returned by the cpuid instruction for
1731 this function/index combination
1733 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1734 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1735 support. Instead it is reported via::
1737 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1739 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1740 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1743 4.47 KVM_PPC_GET_PVINFO
1744 -----------------------
1746 :Capability: KVM_CAP_PPC_GET_PVINFO
1749 :Parameters: struct kvm_ppc_pvinfo (out)
1750 :Returns: 0 on success, !0 on error
1754 struct kvm_ppc_pvinfo {
1760 This ioctl fetches PV specific information that need to be passed to the guest
1761 using the device tree or other means from vm context.
1763 The hcall array defines 4 instructions that make up a hypercall.
1765 If any additional field gets added to this structure later on, a bit for that
1766 additional piece of information will be set in the flags bitmap.
1768 The flags bitmap is defined as::
1770 /* the host supports the ePAPR idle hcall
1771 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1773 4.52 KVM_SET_GSI_ROUTING
1774 ------------------------
1776 :Capability: KVM_CAP_IRQ_ROUTING
1777 :Architectures: x86 s390 arm64
1779 :Parameters: struct kvm_irq_routing (in)
1780 :Returns: 0 on success, -1 on error
1782 Sets the GSI routing table entries, overwriting any previously set entries.
1784 On arm64, GSI routing has the following limitation:
1786 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1790 struct kvm_irq_routing {
1793 struct kvm_irq_routing_entry entries[0];
1796 No flags are specified so far, the corresponding field must be set to zero.
1800 struct kvm_irq_routing_entry {
1806 struct kvm_irq_routing_irqchip irqchip;
1807 struct kvm_irq_routing_msi msi;
1808 struct kvm_irq_routing_s390_adapter adapter;
1809 struct kvm_irq_routing_hv_sint hv_sint;
1810 struct kvm_irq_routing_xen_evtchn xen_evtchn;
1815 /* gsi routing entry types */
1816 #define KVM_IRQ_ROUTING_IRQCHIP 1
1817 #define KVM_IRQ_ROUTING_MSI 2
1818 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1819 #define KVM_IRQ_ROUTING_HV_SINT 4
1820 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5
1824 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1825 type, specifies that the devid field contains a valid value. The per-VM
1826 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1827 the device ID. If this capability is not available, userspace should
1828 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1833 struct kvm_irq_routing_irqchip {
1838 struct kvm_irq_routing_msi {
1848 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1849 for the device that wrote the MSI message. For PCI, this is usually a
1850 BFD identifier in the lower 16 bits.
1852 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1853 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1854 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1855 address_hi must be zero.
1859 struct kvm_irq_routing_s390_adapter {
1863 __u32 summary_offset;
1867 struct kvm_irq_routing_hv_sint {
1872 struct kvm_irq_routing_xen_evtchn {
1879 When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit
1880 in its indication of supported features, routing to Xen event channels
1881 is supported. Although the priority field is present, only the value
1882 KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by
1883 2 level event channels. FIFO event channel support may be added in
1887 4.55 KVM_SET_TSC_KHZ
1888 --------------------
1890 :Capability: KVM_CAP_TSC_CONTROL
1893 :Parameters: virtual tsc_khz
1894 :Returns: 0 on success, -1 on error
1896 Specifies the tsc frequency for the virtual machine. The unit of the
1900 4.56 KVM_GET_TSC_KHZ
1901 --------------------
1903 :Capability: KVM_CAP_GET_TSC_KHZ
1907 :Returns: virtual tsc-khz on success, negative value on error
1909 Returns the tsc frequency of the guest. The unit of the return value is
1910 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1917 :Capability: KVM_CAP_IRQCHIP
1920 :Parameters: struct kvm_lapic_state (out)
1921 :Returns: 0 on success, -1 on error
1925 #define KVM_APIC_REG_SIZE 0x400
1926 struct kvm_lapic_state {
1927 char regs[KVM_APIC_REG_SIZE];
1930 Reads the Local APIC registers and copies them into the input argument. The
1931 data format and layout are the same as documented in the architecture manual.
1933 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1934 enabled, then the format of APIC_ID register depends on the APIC mode
1935 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1936 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1937 which is stored in bits 31-24 of the APIC register, or equivalently in
1938 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1939 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1941 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1942 always uses xAPIC format.
1948 :Capability: KVM_CAP_IRQCHIP
1951 :Parameters: struct kvm_lapic_state (in)
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 Copies the input argument into the Local APIC registers. The data format
1962 and layout are the same as documented in the architecture manual.
1964 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1965 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1966 See the note in KVM_GET_LAPIC.
1972 :Capability: KVM_CAP_IOEVENTFD
1975 :Parameters: struct kvm_ioeventfd (in)
1976 :Returns: 0 on success, !0 on error
1978 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1979 within the guest. A guest write in the registered address will signal the
1980 provided event instead of triggering an exit.
1984 struct kvm_ioeventfd {
1986 __u64 addr; /* legal pio/mmio address */
1987 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
1993 For the special case of virtio-ccw devices on s390, the ioevent is matched
1994 to a subchannel/virtqueue tuple instead.
1996 The following flags are defined::
1998 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1999 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
2000 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
2001 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2002 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2004 If datamatch flag is set, the event will be signaled only if the written value
2005 to the registered address is equal to datamatch in struct kvm_ioeventfd.
2007 For virtio-ccw devices, addr contains the subchannel id and datamatch the
2010 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2011 the kernel will ignore the length of guest write and may get a faster vmexit.
2012 The speedup may only apply to specific architectures, but the ioeventfd will
2018 :Capability: KVM_CAP_SW_TLB
2021 :Parameters: struct kvm_dirty_tlb (in)
2022 :Returns: 0 on success, -1 on error
2026 struct kvm_dirty_tlb {
2031 This must be called whenever userspace has changed an entry in the shared
2032 TLB, prior to calling KVM_RUN on the associated vcpu.
2034 The "bitmap" field is the userspace address of an array. This array
2035 consists of a number of bits, equal to the total number of TLB entries as
2036 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
2037 nearest multiple of 64.
2039 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2042 The array is little-endian: the bit 0 is the least significant bit of the
2043 first byte, bit 8 is the least significant bit of the second byte, etc.
2044 This avoids any complications with differing word sizes.
2046 The "num_dirty" field is a performance hint for KVM to determine whether it
2047 should skip processing the bitmap and just invalidate everything. It must
2048 be set to the number of set bits in the bitmap.
2051 4.62 KVM_CREATE_SPAPR_TCE
2052 -------------------------
2054 :Capability: KVM_CAP_SPAPR_TCE
2055 :Architectures: powerpc
2057 :Parameters: struct kvm_create_spapr_tce (in)
2058 :Returns: file descriptor for manipulating the created TCE table
2060 This creates a virtual TCE (translation control entry) table, which
2061 is an IOMMU for PAPR-style virtual I/O. It is used to translate
2062 logical addresses used in virtual I/O into guest physical addresses,
2063 and provides a scatter/gather capability for PAPR virtual I/O.
2067 /* for KVM_CAP_SPAPR_TCE */
2068 struct kvm_create_spapr_tce {
2073 The liobn field gives the logical IO bus number for which to create a
2074 TCE table. The window_size field specifies the size of the DMA window
2075 which this TCE table will translate - the table will contain one 64
2076 bit TCE entry for every 4kiB of the DMA window.
2078 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2079 table has been created using this ioctl(), the kernel will handle it
2080 in real mode, updating the TCE table. H_PUT_TCE calls for other
2081 liobns will cause a vm exit and must be handled by userspace.
2083 The return value is a file descriptor which can be passed to mmap(2)
2084 to map the created TCE table into userspace. This lets userspace read
2085 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2086 userspace update the TCE table directly which is useful in some
2090 4.63 KVM_ALLOCATE_RMA
2091 ---------------------
2093 :Capability: KVM_CAP_PPC_RMA
2094 :Architectures: powerpc
2096 :Parameters: struct kvm_allocate_rma (out)
2097 :Returns: file descriptor for mapping the allocated RMA
2099 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2100 time by the kernel. An RMA is a physically-contiguous, aligned region
2101 of memory used on older POWER processors to provide the memory which
2102 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2103 POWER processors support a set of sizes for the RMA that usually
2104 includes 64MB, 128MB, 256MB and some larger powers of two.
2108 /* for KVM_ALLOCATE_RMA */
2109 struct kvm_allocate_rma {
2113 The return value is a file descriptor which can be passed to mmap(2)
2114 to map the allocated RMA into userspace. The mapped area can then be
2115 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2116 RMA for a virtual machine. The size of the RMA in bytes (which is
2117 fixed at host kernel boot time) is returned in the rma_size field of
2118 the argument structure.
2120 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2121 is supported; 2 if the processor requires all virtual machines to have
2122 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2123 because it supports the Virtual RMA (VRMA) facility.
2129 :Capability: KVM_CAP_USER_NMI
2133 :Returns: 0 on success, -1 on error
2135 Queues an NMI on the thread's vcpu. Note this is well defined only
2136 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2137 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2138 has been called, this interface is completely emulated within the kernel.
2140 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2141 following algorithm:
2144 - read the local APIC's state (KVM_GET_LAPIC)
2145 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2146 - if so, issue KVM_NMI
2149 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2153 4.65 KVM_S390_UCAS_MAP
2154 ----------------------
2156 :Capability: KVM_CAP_S390_UCONTROL
2157 :Architectures: s390
2159 :Parameters: struct kvm_s390_ucas_mapping (in)
2160 :Returns: 0 in case of success
2162 The parameter is defined like this::
2164 struct kvm_s390_ucas_mapping {
2170 This ioctl maps the memory at "user_addr" with the length "length" to
2171 the vcpu's address space starting at "vcpu_addr". All parameters need to
2172 be aligned by 1 megabyte.
2175 4.66 KVM_S390_UCAS_UNMAP
2176 ------------------------
2178 :Capability: KVM_CAP_S390_UCONTROL
2179 :Architectures: s390
2181 :Parameters: struct kvm_s390_ucas_mapping (in)
2182 :Returns: 0 in case of success
2184 The parameter is defined like this::
2186 struct kvm_s390_ucas_mapping {
2192 This ioctl unmaps the memory in the vcpu's address space starting at
2193 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2194 All parameters need to be aligned by 1 megabyte.
2197 4.67 KVM_S390_VCPU_FAULT
2198 ------------------------
2200 :Capability: KVM_CAP_S390_UCONTROL
2201 :Architectures: s390
2203 :Parameters: vcpu absolute address (in)
2204 :Returns: 0 in case of success
2206 This call creates a page table entry on the virtual cpu's address space
2207 (for user controlled virtual machines) or the virtual machine's address
2208 space (for regular virtual machines). This only works for minor faults,
2209 thus it's recommended to access subject memory page via the user page
2210 table upfront. This is useful to handle validity intercepts for user
2211 controlled virtual machines to fault in the virtual cpu's lowcore pages
2212 prior to calling the KVM_RUN ioctl.
2215 4.68 KVM_SET_ONE_REG
2216 --------------------
2218 :Capability: KVM_CAP_ONE_REG
2221 :Parameters: struct kvm_one_reg (in)
2222 :Returns: 0 on success, negative value on failure
2226 ====== ============================================================
2227 ENOENT no such register
2228 EINVAL invalid register ID, or no such register or used with VMs in
2229 protected virtualization mode on s390
2230 EPERM (arm64) register access not allowed before vcpu finalization
2231 ====== ============================================================
2233 (These error codes are indicative only: do not rely on a specific error
2234 code being returned in a specific situation.)
2238 struct kvm_one_reg {
2243 Using this ioctl, a single vcpu register can be set to a specific value
2244 defined by user space with the passed in struct kvm_one_reg, where id
2245 refers to the register identifier as described below and addr is a pointer
2246 to a variable with the respective size. There can be architecture agnostic
2247 and architecture specific registers. Each have their own range of operation
2248 and their own constants and width. To keep track of the implemented
2249 registers, find a list below:
2251 ======= =============================== ============
2252 Arch Register Width (bits)
2253 ======= =============================== ============
2254 PPC KVM_REG_PPC_HIOR 64
2255 PPC KVM_REG_PPC_IAC1 64
2256 PPC KVM_REG_PPC_IAC2 64
2257 PPC KVM_REG_PPC_IAC3 64
2258 PPC KVM_REG_PPC_IAC4 64
2259 PPC KVM_REG_PPC_DAC1 64
2260 PPC KVM_REG_PPC_DAC2 64
2261 PPC KVM_REG_PPC_DABR 64
2262 PPC KVM_REG_PPC_DSCR 64
2263 PPC KVM_REG_PPC_PURR 64
2264 PPC KVM_REG_PPC_SPURR 64
2265 PPC KVM_REG_PPC_DAR 64
2266 PPC KVM_REG_PPC_DSISR 32
2267 PPC KVM_REG_PPC_AMR 64
2268 PPC KVM_REG_PPC_UAMOR 64
2269 PPC KVM_REG_PPC_MMCR0 64
2270 PPC KVM_REG_PPC_MMCR1 64
2271 PPC KVM_REG_PPC_MMCRA 64
2272 PPC KVM_REG_PPC_MMCR2 64
2273 PPC KVM_REG_PPC_MMCRS 64
2274 PPC KVM_REG_PPC_MMCR3 64
2275 PPC KVM_REG_PPC_SIAR 64
2276 PPC KVM_REG_PPC_SDAR 64
2277 PPC KVM_REG_PPC_SIER 64
2278 PPC KVM_REG_PPC_SIER2 64
2279 PPC KVM_REG_PPC_SIER3 64
2280 PPC KVM_REG_PPC_PMC1 32
2281 PPC KVM_REG_PPC_PMC2 32
2282 PPC KVM_REG_PPC_PMC3 32
2283 PPC KVM_REG_PPC_PMC4 32
2284 PPC KVM_REG_PPC_PMC5 32
2285 PPC KVM_REG_PPC_PMC6 32
2286 PPC KVM_REG_PPC_PMC7 32
2287 PPC KVM_REG_PPC_PMC8 32
2288 PPC KVM_REG_PPC_FPR0 64
2290 PPC KVM_REG_PPC_FPR31 64
2291 PPC KVM_REG_PPC_VR0 128
2293 PPC KVM_REG_PPC_VR31 128
2294 PPC KVM_REG_PPC_VSR0 128
2296 PPC KVM_REG_PPC_VSR31 128
2297 PPC KVM_REG_PPC_FPSCR 64
2298 PPC KVM_REG_PPC_VSCR 32
2299 PPC KVM_REG_PPC_VPA_ADDR 64
2300 PPC KVM_REG_PPC_VPA_SLB 128
2301 PPC KVM_REG_PPC_VPA_DTL 128
2302 PPC KVM_REG_PPC_EPCR 32
2303 PPC KVM_REG_PPC_EPR 32
2304 PPC KVM_REG_PPC_TCR 32
2305 PPC KVM_REG_PPC_TSR 32
2306 PPC KVM_REG_PPC_OR_TSR 32
2307 PPC KVM_REG_PPC_CLEAR_TSR 32
2308 PPC KVM_REG_PPC_MAS0 32
2309 PPC KVM_REG_PPC_MAS1 32
2310 PPC KVM_REG_PPC_MAS2 64
2311 PPC KVM_REG_PPC_MAS7_3 64
2312 PPC KVM_REG_PPC_MAS4 32
2313 PPC KVM_REG_PPC_MAS6 32
2314 PPC KVM_REG_PPC_MMUCFG 32
2315 PPC KVM_REG_PPC_TLB0CFG 32
2316 PPC KVM_REG_PPC_TLB1CFG 32
2317 PPC KVM_REG_PPC_TLB2CFG 32
2318 PPC KVM_REG_PPC_TLB3CFG 32
2319 PPC KVM_REG_PPC_TLB0PS 32
2320 PPC KVM_REG_PPC_TLB1PS 32
2321 PPC KVM_REG_PPC_TLB2PS 32
2322 PPC KVM_REG_PPC_TLB3PS 32
2323 PPC KVM_REG_PPC_EPTCFG 32
2324 PPC KVM_REG_PPC_ICP_STATE 64
2325 PPC KVM_REG_PPC_VP_STATE 128
2326 PPC KVM_REG_PPC_TB_OFFSET 64
2327 PPC KVM_REG_PPC_SPMC1 32
2328 PPC KVM_REG_PPC_SPMC2 32
2329 PPC KVM_REG_PPC_IAMR 64
2330 PPC KVM_REG_PPC_TFHAR 64
2331 PPC KVM_REG_PPC_TFIAR 64
2332 PPC KVM_REG_PPC_TEXASR 64
2333 PPC KVM_REG_PPC_FSCR 64
2334 PPC KVM_REG_PPC_PSPB 32
2335 PPC KVM_REG_PPC_EBBHR 64
2336 PPC KVM_REG_PPC_EBBRR 64
2337 PPC KVM_REG_PPC_BESCR 64
2338 PPC KVM_REG_PPC_TAR 64
2339 PPC KVM_REG_PPC_DPDES 64
2340 PPC KVM_REG_PPC_DAWR 64
2341 PPC KVM_REG_PPC_DAWRX 64
2342 PPC KVM_REG_PPC_CIABR 64
2343 PPC KVM_REG_PPC_IC 64
2344 PPC KVM_REG_PPC_VTB 64
2345 PPC KVM_REG_PPC_CSIGR 64
2346 PPC KVM_REG_PPC_TACR 64
2347 PPC KVM_REG_PPC_TCSCR 64
2348 PPC KVM_REG_PPC_PID 64
2349 PPC KVM_REG_PPC_ACOP 64
2350 PPC KVM_REG_PPC_VRSAVE 32
2351 PPC KVM_REG_PPC_LPCR 32
2352 PPC KVM_REG_PPC_LPCR_64 64
2353 PPC KVM_REG_PPC_PPR 64
2354 PPC KVM_REG_PPC_ARCH_COMPAT 32
2355 PPC KVM_REG_PPC_DABRX 32
2356 PPC KVM_REG_PPC_WORT 64
2357 PPC KVM_REG_PPC_SPRG9 64
2358 PPC KVM_REG_PPC_DBSR 32
2359 PPC KVM_REG_PPC_TIDR 64
2360 PPC KVM_REG_PPC_PSSCR 64
2361 PPC KVM_REG_PPC_DEC_EXPIRY 64
2362 PPC KVM_REG_PPC_PTCR 64
2363 PPC KVM_REG_PPC_DAWR1 64
2364 PPC KVM_REG_PPC_DAWRX1 64
2365 PPC KVM_REG_PPC_TM_GPR0 64
2367 PPC KVM_REG_PPC_TM_GPR31 64
2368 PPC KVM_REG_PPC_TM_VSR0 128
2370 PPC KVM_REG_PPC_TM_VSR63 128
2371 PPC KVM_REG_PPC_TM_CR 64
2372 PPC KVM_REG_PPC_TM_LR 64
2373 PPC KVM_REG_PPC_TM_CTR 64
2374 PPC KVM_REG_PPC_TM_FPSCR 64
2375 PPC KVM_REG_PPC_TM_AMR 64
2376 PPC KVM_REG_PPC_TM_PPR 64
2377 PPC KVM_REG_PPC_TM_VRSAVE 64
2378 PPC KVM_REG_PPC_TM_VSCR 32
2379 PPC KVM_REG_PPC_TM_DSCR 64
2380 PPC KVM_REG_PPC_TM_TAR 64
2381 PPC KVM_REG_PPC_TM_XER 64
2383 MIPS KVM_REG_MIPS_R0 64
2385 MIPS KVM_REG_MIPS_R31 64
2386 MIPS KVM_REG_MIPS_HI 64
2387 MIPS KVM_REG_MIPS_LO 64
2388 MIPS KVM_REG_MIPS_PC 64
2389 MIPS KVM_REG_MIPS_CP0_INDEX 32
2390 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2391 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2392 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2393 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2394 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2395 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2396 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2397 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2398 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2399 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2400 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2401 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2402 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2403 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2404 MIPS KVM_REG_MIPS_CP0_WIRED 32
2405 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2406 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2407 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2408 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2409 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2410 MIPS KVM_REG_MIPS_CP0_COUNT 32
2411 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2412 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2413 MIPS KVM_REG_MIPS_CP0_STATUS 32
2414 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2415 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2416 MIPS KVM_REG_MIPS_CP0_EPC 64
2417 MIPS KVM_REG_MIPS_CP0_PRID 32
2418 MIPS KVM_REG_MIPS_CP0_EBASE 64
2419 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2420 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2421 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2422 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2423 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2424 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2425 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2426 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2427 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2428 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2429 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2430 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2431 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2432 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2433 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2434 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2435 MIPS KVM_REG_MIPS_COUNT_CTL 64
2436 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2437 MIPS KVM_REG_MIPS_COUNT_HZ 64
2438 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2439 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2440 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2441 MIPS KVM_REG_MIPS_FCR_IR 32
2442 MIPS KVM_REG_MIPS_FCR_CSR 32
2443 MIPS KVM_REG_MIPS_MSA_IR 32
2444 MIPS KVM_REG_MIPS_MSA_CSR 32
2445 ======= =============================== ============
2447 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2448 is the register group type, or coprocessor number:
2450 ARM core registers have the following id bit patterns::
2452 0x4020 0000 0010 <index into the kvm_regs struct:16>
2454 ARM 32-bit CP15 registers have the following id bit patterns::
2456 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2458 ARM 64-bit CP15 registers have the following id bit patterns::
2460 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2462 ARM CCSIDR registers are demultiplexed by CSSELR value::
2464 0x4020 0000 0011 00 <csselr:8>
2466 ARM 32-bit VFP control registers have the following id bit patterns::
2468 0x4020 0000 0012 1 <regno:12>
2470 ARM 64-bit FP registers have the following id bit patterns::
2472 0x4030 0000 0012 0 <regno:12>
2474 ARM firmware pseudo-registers have the following bit pattern::
2476 0x4030 0000 0014 <regno:16>
2479 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2480 that is the register group type, or coprocessor number:
2482 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2483 that the size of the access is variable, as the kvm_regs structure
2484 contains elements ranging from 32 to 128 bits. The index is a 32bit
2485 value in the kvm_regs structure seen as a 32bit array::
2487 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2491 ======================= ========= ===== =======================================
2492 Encoding Register Bits kvm_regs member
2493 ======================= ========= ===== =======================================
2494 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2495 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2497 0x6030 0000 0010 003c X30 64 regs.regs[30]
2498 0x6030 0000 0010 003e SP 64 regs.sp
2499 0x6030 0000 0010 0040 PC 64 regs.pc
2500 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2501 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2502 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2503 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2504 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2505 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2506 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2507 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2508 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2509 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2511 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2512 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2513 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2514 ======================= ========= ===== =======================================
2516 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2519 The equivalent register content can be accessed via bits [127:0] of
2520 the corresponding SVE Zn registers instead for vcpus that have SVE
2521 enabled (see below).
2523 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2525 0x6020 0000 0011 00 <csselr:8>
2527 arm64 system registers have the following id bit patterns::
2529 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2533 Two system register IDs do not follow the specified pattern. These
2534 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2535 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2536 two had their values accidentally swapped, which means TIMER_CVAL is
2537 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2538 derived from the register encoding for CNTV_CVAL_EL0. As this is
2539 API, it must remain this way.
2541 arm64 firmware pseudo-registers have the following bit pattern::
2543 0x6030 0000 0014 <regno:16>
2545 arm64 SVE registers have the following bit patterns::
2547 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2548 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2549 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2550 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2552 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2553 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2554 quadwords: see [2]_ below.
2556 These registers are only accessible on vcpus for which SVE is enabled.
2557 See KVM_ARM_VCPU_INIT for details.
2559 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2560 accessible until the vcpu's SVE configuration has been finalized
2561 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2562 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2564 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2565 lengths supported by the vcpu to be discovered and configured by
2566 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2567 or KVM_SET_ONE_REG, the value of this register is of type
2568 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2571 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2573 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2574 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2575 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2576 /* Vector length vq * 16 bytes supported */
2578 /* Vector length vq * 16 bytes not supported */
2580 .. [2] The maximum value vq for which the above condition is true is
2581 max_vq. This is the maximum vector length available to the guest on
2582 this vcpu, and determines which register slices are visible through
2583 this ioctl interface.
2585 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2588 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2589 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2592 Userspace may subsequently modify it if desired until the vcpu's SVE
2593 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2595 Apart from simply removing all vector lengths from the host set that
2596 exceed some value, support for arbitrarily chosen sets of vector lengths
2597 is hardware-dependent and may not be available. Attempting to configure
2598 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2601 After the vcpu's SVE configuration is finalized, further attempts to
2602 write this register will fail with EPERM.
2605 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2606 the register group type:
2608 MIPS core registers (see above) have the following id bit patterns::
2610 0x7030 0000 0000 <reg:16>
2612 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2613 patterns depending on whether they're 32-bit or 64-bit registers::
2615 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2616 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2618 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2619 versions of the EntryLo registers regardless of the word size of the host
2620 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2621 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2622 the PFNX field starting at bit 30.
2624 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2627 0x7030 0000 0001 01 <reg:8>
2629 MIPS KVM control registers (see above) have the following id bit patterns::
2631 0x7030 0000 0002 <reg:16>
2633 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2634 id bit patterns depending on the size of the register being accessed. They are
2635 always accessed according to the current guest FPU mode (Status.FR and
2636 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2637 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2638 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2639 overlap the FPU registers::
2641 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2642 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2643 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2645 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2646 following id bit patterns::
2648 0x7020 0000 0003 01 <0:3> <reg:5>
2650 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2651 following id bit patterns::
2653 0x7020 0000 0003 02 <0:3> <reg:5>
2655 RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of
2656 that is the register group type.
2658 RISC-V config registers are meant for configuring a Guest VCPU and it has
2659 the following id bit patterns::
2661 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host)
2662 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host)
2664 Following are the RISC-V config registers:
2666 ======================= ========= =============================================
2667 Encoding Register Description
2668 ======================= ========= =============================================
2669 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU
2670 ======================= ========= =============================================
2672 The isa config register can be read anytime but can only be written before
2673 a Guest VCPU runs. It will have ISA feature bits matching underlying host
2676 RISC-V core registers represent the general excution state of a Guest VCPU
2677 and it has the following id bit patterns::
2679 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host)
2680 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host)
2682 Following are the RISC-V core registers:
2684 ======================= ========= =============================================
2685 Encoding Register Description
2686 ======================= ========= =============================================
2687 0x80x0 0000 0200 0000 regs.pc Program counter
2688 0x80x0 0000 0200 0001 regs.ra Return address
2689 0x80x0 0000 0200 0002 regs.sp Stack pointer
2690 0x80x0 0000 0200 0003 regs.gp Global pointer
2691 0x80x0 0000 0200 0004 regs.tp Task pointer
2692 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0
2693 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1
2694 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2
2695 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0
2696 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1
2697 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0
2698 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1
2699 0x80x0 0000 0200 000c regs.a2 Function argument 2
2700 0x80x0 0000 0200 000d regs.a3 Function argument 3
2701 0x80x0 0000 0200 000e regs.a4 Function argument 4
2702 0x80x0 0000 0200 000f regs.a5 Function argument 5
2703 0x80x0 0000 0200 0010 regs.a6 Function argument 6
2704 0x80x0 0000 0200 0011 regs.a7 Function argument 7
2705 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2
2706 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3
2707 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4
2708 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5
2709 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6
2710 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7
2711 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8
2712 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9
2713 0x80x0 0000 0200 001a regs.s10 Callee saved register 10
2714 0x80x0 0000 0200 001b regs.s11 Callee saved register 11
2715 0x80x0 0000 0200 001c regs.t3 Caller saved register 3
2716 0x80x0 0000 0200 001d regs.t4 Caller saved register 4
2717 0x80x0 0000 0200 001e regs.t5 Caller saved register 5
2718 0x80x0 0000 0200 001f regs.t6 Caller saved register 6
2719 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode)
2720 ======================= ========= =============================================
2722 RISC-V csr registers represent the supervisor mode control/status registers
2723 of a Guest VCPU and it has the following id bit patterns::
2725 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host)
2726 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host)
2728 Following are the RISC-V csr registers:
2730 ======================= ========= =============================================
2731 Encoding Register Description
2732 ======================= ========= =============================================
2733 0x80x0 0000 0300 0000 sstatus Supervisor status
2734 0x80x0 0000 0300 0001 sie Supervisor interrupt enable
2735 0x80x0 0000 0300 0002 stvec Supervisor trap vector base
2736 0x80x0 0000 0300 0003 sscratch Supervisor scratch register
2737 0x80x0 0000 0300 0004 sepc Supervisor exception program counter
2738 0x80x0 0000 0300 0005 scause Supervisor trap cause
2739 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction
2740 0x80x0 0000 0300 0007 sip Supervisor interrupt pending
2741 0x80x0 0000 0300 0008 satp Supervisor address translation and protection
2742 ======================= ========= =============================================
2744 RISC-V timer registers represent the timer state of a Guest VCPU and it has
2745 the following id bit patterns::
2747 0x8030 0000 04 <index into the kvm_riscv_timer struct:24>
2749 Following are the RISC-V timer registers:
2751 ======================= ========= =============================================
2752 Encoding Register Description
2753 ======================= ========= =============================================
2754 0x8030 0000 0400 0000 frequency Time base frequency (read-only)
2755 0x8030 0000 0400 0001 time Time value visible to Guest
2756 0x8030 0000 0400 0002 compare Time compare programmed by Guest
2757 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF)
2758 ======================= ========= =============================================
2760 RISC-V F-extension registers represent the single precision floating point
2761 state of a Guest VCPU and it has the following id bit patterns::
2763 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24>
2765 Following are the RISC-V F-extension registers:
2767 ======================= ========= =============================================
2768 Encoding Register Description
2769 ======================= ========= =============================================
2770 0x8020 0000 0500 0000 f[0] Floating point register 0
2772 0x8020 0000 0500 001f f[31] Floating point register 31
2773 0x8020 0000 0500 0020 fcsr Floating point control and status register
2774 ======================= ========= =============================================
2776 RISC-V D-extension registers represent the double precision floating point
2777 state of a Guest VCPU and it has the following id bit patterns::
2779 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr)
2780 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr)
2782 Following are the RISC-V D-extension registers:
2784 ======================= ========= =============================================
2785 Encoding Register Description
2786 ======================= ========= =============================================
2787 0x8030 0000 0600 0000 f[0] Floating point register 0
2789 0x8030 0000 0600 001f f[31] Floating point register 31
2790 0x8020 0000 0600 0020 fcsr Floating point control and status register
2791 ======================= ========= =============================================
2794 4.69 KVM_GET_ONE_REG
2795 --------------------
2797 :Capability: KVM_CAP_ONE_REG
2800 :Parameters: struct kvm_one_reg (in and out)
2801 :Returns: 0 on success, negative value on failure
2805 ======== ============================================================
2806 ENOENT no such register
2807 EINVAL invalid register ID, or no such register or used with VMs in
2808 protected virtualization mode on s390
2809 EPERM (arm64) register access not allowed before vcpu finalization
2810 ======== ============================================================
2812 (These error codes are indicative only: do not rely on a specific error
2813 code being returned in a specific situation.)
2815 This ioctl allows to receive the value of a single register implemented
2816 in a vcpu. The register to read is indicated by the "id" field of the
2817 kvm_one_reg struct passed in. On success, the register value can be found
2818 at the memory location pointed to by "addr".
2820 The list of registers accessible using this interface is identical to the
2824 4.70 KVM_KVMCLOCK_CTRL
2825 ----------------------
2827 :Capability: KVM_CAP_KVMCLOCK_CTRL
2828 :Architectures: Any that implement pvclocks (currently x86 only)
2831 :Returns: 0 on success, -1 on error
2833 This ioctl sets a flag accessible to the guest indicating that the specified
2834 vCPU has been paused by the host userspace.
2836 The host will set a flag in the pvclock structure that is checked from the
2837 soft lockup watchdog. The flag is part of the pvclock structure that is
2838 shared between guest and host, specifically the second bit of the flags
2839 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2840 the host and read/cleared exclusively by the guest. The guest operation of
2841 checking and clearing the flag must be an atomic operation so
2842 load-link/store-conditional, or equivalent must be used. There are two cases
2843 where the guest will clear the flag: when the soft lockup watchdog timer resets
2844 itself or when a soft lockup is detected. This ioctl can be called any time
2845 after pausing the vcpu, but before it is resumed.
2851 :Capability: KVM_CAP_SIGNAL_MSI
2852 :Architectures: x86 arm64
2854 :Parameters: struct kvm_msi (in)
2855 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2857 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2872 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2873 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2874 the device ID. If this capability is not available, userspace
2875 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2877 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2878 for the device that wrote the MSI message. For PCI, this is usually a
2879 BFD identifier in the lower 16 bits.
2881 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2882 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2883 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2884 address_hi must be zero.
2887 4.71 KVM_CREATE_PIT2
2888 --------------------
2890 :Capability: KVM_CAP_PIT2
2893 :Parameters: struct kvm_pit_config (in)
2894 :Returns: 0 on success, -1 on error
2896 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2897 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2898 parameters have to be passed::
2900 struct kvm_pit_config {
2907 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2909 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2910 exists, this thread will have a name of the following pattern::
2912 kvm-pit/<owner-process-pid>
2914 When running a guest with elevated priorities, the scheduling parameters of
2915 this thread may have to be adjusted accordingly.
2917 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2923 :Capability: KVM_CAP_PIT_STATE2
2926 :Parameters: struct kvm_pit_state2 (out)
2927 :Returns: 0 on success, -1 on error
2929 Retrieves the state of the in-kernel PIT model. Only valid after
2930 KVM_CREATE_PIT2. The state is returned in the following structure::
2932 struct kvm_pit_state2 {
2933 struct kvm_pit_channel_state channels[3];
2940 /* disable PIT in HPET legacy mode */
2941 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
2943 This IOCTL replaces the obsolete KVM_GET_PIT.
2949 :Capability: KVM_CAP_PIT_STATE2
2952 :Parameters: struct kvm_pit_state2 (in)
2953 :Returns: 0 on success, -1 on error
2955 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2956 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2958 This IOCTL replaces the obsolete KVM_SET_PIT.
2961 4.74 KVM_PPC_GET_SMMU_INFO
2962 --------------------------
2964 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
2965 :Architectures: powerpc
2968 :Returns: 0 on success, -1 on error
2970 This populates and returns a structure describing the features of
2971 the "Server" class MMU emulation supported by KVM.
2972 This can in turn be used by userspace to generate the appropriate
2973 device-tree properties for the guest operating system.
2975 The structure contains some global information, followed by an
2976 array of supported segment page sizes::
2978 struct kvm_ppc_smmu_info {
2982 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2985 The supported flags are:
2987 - KVM_PPC_PAGE_SIZES_REAL:
2988 When that flag is set, guest page sizes must "fit" the backing
2989 store page sizes. When not set, any page size in the list can
2990 be used regardless of how they are backed by userspace.
2992 - KVM_PPC_1T_SEGMENTS
2993 The emulated MMU supports 1T segments in addition to the
2997 This flag indicates that HPT guests are not supported by KVM,
2998 thus all guests must use radix MMU mode.
3000 The "slb_size" field indicates how many SLB entries are supported
3002 The "sps" array contains 8 entries indicating the supported base
3003 page sizes for a segment in increasing order. Each entry is defined
3006 struct kvm_ppc_one_seg_page_size {
3007 __u32 page_shift; /* Base page shift of segment (or 0) */
3008 __u32 slb_enc; /* SLB encoding for BookS */
3009 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3012 An entry with a "page_shift" of 0 is unused. Because the array is
3013 organized in increasing order, a lookup can stop when encoutering
3016 The "slb_enc" field provides the encoding to use in the SLB for the
3017 page size. The bits are in positions such as the value can directly
3018 be OR'ed into the "vsid" argument of the slbmte instruction.
3020 The "enc" array is a list which for each of those segment base page
3021 size provides the list of supported actual page sizes (which can be
3022 only larger or equal to the base page size), along with the
3023 corresponding encoding in the hash PTE. Similarly, the array is
3024 8 entries sorted by increasing sizes and an entry with a "0" shift
3025 is an empty entry and a terminator::
3027 struct kvm_ppc_one_page_size {
3028 __u32 page_shift; /* Page shift (or 0) */
3029 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
3032 The "pte_enc" field provides a value that can OR'ed into the hash
3033 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3034 into the hash PTE second double word).
3039 :Capability: KVM_CAP_IRQFD
3040 :Architectures: x86 s390 arm64
3042 :Parameters: struct kvm_irqfd (in)
3043 :Returns: 0 on success, -1 on error
3045 Allows setting an eventfd to directly trigger a guest interrupt.
3046 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3047 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
3048 an event is triggered on the eventfd, an interrupt is injected into
3049 the guest using the specified gsi pin. The irqfd is removed using
3050 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3053 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3054 mechanism allowing emulation of level-triggered, irqfd-based
3055 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3056 additional eventfd in the kvm_irqfd.resamplefd field. When operating
3057 in resample mode, posting of an interrupt through kvm_irq.fd asserts
3058 the specified gsi in the irqchip. When the irqchip is resampled, such
3059 as from an EOI, the gsi is de-asserted and the user is notified via
3060 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
3061 the interrupt if the device making use of it still requires service.
3062 Note that closing the resamplefd is not sufficient to disable the
3063 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3064 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3066 On arm64, gsi routing being supported, the following can happen:
3068 - in case no routing entry is associated to this gsi, injection fails
3069 - in case the gsi is associated to an irqchip routing entry,
3070 irqchip.pin + 32 corresponds to the injected SPI ID.
3071 - in case the gsi is associated to an MSI routing entry, the MSI
3072 message and device ID are translated into an LPI (support restricted
3073 to GICv3 ITS in-kernel emulation).
3075 4.76 KVM_PPC_ALLOCATE_HTAB
3076 --------------------------
3078 :Capability: KVM_CAP_PPC_ALLOC_HTAB
3079 :Architectures: powerpc
3081 :Parameters: Pointer to u32 containing hash table order (in/out)
3082 :Returns: 0 on success, -1 on error
3084 This requests the host kernel to allocate an MMU hash table for a
3085 guest using the PAPR paravirtualization interface. This only does
3086 anything if the kernel is configured to use the Book 3S HV style of
3087 virtualization. Otherwise the capability doesn't exist and the ioctl
3088 returns an ENOTTY error. The rest of this description assumes Book 3S
3091 There must be no vcpus running when this ioctl is called; if there
3092 are, it will do nothing and return an EBUSY error.
3094 The parameter is a pointer to a 32-bit unsigned integer variable
3095 containing the order (log base 2) of the desired size of the hash
3096 table, which must be between 18 and 46. On successful return from the
3097 ioctl, the value will not be changed by the kernel.
3099 If no hash table has been allocated when any vcpu is asked to run
3100 (with the KVM_RUN ioctl), the host kernel will allocate a
3101 default-sized hash table (16 MB).
3103 If this ioctl is called when a hash table has already been allocated,
3104 with a different order from the existing hash table, the existing hash
3105 table will be freed and a new one allocated. If this is ioctl is
3106 called when a hash table has already been allocated of the same order
3107 as specified, the kernel will clear out the existing hash table (zero
3108 all HPTEs). In either case, if the guest is using the virtualized
3109 real-mode area (VRMA) facility, the kernel will re-create the VMRA
3110 HPTEs on the next KVM_RUN of any vcpu.
3112 4.77 KVM_S390_INTERRUPT
3113 -----------------------
3116 :Architectures: s390
3117 :Type: vm ioctl, vcpu ioctl
3118 :Parameters: struct kvm_s390_interrupt (in)
3119 :Returns: 0 on success, -1 on error
3121 Allows to inject an interrupt to the guest. Interrupts can be floating
3122 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3124 Interrupt parameters are passed via kvm_s390_interrupt::
3126 struct kvm_s390_interrupt {
3132 type can be one of the following:
3134 KVM_S390_SIGP_STOP (vcpu)
3135 - sigp stop; optional flags in parm
3136 KVM_S390_PROGRAM_INT (vcpu)
3137 - program check; code in parm
3138 KVM_S390_SIGP_SET_PREFIX (vcpu)
3139 - sigp set prefix; prefix address in parm
3140 KVM_S390_RESTART (vcpu)
3142 KVM_S390_INT_CLOCK_COMP (vcpu)
3143 - clock comparator interrupt
3144 KVM_S390_INT_CPU_TIMER (vcpu)
3145 - CPU timer interrupt
3146 KVM_S390_INT_VIRTIO (vm)
3147 - virtio external interrupt; external interrupt
3148 parameters in parm and parm64
3149 KVM_S390_INT_SERVICE (vm)
3150 - sclp external interrupt; sclp parameter in parm
3151 KVM_S390_INT_EMERGENCY (vcpu)
3152 - sigp emergency; source cpu in parm
3153 KVM_S390_INT_EXTERNAL_CALL (vcpu)
3154 - sigp external call; source cpu in parm
3155 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3156 - compound value to indicate an
3157 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3158 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3159 interruption subclass)
3160 KVM_S390_MCHK (vm, vcpu)
3161 - machine check interrupt; cr 14 bits in parm, machine check interrupt
3162 code in parm64 (note that machine checks needing further payload are not
3163 supported by this ioctl)
3165 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3167 4.78 KVM_PPC_GET_HTAB_FD
3168 ------------------------
3170 :Capability: KVM_CAP_PPC_HTAB_FD
3171 :Architectures: powerpc
3173 :Parameters: Pointer to struct kvm_get_htab_fd (in)
3174 :Returns: file descriptor number (>= 0) on success, -1 on error
3176 This returns a file descriptor that can be used either to read out the
3177 entries in the guest's hashed page table (HPT), or to write entries to
3178 initialize the HPT. The returned fd can only be written to if the
3179 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3180 can only be read if that bit is clear. The argument struct looks like
3183 /* For KVM_PPC_GET_HTAB_FD */
3184 struct kvm_get_htab_fd {
3190 /* Values for kvm_get_htab_fd.flags */
3191 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
3192 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
3194 The 'start_index' field gives the index in the HPT of the entry at
3195 which to start reading. It is ignored when writing.
3197 Reads on the fd will initially supply information about all
3198 "interesting" HPT entries. Interesting entries are those with the
3199 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3200 all entries. When the end of the HPT is reached, the read() will
3201 return. If read() is called again on the fd, it will start again from
3202 the beginning of the HPT, but will only return HPT entries that have
3203 changed since they were last read.
3205 Data read or written is structured as a header (8 bytes) followed by a
3206 series of valid HPT entries (16 bytes) each. The header indicates how
3207 many valid HPT entries there are and how many invalid entries follow
3208 the valid entries. The invalid entries are not represented explicitly
3209 in the stream. The header format is::
3211 struct kvm_get_htab_header {
3217 Writes to the fd create HPT entries starting at the index given in the
3218 header; first 'n_valid' valid entries with contents from the data
3219 written, then 'n_invalid' invalid entries, invalidating any previously
3220 valid entries found.
3222 4.79 KVM_CREATE_DEVICE
3223 ----------------------
3225 :Capability: KVM_CAP_DEVICE_CTRL
3227 :Parameters: struct kvm_create_device (in/out)
3228 :Returns: 0 on success, -1 on error
3232 ====== =======================================================
3233 ENODEV The device type is unknown or unsupported
3234 EEXIST Device already created, and this type of device may not
3235 be instantiated multiple times
3236 ====== =======================================================
3238 Other error conditions may be defined by individual device types or
3239 have their standard meanings.
3241 Creates an emulated device in the kernel. The file descriptor returned
3242 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3244 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3245 device type is supported (not necessarily whether it can be created
3248 Individual devices should not define flags. Attributes should be used
3249 for specifying any behavior that is not implied by the device type
3254 struct kvm_create_device {
3255 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3256 __u32 fd; /* out: device handle */
3257 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3260 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3261 --------------------------------------------
3263 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3264 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3265 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set)
3266 :Type: device ioctl, vm ioctl, vcpu ioctl
3267 :Parameters: struct kvm_device_attr
3268 :Returns: 0 on success, -1 on error
3272 ===== =============================================================
3273 ENXIO The group or attribute is unknown/unsupported for this device
3274 or hardware support is missing.
3275 EPERM The attribute cannot (currently) be accessed this way
3276 (e.g. read-only attribute, or attribute that only makes
3277 sense when the device is in a different state)
3278 ===== =============================================================
3280 Other error conditions may be defined by individual device types.
3282 Gets/sets a specified piece of device configuration and/or state. The
3283 semantics are device-specific. See individual device documentation in
3284 the "devices" directory. As with ONE_REG, the size of the data
3285 transferred is defined by the particular attribute.
3289 struct kvm_device_attr {
3290 __u32 flags; /* no flags currently defined */
3291 __u32 group; /* device-defined */
3292 __u64 attr; /* group-defined */
3293 __u64 addr; /* userspace address of attr data */
3296 4.81 KVM_HAS_DEVICE_ATTR
3297 ------------------------
3299 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3300 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3301 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device
3302 :Type: device ioctl, vm ioctl, vcpu ioctl
3303 :Parameters: struct kvm_device_attr
3304 :Returns: 0 on success, -1 on error
3308 ===== =============================================================
3309 ENXIO The group or attribute is unknown/unsupported for this device
3310 or hardware support is missing.
3311 ===== =============================================================
3313 Tests whether a device supports a particular attribute. A successful
3314 return indicates the attribute is implemented. It does not necessarily
3315 indicate that the attribute can be read or written in the device's
3316 current state. "addr" is ignored.
3318 4.82 KVM_ARM_VCPU_INIT
3319 ----------------------
3322 :Architectures: arm64
3324 :Parameters: struct kvm_vcpu_init (in)
3325 :Returns: 0 on success; -1 on error
3329 ====== =================================================================
3330 EINVAL the target is unknown, or the combination of features is invalid.
3331 ENOENT a features bit specified is unknown.
3332 ====== =================================================================
3334 This tells KVM what type of CPU to present to the guest, and what
3335 optional features it should have. This will cause a reset of the cpu
3336 registers to their initial values. If this is not called, KVM_RUN will
3337 return ENOEXEC for that vcpu.
3339 The initial values are defined as:
3341 * AArch64: EL1h, D, A, I and F bits set. All other bits
3343 * AArch32: SVC, A, I and F bits set. All other bits are
3345 - General Purpose registers, including PC and SP: set to 0
3346 - FPSIMD/NEON registers: set to 0
3347 - SVE registers: set to 0
3348 - System registers: Reset to their architecturally defined
3349 values as for a warm reset to EL1 (resp. SVC)
3351 Note that because some registers reflect machine topology, all vcpus
3352 should be created before this ioctl is invoked.
3354 Userspace can call this function multiple times for a given vcpu, including
3355 after the vcpu has been run. This will reset the vcpu to its initial
3356 state. All calls to this function after the initial call must use the same
3357 target and same set of feature flags, otherwise EINVAL will be returned.
3361 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3362 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3363 and execute guest code when KVM_RUN is called.
3364 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3365 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3366 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3367 backward compatible with v0.2) for the CPU.
3368 Depends on KVM_CAP_ARM_PSCI_0_2.
3369 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3370 Depends on KVM_CAP_ARM_PMU_V3.
3372 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3374 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3375 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3376 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3377 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3380 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3382 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3383 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3384 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3385 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3388 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3389 Depends on KVM_CAP_ARM_SVE.
3390 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3392 * After KVM_ARM_VCPU_INIT:
3394 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3395 initial value of this pseudo-register indicates the best set of
3396 vector lengths possible for a vcpu on this host.
3398 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3400 - KVM_RUN and KVM_GET_REG_LIST are not available;
3402 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3403 the scalable archietctural SVE registers
3404 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3405 KVM_REG_ARM64_SVE_FFR;
3407 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3408 KVM_SET_ONE_REG, to modify the set of vector lengths available
3411 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3413 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3414 no longer be written using KVM_SET_ONE_REG.
3416 4.83 KVM_ARM_PREFERRED_TARGET
3417 -----------------------------
3420 :Architectures: arm64
3422 :Parameters: struct kvm_vcpu_init (out)
3423 :Returns: 0 on success; -1 on error
3427 ====== ==========================================
3428 ENODEV no preferred target available for the host
3429 ====== ==========================================
3431 This queries KVM for preferred CPU target type which can be emulated
3432 by KVM on underlying host.
3434 The ioctl returns struct kvm_vcpu_init instance containing information
3435 about preferred CPU target type and recommended features for it. The
3436 kvm_vcpu_init->features bitmap returned will have feature bits set if
3437 the preferred target recommends setting these features, but this is
3440 The information returned by this ioctl can be used to prepare an instance
3441 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3442 VCPU matching underlying host.
3445 4.84 KVM_GET_REG_LIST
3446 ---------------------
3449 :Architectures: arm64, mips
3451 :Parameters: struct kvm_reg_list (in/out)
3452 :Returns: 0 on success; -1 on error
3456 ===== ==============================================================
3457 E2BIG the reg index list is too big to fit in the array specified by
3458 the user (the number required will be written into n).
3459 ===== ==============================================================
3463 struct kvm_reg_list {
3464 __u64 n; /* number of registers in reg[] */
3468 This ioctl returns the guest registers that are supported for the
3469 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3472 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3473 -----------------------------------------
3475 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3476 :Architectures: arm64
3478 :Parameters: struct kvm_arm_device_address (in)
3479 :Returns: 0 on success, -1 on error
3483 ====== ============================================
3484 ENODEV The device id is unknown
3485 ENXIO Device not supported on current system
3486 EEXIST Address already set
3487 E2BIG Address outside guest physical address space
3488 EBUSY Address overlaps with other device range
3489 ====== ============================================
3493 struct kvm_arm_device_addr {
3498 Specify a device address in the guest's physical address space where guests
3499 can access emulated or directly exposed devices, which the host kernel needs
3500 to know about. The id field is an architecture specific identifier for a
3503 arm64 divides the id field into two parts, a device id and an
3504 address type id specific to the individual device::
3506 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3507 field: | 0x00000000 | device id | addr type id |
3509 arm64 currently only require this when using the in-kernel GIC
3510 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3511 as the device id. When setting the base address for the guest's
3512 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3513 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3514 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3515 base addresses will return -EEXIST.
3517 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3518 should be used instead.
3521 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3522 ------------------------------
3524 :Capability: KVM_CAP_PPC_RTAS
3527 :Parameters: struct kvm_rtas_token_args
3528 :Returns: 0 on success, -1 on error
3530 Defines a token value for a RTAS (Run Time Abstraction Services)
3531 service in order to allow it to be handled in the kernel. The
3532 argument struct gives the name of the service, which must be the name
3533 of a service that has a kernel-side implementation. If the token
3534 value is non-zero, it will be associated with that service, and
3535 subsequent RTAS calls by the guest specifying that token will be
3536 handled by the kernel. If the token value is 0, then any token
3537 associated with the service will be forgotten, and subsequent RTAS
3538 calls by the guest for that service will be passed to userspace to be
3541 4.87 KVM_SET_GUEST_DEBUG
3542 ------------------------
3544 :Capability: KVM_CAP_SET_GUEST_DEBUG
3545 :Architectures: x86, s390, ppc, arm64
3547 :Parameters: struct kvm_guest_debug (in)
3548 :Returns: 0 on success; -1 on error
3552 struct kvm_guest_debug {
3555 struct kvm_guest_debug_arch arch;
3558 Set up the processor specific debug registers and configure vcpu for
3559 handling guest debug events. There are two parts to the structure, the
3560 first a control bitfield indicates the type of debug events to handle
3561 when running. Common control bits are:
3563 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3564 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3566 The top 16 bits of the control field are architecture specific control
3567 flags which can include the following:
3569 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3570 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390]
3571 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64]
3572 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3573 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3574 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3575 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86]
3577 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3578 are enabled in memory so we need to ensure breakpoint exceptions are
3579 correctly trapped and the KVM run loop exits at the breakpoint and not
3580 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3581 we need to ensure the guest vCPUs architecture specific registers are
3582 updated to the correct (supplied) values.
3584 The second part of the structure is architecture specific and
3585 typically contains a set of debug registers.
3587 For arm64 the number of debug registers is implementation defined and
3588 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3589 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3590 indicating the number of supported registers.
3592 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3593 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3595 Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3596 supported KVM_GUESTDBG_* bits in the control field.
3598 When debug events exit the main run loop with the reason
3599 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3600 structure containing architecture specific debug information.
3602 4.88 KVM_GET_EMULATED_CPUID
3603 ---------------------------
3605 :Capability: KVM_CAP_EXT_EMUL_CPUID
3608 :Parameters: struct kvm_cpuid2 (in/out)
3609 :Returns: 0 on success, -1 on error
3616 struct kvm_cpuid_entry2 entries[0];
3619 The member 'flags' is used for passing flags from userspace.
3623 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3624 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3625 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3627 struct kvm_cpuid_entry2 {
3638 This ioctl returns x86 cpuid features which are emulated by
3639 kvm.Userspace can use the information returned by this ioctl to query
3640 which features are emulated by kvm instead of being present natively.
3642 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3643 structure with the 'nent' field indicating the number of entries in
3644 the variable-size array 'entries'. If the number of entries is too low
3645 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3646 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3647 is returned. If the number is just right, the 'nent' field is adjusted
3648 to the number of valid entries in the 'entries' array, which is then
3651 The entries returned are the set CPUID bits of the respective features
3652 which kvm emulates, as returned by the CPUID instruction, with unknown
3653 or unsupported feature bits cleared.
3655 Features like x2apic, for example, may not be present in the host cpu
3656 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3657 emulated efficiently and thus not included here.
3659 The fields in each entry are defined as follows:
3662 the eax value used to obtain the entry
3664 the ecx value used to obtain the entry (for entries that are
3667 an OR of zero or more of the following:
3669 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3670 if the index field is valid
3674 the values returned by the cpuid instruction for
3675 this function/index combination
3677 4.89 KVM_S390_MEM_OP
3678 --------------------
3680 :Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION
3681 :Architectures: s390
3682 :Type: vm ioctl, vcpu ioctl
3683 :Parameters: struct kvm_s390_mem_op (in)
3684 :Returns: = 0 on success,
3685 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3686 > 0 if an exception occurred while walking the page tables
3688 Read or write data from/to the VM's memory.
3689 The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is
3692 Parameters are specified via the following structure::
3694 struct kvm_s390_mem_op {
3695 __u64 gaddr; /* the guest address */
3696 __u64 flags; /* flags */
3697 __u32 size; /* amount of bytes */
3698 __u32 op; /* type of operation */
3699 __u64 buf; /* buffer in userspace */
3702 __u8 ar; /* the access register number */
3703 __u8 key; /* access key, ignored if flag unset */
3705 __u32 sida_offset; /* offset into the sida */
3706 __u8 reserved[32]; /* ignored */
3710 The start address of the memory region has to be specified in the "gaddr"
3711 field, and the length of the region in the "size" field (which must not
3712 be 0). The maximum value for "size" can be obtained by checking the
3713 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3714 userspace application where the read data should be written to for
3715 a read access, or where the data that should be written is stored for
3716 a write access. The "reserved" field is meant for future extensions.
3717 Reserved and unused values are ignored. Future extension that add members must
3718 introduce new flags.
3720 The type of operation is specified in the "op" field. Flags modifying
3721 their behavior can be set in the "flags" field. Undefined flag bits must
3724 Possible operations are:
3725 * ``KVM_S390_MEMOP_LOGICAL_READ``
3726 * ``KVM_S390_MEMOP_LOGICAL_WRITE``
3727 * ``KVM_S390_MEMOP_ABSOLUTE_READ``
3728 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE``
3729 * ``KVM_S390_MEMOP_SIDA_READ``
3730 * ``KVM_S390_MEMOP_SIDA_WRITE``
3735 Access logical memory, i.e. translate the given guest address to an absolute
3736 address given the state of the VCPU and use the absolute address as target of
3737 the access. "ar" designates the access register number to be used; the valid
3739 Logical accesses are permitted for the VCPU ioctl only.
3740 Logical accesses are permitted for non-protected guests only.
3743 * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3744 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION``
3745 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3747 The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the
3748 corresponding memory access would cause an access exception; however,
3749 no actual access to the data in memory at the destination is performed.
3750 In this case, "buf" is unused and can be NULL.
3752 In case an access exception occurred during the access (or would occur
3753 in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive
3754 error number indicating the type of exception. This exception is also
3755 raised directly at the corresponding VCPU if the flag
3756 KVM_S390_MEMOP_F_INJECT_EXCEPTION is set.
3758 If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key
3759 protection is also in effect and may cause exceptions if accesses are
3760 prohibited given the access key designated by "key"; the valid range is 0..15.
3761 KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION
3764 Absolute read/write:
3765 ^^^^^^^^^^^^^^^^^^^^
3767 Access absolute memory. This operation is intended to be used with the
3768 KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing
3769 the checks required for storage key protection as one operation (as opposed to
3770 user space getting the storage keys, performing the checks, and accessing
3771 memory thereafter, which could lead to a delay between check and access).
3772 Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION
3774 Currently absolute accesses are not permitted for VCPU ioctls.
3775 Absolute accesses are permitted for non-protected guests only.
3778 * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3779 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3781 The semantics of the flags are as for logical accesses.
3786 Access the secure instruction data area which contains memory operands necessary
3787 for instruction emulation for protected guests.
3788 SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available.
3789 SIDA accesses are permitted for the VCPU ioctl only.
3790 SIDA accesses are permitted for protected guests only.
3792 No flags are supported.
3794 4.90 KVM_S390_GET_SKEYS
3795 -----------------------
3797 :Capability: KVM_CAP_S390_SKEYS
3798 :Architectures: s390
3800 :Parameters: struct kvm_s390_skeys
3801 :Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage
3802 keys, negative value on error
3804 This ioctl is used to get guest storage key values on the s390
3805 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3807 struct kvm_s390_skeys {
3810 __u64 skeydata_addr;
3815 The start_gfn field is the number of the first guest frame whose storage keys
3818 The count field is the number of consecutive frames (starting from start_gfn)
3819 whose storage keys to get. The count field must be at least 1 and the maximum
3820 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3821 will cause the ioctl to return -EINVAL.
3823 The skeydata_addr field is the address to a buffer large enough to hold count
3824 bytes. This buffer will be filled with storage key data by the ioctl.
3826 4.91 KVM_S390_SET_SKEYS
3827 -----------------------
3829 :Capability: KVM_CAP_S390_SKEYS
3830 :Architectures: s390
3832 :Parameters: struct kvm_s390_skeys
3833 :Returns: 0 on success, negative value on error
3835 This ioctl is used to set guest storage key values on the s390
3836 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3837 See section on KVM_S390_GET_SKEYS for struct definition.
3839 The start_gfn field is the number of the first guest frame whose storage keys
3842 The count field is the number of consecutive frames (starting from start_gfn)
3843 whose storage keys to get. The count field must be at least 1 and the maximum
3844 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3845 will cause the ioctl to return -EINVAL.
3847 The skeydata_addr field is the address to a buffer containing count bytes of
3848 storage keys. Each byte in the buffer will be set as the storage key for a
3849 single frame starting at start_gfn for count frames.
3851 Note: If any architecturally invalid key value is found in the given data then
3852 the ioctl will return -EINVAL.
3857 :Capability: KVM_CAP_S390_INJECT_IRQ
3858 :Architectures: s390
3860 :Parameters: struct kvm_s390_irq (in)
3861 :Returns: 0 on success, -1 on error
3866 ====== =================================================================
3867 EINVAL interrupt type is invalid
3868 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3869 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3870 than the maximum of VCPUs
3871 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3872 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3873 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3875 ====== =================================================================
3877 Allows to inject an interrupt to the guest.
3879 Using struct kvm_s390_irq as a parameter allows
3880 to inject additional payload which is not
3881 possible via KVM_S390_INTERRUPT.
3883 Interrupt parameters are passed via kvm_s390_irq::
3885 struct kvm_s390_irq {
3888 struct kvm_s390_io_info io;
3889 struct kvm_s390_ext_info ext;
3890 struct kvm_s390_pgm_info pgm;
3891 struct kvm_s390_emerg_info emerg;
3892 struct kvm_s390_extcall_info extcall;
3893 struct kvm_s390_prefix_info prefix;
3894 struct kvm_s390_stop_info stop;
3895 struct kvm_s390_mchk_info mchk;
3900 type can be one of the following:
3902 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3903 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3904 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3905 - KVM_S390_RESTART - restart; no parameters
3906 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3907 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3908 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3909 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3910 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3912 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3914 4.94 KVM_S390_GET_IRQ_STATE
3915 ---------------------------
3917 :Capability: KVM_CAP_S390_IRQ_STATE
3918 :Architectures: s390
3920 :Parameters: struct kvm_s390_irq_state (out)
3921 :Returns: >= number of bytes copied into buffer,
3922 -EINVAL if buffer size is 0,
3923 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3924 -EFAULT if the buffer address was invalid
3926 This ioctl allows userspace to retrieve the complete state of all currently
3927 pending interrupts in a single buffer. Use cases include migration
3928 and introspection. The parameter structure contains the address of a
3929 userspace buffer and its length::
3931 struct kvm_s390_irq_state {
3933 __u32 flags; /* will stay unused for compatibility reasons */
3935 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3938 Userspace passes in the above struct and for each pending interrupt a
3939 struct kvm_s390_irq is copied to the provided buffer.
3941 The structure contains a flags and a reserved field for future extensions. As
3942 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3943 reserved, these fields can not be used in the future without breaking
3946 If -ENOBUFS is returned the buffer provided was too small and userspace
3947 may retry with a bigger buffer.
3949 4.95 KVM_S390_SET_IRQ_STATE
3950 ---------------------------
3952 :Capability: KVM_CAP_S390_IRQ_STATE
3953 :Architectures: s390
3955 :Parameters: struct kvm_s390_irq_state (in)
3956 :Returns: 0 on success,
3957 -EFAULT if the buffer address was invalid,
3958 -EINVAL for an invalid buffer length (see below),
3959 -EBUSY if there were already interrupts pending,
3960 errors occurring when actually injecting the
3961 interrupt. See KVM_S390_IRQ.
3963 This ioctl allows userspace to set the complete state of all cpu-local
3964 interrupts currently pending for the vcpu. It is intended for restoring
3965 interrupt state after a migration. The input parameter is a userspace buffer
3966 containing a struct kvm_s390_irq_state::
3968 struct kvm_s390_irq_state {
3970 __u32 flags; /* will stay unused for compatibility reasons */
3972 __u32 reserved[4]; /* will stay unused for compatibility reasons */
3975 The restrictions for flags and reserved apply as well.
3976 (see KVM_S390_GET_IRQ_STATE)
3978 The userspace memory referenced by buf contains a struct kvm_s390_irq
3979 for each interrupt to be injected into the guest.
3980 If one of the interrupts could not be injected for some reason the
3983 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3984 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3985 which is the maximum number of possibly pending cpu-local interrupts.
3990 :Capability: KVM_CAP_X86_SMM
3994 :Returns: 0 on success, -1 on error
3996 Queues an SMI on the thread's vcpu.
3998 4.97 KVM_X86_SET_MSR_FILTER
3999 ----------------------------
4001 :Capability: KVM_X86_SET_MSR_FILTER
4004 :Parameters: struct kvm_msr_filter
4005 :Returns: 0 on success, < 0 on error
4009 struct kvm_msr_filter_range {
4010 #define KVM_MSR_FILTER_READ (1 << 0)
4011 #define KVM_MSR_FILTER_WRITE (1 << 1)
4013 __u32 nmsrs; /* number of msrs in bitmap */
4014 __u32 base; /* MSR index the bitmap starts at */
4015 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4018 #define KVM_MSR_FILTER_MAX_RANGES 16
4019 struct kvm_msr_filter {
4020 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4021 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4023 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4026 flags values for ``struct kvm_msr_filter_range``:
4028 ``KVM_MSR_FILTER_READ``
4030 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4031 indicates that a read should immediately fail, while a 1 indicates that
4032 a read for a particular MSR should be handled regardless of the default
4035 ``KVM_MSR_FILTER_WRITE``
4037 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4038 indicates that a write should immediately fail, while a 1 indicates that
4039 a write for a particular MSR should be handled regardless of the default
4042 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4044 Filter both read and write accesses to MSRs using the given bitmap. A 0
4045 in the bitmap indicates that both reads and writes should immediately fail,
4046 while a 1 indicates that reads and writes for a particular MSR are not
4047 filtered by this range.
4049 flags values for ``struct kvm_msr_filter``:
4051 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4053 If no filter range matches an MSR index that is getting accessed, KVM will
4054 fall back to allowing access to the MSR.
4056 ``KVM_MSR_FILTER_DEFAULT_DENY``
4058 If no filter range matches an MSR index that is getting accessed, KVM will
4059 fall back to rejecting access to the MSR. In this mode, all MSRs that should
4060 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4062 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4063 specify whether a certain MSR access should be explicitly filtered for or not.
4065 If this ioctl has never been invoked, MSR accesses are not guarded and the
4066 default KVM in-kernel emulation behavior is fully preserved.
4068 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4069 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4072 As soon as the filtering is in place, every MSR access is processed through
4073 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4074 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4075 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4079 MSR accesses coming from nested vmentry/vmexit are not filtered.
4080 This includes both writes to individual VMCS fields and reads/writes
4081 through the MSR lists pointed to by the VMCS.
4083 If a bit is within one of the defined ranges, read and write accesses are
4084 guarded by the bitmap's value for the MSR index if the kind of access
4085 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4086 cover this particular access, the behavior is determined by the flags
4087 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4088 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4090 Each bitmap range specifies a range of MSRs to potentially allow access on.
4091 The range goes from MSR index [base .. base+nmsrs]. The flags field
4092 indicates whether reads, writes or both reads and writes are filtered
4093 by setting a 1 bit in the bitmap for the corresponding MSR index.
4095 If an MSR access is not permitted through the filtering, it generates a
4096 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4097 allows user space to deflect and potentially handle various MSR accesses
4100 If a vCPU is in running state while this ioctl is invoked, the vCPU may
4101 experience inconsistent filtering behavior on MSR accesses.
4103 4.98 KVM_CREATE_SPAPR_TCE_64
4104 ----------------------------
4106 :Capability: KVM_CAP_SPAPR_TCE_64
4107 :Architectures: powerpc
4109 :Parameters: struct kvm_create_spapr_tce_64 (in)
4110 :Returns: file descriptor for manipulating the created TCE table
4112 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4113 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4115 This capability uses extended struct in ioctl interface::
4117 /* for KVM_CAP_SPAPR_TCE_64 */
4118 struct kvm_create_spapr_tce_64 {
4122 __u64 offset; /* in pages */
4123 __u64 size; /* in pages */
4126 The aim of extension is to support an additional bigger DMA window with
4127 a variable page size.
4128 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4129 a bus offset of the corresponding DMA window, @size and @offset are numbers
4132 @flags are not used at the moment.
4134 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4136 4.99 KVM_REINJECT_CONTROL
4137 -------------------------
4139 :Capability: KVM_CAP_REINJECT_CONTROL
4142 :Parameters: struct kvm_reinject_control (in)
4143 :Returns: 0 on success,
4144 -EFAULT if struct kvm_reinject_control cannot be read,
4145 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4147 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
4148 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4149 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
4150 interrupt whenever there isn't a pending interrupt from i8254.
4151 !reinject mode injects an interrupt as soon as a tick arrives.
4155 struct kvm_reinject_control {
4160 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4161 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4163 4.100 KVM_PPC_CONFIGURE_V3_MMU
4164 ------------------------------
4166 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
4169 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
4170 :Returns: 0 on success,
4171 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4172 -EINVAL if the configuration is invalid
4174 This ioctl controls whether the guest will use radix or HPT (hashed
4175 page table) translation, and sets the pointer to the process table for
4180 struct kvm_ppc_mmuv3_cfg {
4182 __u64 process_table;
4185 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4186 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
4187 to use radix tree translation, and if clear, to use HPT translation.
4188 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4189 to be able to use the global TLB and SLB invalidation instructions;
4190 if clear, the guest may not use these instructions.
4192 The process_table field specifies the address and size of the guest
4193 process table, which is in the guest's space. This field is formatted
4194 as the second doubleword of the partition table entry, as defined in
4195 the Power ISA V3.00, Book III section 5.7.6.1.
4197 4.101 KVM_PPC_GET_RMMU_INFO
4198 ---------------------------
4200 :Capability: KVM_CAP_PPC_RADIX_MMU
4203 :Parameters: struct kvm_ppc_rmmu_info (out)
4204 :Returns: 0 on success,
4205 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4206 -EINVAL if no useful information can be returned
4208 This ioctl returns a structure containing two things: (a) a list
4209 containing supported radix tree geometries, and (b) a list that maps
4210 page sizes to put in the "AP" (actual page size) field for the tlbie
4211 (TLB invalidate entry) instruction.
4215 struct kvm_ppc_rmmu_info {
4216 struct kvm_ppc_radix_geom {
4221 __u32 ap_encodings[8];
4224 The geometries[] field gives up to 8 supported geometries for the
4225 radix page table, in terms of the log base 2 of the smallest page
4226 size, and the number of bits indexed at each level of the tree, from
4227 the PTE level up to the PGD level in that order. Any unused entries
4228 will have 0 in the page_shift field.
4230 The ap_encodings gives the supported page sizes and their AP field
4231 encodings, encoded with the AP value in the top 3 bits and the log
4232 base 2 of the page size in the bottom 6 bits.
4234 4.102 KVM_PPC_RESIZE_HPT_PREPARE
4235 --------------------------------
4237 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4238 :Architectures: powerpc
4240 :Parameters: struct kvm_ppc_resize_hpt (in)
4241 :Returns: 0 on successful completion,
4242 >0 if a new HPT is being prepared, the value is an estimated
4243 number of milliseconds until preparation is complete,
4244 -EFAULT if struct kvm_reinject_control cannot be read,
4245 -EINVAL if the supplied shift or flags are invalid,
4246 -ENOMEM if unable to allocate the new HPT,
4248 Used to implement the PAPR extension for runtime resizing of a guest's
4249 Hashed Page Table (HPT). Specifically this starts, stops or monitors
4250 the preparation of a new potential HPT for the guest, essentially
4251 implementing the H_RESIZE_HPT_PREPARE hypercall.
4255 struct kvm_ppc_resize_hpt {
4261 If called with shift > 0 when there is no pending HPT for the guest,
4262 this begins preparation of a new pending HPT of size 2^(shift) bytes.
4263 It then returns a positive integer with the estimated number of
4264 milliseconds until preparation is complete.
4266 If called when there is a pending HPT whose size does not match that
4267 requested in the parameters, discards the existing pending HPT and
4268 creates a new one as above.
4270 If called when there is a pending HPT of the size requested, will:
4272 * If preparation of the pending HPT is already complete, return 0
4273 * If preparation of the pending HPT has failed, return an error
4274 code, then discard the pending HPT.
4275 * If preparation of the pending HPT is still in progress, return an
4276 estimated number of milliseconds until preparation is complete.
4278 If called with shift == 0, discards any currently pending HPT and
4279 returns 0 (i.e. cancels any in-progress preparation).
4281 flags is reserved for future expansion, currently setting any bits in
4282 flags will result in an -EINVAL.
4284 Normally this will be called repeatedly with the same parameters until
4285 it returns <= 0. The first call will initiate preparation, subsequent
4286 ones will monitor preparation until it completes or fails.
4288 4.103 KVM_PPC_RESIZE_HPT_COMMIT
4289 -------------------------------
4291 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4292 :Architectures: powerpc
4294 :Parameters: struct kvm_ppc_resize_hpt (in)
4295 :Returns: 0 on successful completion,
4296 -EFAULT if struct kvm_reinject_control cannot be read,
4297 -EINVAL if the supplied shift or flags are invalid,
4298 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4299 have the requested size,
4300 -EBUSY if the pending HPT is not fully prepared,
4301 -ENOSPC if there was a hash collision when moving existing
4302 HPT entries to the new HPT,
4303 -EIO on other error conditions
4305 Used to implement the PAPR extension for runtime resizing of a guest's
4306 Hashed Page Table (HPT). Specifically this requests that the guest be
4307 transferred to working with the new HPT, essentially implementing the
4308 H_RESIZE_HPT_COMMIT hypercall.
4312 struct kvm_ppc_resize_hpt {
4318 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4319 returned 0 with the same parameters. In other cases
4320 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4321 -EBUSY, though others may be possible if the preparation was started,
4324 This will have undefined effects on the guest if it has not already
4325 placed itself in a quiescent state where no vcpu will make MMU enabled
4328 On succsful completion, the pending HPT will become the guest's active
4329 HPT and the previous HPT will be discarded.
4331 On failure, the guest will still be operating on its previous HPT.
4333 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4334 -----------------------------------
4336 :Capability: KVM_CAP_MCE
4339 :Parameters: u64 mce_cap (out)
4340 :Returns: 0 on success, -1 on error
4342 Returns supported MCE capabilities. The u64 mce_cap parameter
4343 has the same format as the MSR_IA32_MCG_CAP register. Supported
4344 capabilities will have the corresponding bits set.
4346 4.105 KVM_X86_SETUP_MCE
4347 -----------------------
4349 :Capability: KVM_CAP_MCE
4352 :Parameters: u64 mcg_cap (in)
4353 :Returns: 0 on success,
4354 -EFAULT if u64 mcg_cap cannot be read,
4355 -EINVAL if the requested number of banks is invalid,
4356 -EINVAL if requested MCE capability is not supported.
4358 Initializes MCE support for use. The u64 mcg_cap parameter
4359 has the same format as the MSR_IA32_MCG_CAP register and
4360 specifies which capabilities should be enabled. The maximum
4361 supported number of error-reporting banks can be retrieved when
4362 checking for KVM_CAP_MCE. The supported capabilities can be
4363 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4365 4.106 KVM_X86_SET_MCE
4366 ---------------------
4368 :Capability: KVM_CAP_MCE
4371 :Parameters: struct kvm_x86_mce (in)
4372 :Returns: 0 on success,
4373 -EFAULT if struct kvm_x86_mce cannot be read,
4374 -EINVAL if the bank number is invalid,
4375 -EINVAL if VAL bit is not set in status field.
4377 Inject a machine check error (MCE) into the guest. The input
4380 struct kvm_x86_mce {
4390 If the MCE being reported is an uncorrected error, KVM will
4391 inject it as an MCE exception into the guest. If the guest
4392 MCG_STATUS register reports that an MCE is in progress, KVM
4393 causes an KVM_EXIT_SHUTDOWN vmexit.
4395 Otherwise, if the MCE is a corrected error, KVM will just
4396 store it in the corresponding bank (provided this bank is
4397 not holding a previously reported uncorrected error).
4399 4.107 KVM_S390_GET_CMMA_BITS
4400 ----------------------------
4402 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4403 :Architectures: s390
4405 :Parameters: struct kvm_s390_cmma_log (in, out)
4406 :Returns: 0 on success, a negative value on error
4408 This ioctl is used to get the values of the CMMA bits on the s390
4409 architecture. It is meant to be used in two scenarios:
4411 - During live migration to save the CMMA values. Live migration needs
4412 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4413 - To non-destructively peek at the CMMA values, with the flag
4414 KVM_S390_CMMA_PEEK set.
4416 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4417 values are written to a buffer whose location is indicated via the "values"
4418 member in the kvm_s390_cmma_log struct. The values in the input struct are
4419 also updated as needed.
4421 Each CMMA value takes up one byte.
4425 struct kvm_s390_cmma_log {
4436 start_gfn is the number of the first guest frame whose CMMA values are
4439 count is the length of the buffer in bytes,
4441 values points to the buffer where the result will be written to.
4443 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4444 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4447 The result is written in the buffer pointed to by the field values, and
4448 the values of the input parameter are updated as follows.
4450 Depending on the flags, different actions are performed. The only
4451 supported flag so far is KVM_S390_CMMA_PEEK.
4453 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4454 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4455 It is not necessarily the same as the one passed as input, as clean pages
4458 count will indicate the number of bytes actually written in the buffer.
4459 It can (and very often will) be smaller than the input value, since the
4460 buffer is only filled until 16 bytes of clean values are found (which
4461 are then not copied in the buffer). Since a CMMA migration block needs
4462 the base address and the length, for a total of 16 bytes, we will send
4463 back some clean data if there is some dirty data afterwards, as long as
4464 the size of the clean data does not exceed the size of the header. This
4465 allows to minimize the amount of data to be saved or transferred over
4466 the network at the expense of more roundtrips to userspace. The next
4467 invocation of the ioctl will skip over all the clean values, saving
4468 potentially more than just the 16 bytes we found.
4470 If KVM_S390_CMMA_PEEK is set:
4471 the existing storage attributes are read even when not in migration
4472 mode, and no other action is performed;
4474 the output start_gfn will be equal to the input start_gfn,
4476 the output count will be equal to the input count, except if the end of
4477 memory has been reached.
4480 the field "remaining" will indicate the total number of dirty CMMA values
4481 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4486 values points to the userspace buffer where the result will be stored.
4488 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4489 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4490 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4491 -EFAULT if the userspace address is invalid or if no page table is
4492 present for the addresses (e.g. when using hugepages).
4494 4.108 KVM_S390_SET_CMMA_BITS
4495 ----------------------------
4497 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4498 :Architectures: s390
4500 :Parameters: struct kvm_s390_cmma_log (in)
4501 :Returns: 0 on success, a negative value on error
4503 This ioctl is used to set the values of the CMMA bits on the s390
4504 architecture. It is meant to be used during live migration to restore
4505 the CMMA values, but there are no restrictions on its use.
4506 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4507 Each CMMA value takes up one byte.
4511 struct kvm_s390_cmma_log {
4522 start_gfn indicates the starting guest frame number,
4524 count indicates how many values are to be considered in the buffer,
4526 flags is not used and must be 0.
4528 mask indicates which PGSTE bits are to be considered.
4530 remaining is not used.
4532 values points to the buffer in userspace where to store the values.
4534 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4535 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4536 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4537 if the flags field was not 0, with -EFAULT if the userspace address is
4538 invalid, if invalid pages are written to (e.g. after the end of memory)
4539 or if no page table is present for the addresses (e.g. when using
4542 4.109 KVM_PPC_GET_CPU_CHAR
4543 --------------------------
4545 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4546 :Architectures: powerpc
4548 :Parameters: struct kvm_ppc_cpu_char (out)
4549 :Returns: 0 on successful completion,
4550 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4552 This ioctl gives userspace information about certain characteristics
4553 of the CPU relating to speculative execution of instructions and
4554 possible information leakage resulting from speculative execution (see
4555 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4556 returned in struct kvm_ppc_cpu_char, which looks like this::
4558 struct kvm_ppc_cpu_char {
4559 __u64 character; /* characteristics of the CPU */
4560 __u64 behaviour; /* recommended software behaviour */
4561 __u64 character_mask; /* valid bits in character */
4562 __u64 behaviour_mask; /* valid bits in behaviour */
4565 For extensibility, the character_mask and behaviour_mask fields
4566 indicate which bits of character and behaviour have been filled in by
4567 the kernel. If the set of defined bits is extended in future then
4568 userspace will be able to tell whether it is running on a kernel that
4569 knows about the new bits.
4571 The character field describes attributes of the CPU which can help
4572 with preventing inadvertent information disclosure - specifically,
4573 whether there is an instruction to flash-invalidate the L1 data cache
4574 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4575 to a mode where entries can only be used by the thread that created
4576 them, whether the bcctr[l] instruction prevents speculation, and
4577 whether a speculation barrier instruction (ori 31,31,0) is provided.
4579 The behaviour field describes actions that software should take to
4580 prevent inadvertent information disclosure, and thus describes which
4581 vulnerabilities the hardware is subject to; specifically whether the
4582 L1 data cache should be flushed when returning to user mode from the
4583 kernel, and whether a speculation barrier should be placed between an
4584 array bounds check and the array access.
4586 These fields use the same bit definitions as the new
4587 H_GET_CPU_CHARACTERISTICS hypercall.
4589 4.110 KVM_MEMORY_ENCRYPT_OP
4590 ---------------------------
4595 :Parameters: an opaque platform specific structure (in/out)
4596 :Returns: 0 on success; -1 on error
4598 If the platform supports creating encrypted VMs then this ioctl can be used
4599 for issuing platform-specific memory encryption commands to manage those
4602 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4603 (SEV) commands on AMD Processors. The SEV commands are defined in
4604 Documentation/virt/kvm/amd-memory-encryption.rst.
4606 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4607 -----------------------------------
4612 :Parameters: struct kvm_enc_region (in)
4613 :Returns: 0 on success; -1 on error
4615 This ioctl can be used to register a guest memory region which may
4616 contain encrypted data (e.g. guest RAM, SMRAM etc).
4618 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4619 memory region may contain encrypted data. The SEV memory encryption
4620 engine uses a tweak such that two identical plaintext pages, each at
4621 different locations will have differing ciphertexts. So swapping or
4622 moving ciphertext of those pages will not result in plaintext being
4623 swapped. So relocating (or migrating) physical backing pages for the SEV
4624 guest will require some additional steps.
4626 Note: The current SEV key management spec does not provide commands to
4627 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4628 memory region registered with the ioctl.
4630 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4631 -------------------------------------
4636 :Parameters: struct kvm_enc_region (in)
4637 :Returns: 0 on success; -1 on error
4639 This ioctl can be used to unregister the guest memory region registered
4640 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4642 4.113 KVM_HYPERV_EVENTFD
4643 ------------------------
4645 :Capability: KVM_CAP_HYPERV_EVENTFD
4648 :Parameters: struct kvm_hyperv_eventfd (in)
4650 This ioctl (un)registers an eventfd to receive notifications from the guest on
4651 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4652 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4653 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4657 struct kvm_hyperv_eventfd {
4664 The conn_id field should fit within 24 bits::
4666 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4668 The acceptable values for the flags field are::
4670 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4672 :Returns: 0 on success,
4673 -EINVAL if conn_id or flags is outside the allowed range,
4674 -ENOENT on deassign if the conn_id isn't registered,
4675 -EEXIST on assign if the conn_id is already registered
4677 4.114 KVM_GET_NESTED_STATE
4678 --------------------------
4680 :Capability: KVM_CAP_NESTED_STATE
4683 :Parameters: struct kvm_nested_state (in/out)
4684 :Returns: 0 on success, -1 on error
4688 ===== =============================================================
4689 E2BIG the total state size exceeds the value of 'size' specified by
4690 the user; the size required will be written into size.
4691 ===== =============================================================
4695 struct kvm_nested_state {
4701 struct kvm_vmx_nested_state_hdr vmx;
4702 struct kvm_svm_nested_state_hdr svm;
4704 /* Pad the header to 128 bytes. */
4709 struct kvm_vmx_nested_state_data vmx[0];
4710 struct kvm_svm_nested_state_data svm[0];
4714 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4715 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4716 #define KVM_STATE_NESTED_EVMCS 0x00000004
4718 #define KVM_STATE_NESTED_FORMAT_VMX 0
4719 #define KVM_STATE_NESTED_FORMAT_SVM 1
4721 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4723 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4724 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4726 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4728 struct kvm_vmx_nested_state_hdr {
4737 __u64 preemption_timer_deadline;
4740 struct kvm_vmx_nested_state_data {
4741 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4742 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4745 This ioctl copies the vcpu's nested virtualization state from the kernel to
4748 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4749 to the KVM_CHECK_EXTENSION ioctl().
4751 4.115 KVM_SET_NESTED_STATE
4752 --------------------------
4754 :Capability: KVM_CAP_NESTED_STATE
4757 :Parameters: struct kvm_nested_state (in)
4758 :Returns: 0 on success, -1 on error
4760 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4761 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4763 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4764 -------------------------------------
4766 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4767 KVM_CAP_COALESCED_PIO (for coalesced pio)
4770 :Parameters: struct kvm_coalesced_mmio_zone
4771 :Returns: 0 on success, < 0 on error
4773 Coalesced I/O is a performance optimization that defers hardware
4774 register write emulation so that userspace exits are avoided. It is
4775 typically used to reduce the overhead of emulating frequently accessed
4778 When a hardware register is configured for coalesced I/O, write accesses
4779 do not exit to userspace and their value is recorded in a ring buffer
4780 that is shared between kernel and userspace.
4782 Coalesced I/O is used if one or more write accesses to a hardware
4783 register can be deferred until a read or a write to another hardware
4784 register on the same device. This last access will cause a vmexit and
4785 userspace will process accesses from the ring buffer before emulating
4786 it. That will avoid exiting to userspace on repeated writes.
4788 Coalesced pio is based on coalesced mmio. There is little difference
4789 between coalesced mmio and pio except that coalesced pio records accesses
4792 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4793 ------------------------------------
4795 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4796 :Architectures: x86, arm64, mips
4798 :Parameters: struct kvm_clear_dirty_log (in)
4799 :Returns: 0 on success, -1 on error
4803 /* for KVM_CLEAR_DIRTY_LOG */
4804 struct kvm_clear_dirty_log {
4809 void __user *dirty_bitmap; /* one bit per page */
4814 The ioctl clears the dirty status of pages in a memory slot, according to
4815 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4816 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4817 memory slot, and num_pages is the size in bits of the input bitmap.
4818 first_page must be a multiple of 64; num_pages must also be a multiple of
4819 64 unless first_page + num_pages is the size of the memory slot. For each
4820 bit that is set in the input bitmap, the corresponding page is marked "clean"
4821 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4822 (for example via write-protection, or by clearing the dirty bit in
4823 a page table entry).
4825 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4826 the address space for which you want to clear the dirty status. See
4827 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4829 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4830 is enabled; for more information, see the description of the capability.
4831 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4832 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4834 4.118 KVM_GET_SUPPORTED_HV_CPUID
4835 --------------------------------
4837 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4839 :Type: system ioctl, vcpu ioctl
4840 :Parameters: struct kvm_cpuid2 (in/out)
4841 :Returns: 0 on success, -1 on error
4848 struct kvm_cpuid_entry2 entries[0];
4851 struct kvm_cpuid_entry2 {
4862 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4863 KVM. Userspace can use the information returned by this ioctl to construct
4864 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4865 Windows or Hyper-V guests).
4867 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4868 Functional Specification (TLFS). These leaves can't be obtained with
4869 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4870 leaves (0x40000000, 0x40000001).
4872 Currently, the following list of CPUID leaves are returned:
4874 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4875 - HYPERV_CPUID_INTERFACE
4876 - HYPERV_CPUID_VERSION
4877 - HYPERV_CPUID_FEATURES
4878 - HYPERV_CPUID_ENLIGHTMENT_INFO
4879 - HYPERV_CPUID_IMPLEMENT_LIMITS
4880 - HYPERV_CPUID_NESTED_FEATURES
4881 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4882 - HYPERV_CPUID_SYNDBG_INTERFACE
4883 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4885 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4886 with the 'nent' field indicating the number of entries in the variable-size
4887 array 'entries'. If the number of entries is too low to describe all Hyper-V
4888 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4889 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4890 number of valid entries in the 'entries' array, which is then filled.
4892 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4893 userspace should not expect to get any particular value there.
4895 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4896 system ioctl which exposes all supported feature bits unconditionally, vcpu
4897 version has the following quirks:
4899 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4900 feature bit are only exposed when Enlightened VMCS was previously enabled
4901 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4902 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4903 (presumes KVM_CREATE_IRQCHIP has already been called).
4905 4.119 KVM_ARM_VCPU_FINALIZE
4906 ---------------------------
4908 :Architectures: arm64
4910 :Parameters: int feature (in)
4911 :Returns: 0 on success, -1 on error
4915 ====== ==============================================================
4916 EPERM feature not enabled, needs configuration, or already finalized
4917 EINVAL feature unknown or not present
4918 ====== ==============================================================
4920 Recognised values for feature:
4922 ===== ===========================================
4923 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4924 ===== ===========================================
4926 Finalizes the configuration of the specified vcpu feature.
4928 The vcpu must already have been initialised, enabling the affected feature, by
4929 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4932 For affected vcpu features, this is a mandatory step that must be performed
4933 before the vcpu is fully usable.
4935 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4936 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
4937 that should be performaned and how to do it are feature-dependent.
4939 Other calls that depend on a particular feature being finalized, such as
4940 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4941 -EPERM unless the feature has already been finalized by means of a
4942 KVM_ARM_VCPU_FINALIZE call.
4944 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4947 4.120 KVM_SET_PMU_EVENT_FILTER
4948 ------------------------------
4950 :Capability: KVM_CAP_PMU_EVENT_FILTER
4953 :Parameters: struct kvm_pmu_event_filter (in)
4954 :Returns: 0 on success, -1 on error
4958 struct kvm_pmu_event_filter {
4961 __u32 fixed_counter_bitmap;
4967 This ioctl restricts the set of PMU events that the guest can program.
4968 The argument holds a list of events which will be allowed or denied.
4969 The eventsel+umask of each event the guest attempts to program is compared
4970 against the events field to determine whether the guest should have access.
4971 The events field only controls general purpose counters; fixed purpose
4972 counters are controlled by the fixed_counter_bitmap.
4974 No flags are defined yet, the field must be zero.
4976 Valid values for 'action'::
4978 #define KVM_PMU_EVENT_ALLOW 0
4979 #define KVM_PMU_EVENT_DENY 1
4981 4.121 KVM_PPC_SVM_OFF
4982 ---------------------
4985 :Architectures: powerpc
4988 :Returns: 0 on successful completion,
4992 ====== ================================================================
4993 EINVAL if ultravisor failed to terminate the secure guest
4994 ENOMEM if hypervisor failed to allocate new radix page tables for guest
4995 ====== ================================================================
4997 This ioctl is used to turn off the secure mode of the guest or transition
4998 the guest from secure mode to normal mode. This is invoked when the guest
4999 is reset. This has no effect if called for a normal guest.
5001 This ioctl issues an ultravisor call to terminate the secure guest,
5002 unpins the VPA pages and releases all the device pages that are used to
5003 track the secure pages by hypervisor.
5005 4.122 KVM_S390_NORMAL_RESET
5006 ---------------------------
5008 :Capability: KVM_CAP_S390_VCPU_RESETS
5009 :Architectures: s390
5014 This ioctl resets VCPU registers and control structures according to
5015 the cpu reset definition in the POP (Principles Of Operation).
5017 4.123 KVM_S390_INITIAL_RESET
5018 ----------------------------
5021 :Architectures: s390
5026 This ioctl resets VCPU registers and control structures according to
5027 the initial cpu reset definition in the POP. However, the cpu is not
5028 put into ESA mode. This reset is a superset of the normal reset.
5030 4.124 KVM_S390_CLEAR_RESET
5031 --------------------------
5033 :Capability: KVM_CAP_S390_VCPU_RESETS
5034 :Architectures: s390
5039 This ioctl resets VCPU registers and control structures according to
5040 the clear cpu reset definition in the POP. However, the cpu is not put
5041 into ESA mode. This reset is a superset of the initial reset.
5044 4.125 KVM_S390_PV_COMMAND
5045 -------------------------
5047 :Capability: KVM_CAP_S390_PROTECTED
5048 :Architectures: s390
5050 :Parameters: struct kvm_pv_cmd
5051 :Returns: 0 on success, < 0 on error
5056 __u32 cmd; /* Command to be executed */
5057 __u16 rc; /* Ultravisor return code */
5058 __u16 rrc; /* Ultravisor return reason code */
5059 __u64 data; /* Data or address */
5060 __u32 flags; /* flags for future extensions. Must be 0 for now */
5067 Allocate memory and register the VM with the Ultravisor, thereby
5068 donating memory to the Ultravisor that will become inaccessible to
5069 KVM. All existing CPUs are converted to protected ones. After this
5070 command has succeeded, any CPU added via hotplug will become
5071 protected during its creation as well.
5075 ===== =============================
5076 EINTR an unmasked signal is pending
5077 ===== =============================
5081 Deregister the VM from the Ultravisor and reclaim the memory that
5082 had been donated to the Ultravisor, making it usable by the kernel
5083 again. All registered VCPUs are converted back to non-protected
5086 KVM_PV_VM_SET_SEC_PARMS
5087 Pass the image header from VM memory to the Ultravisor in
5088 preparation of image unpacking and verification.
5091 Unpack (protect and decrypt) a page of the encrypted boot image.
5094 Verify the integrity of the unpacked image. Only if this succeeds,
5095 KVM is allowed to start protected VCPUs.
5097 4.126 KVM_X86_SET_MSR_FILTER
5098 ----------------------------
5100 :Capability: KVM_CAP_X86_MSR_FILTER
5103 :Parameters: struct kvm_msr_filter
5104 :Returns: 0 on success, < 0 on error
5108 struct kvm_msr_filter_range {
5109 #define KVM_MSR_FILTER_READ (1 << 0)
5110 #define KVM_MSR_FILTER_WRITE (1 << 1)
5112 __u32 nmsrs; /* number of msrs in bitmap */
5113 __u32 base; /* MSR index the bitmap starts at */
5114 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
5117 #define KVM_MSR_FILTER_MAX_RANGES 16
5118 struct kvm_msr_filter {
5119 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
5120 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
5122 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
5125 flags values for ``struct kvm_msr_filter_range``:
5127 ``KVM_MSR_FILTER_READ``
5129 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
5130 indicates that a read should immediately fail, while a 1 indicates that
5131 a read for a particular MSR should be handled regardless of the default
5134 ``KVM_MSR_FILTER_WRITE``
5136 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
5137 indicates that a write should immediately fail, while a 1 indicates that
5138 a write for a particular MSR should be handled regardless of the default
5141 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
5143 Filter both read and write accesses to MSRs using the given bitmap. A 0
5144 in the bitmap indicates that both reads and writes should immediately fail,
5145 while a 1 indicates that reads and writes for a particular MSR are not
5146 filtered by this range.
5148 flags values for ``struct kvm_msr_filter``:
5150 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
5152 If no filter range matches an MSR index that is getting accessed, KVM will
5153 fall back to allowing access to the MSR.
5155 ``KVM_MSR_FILTER_DEFAULT_DENY``
5157 If no filter range matches an MSR index that is getting accessed, KVM will
5158 fall back to rejecting access to the MSR. In this mode, all MSRs that should
5159 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
5161 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
5162 specify whether a certain MSR access should be explicitly filtered for or not.
5164 If this ioctl has never been invoked, MSR accesses are not guarded and the
5165 default KVM in-kernel emulation behavior is fully preserved.
5167 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
5168 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
5171 As soon as the filtering is in place, every MSR access is processed through
5172 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
5173 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
5174 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
5177 If a bit is within one of the defined ranges, read and write accesses are
5178 guarded by the bitmap's value for the MSR index if the kind of access
5179 is included in the ``struct kvm_msr_filter_range`` flags. If no range
5180 cover this particular access, the behavior is determined by the flags
5181 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
5182 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
5184 Each bitmap range specifies a range of MSRs to potentially allow access on.
5185 The range goes from MSR index [base .. base+nmsrs]. The flags field
5186 indicates whether reads, writes or both reads and writes are filtered
5187 by setting a 1 bit in the bitmap for the corresponding MSR index.
5189 If an MSR access is not permitted through the filtering, it generates a
5190 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
5191 allows user space to deflect and potentially handle various MSR accesses
5194 Note, invoking this ioctl with a vCPU is running is inherently racy. However,
5195 KVM does guarantee that vCPUs will see either the previous filter or the new
5196 filter, e.g. MSRs with identical settings in both the old and new filter will
5197 have deterministic behavior.
5199 4.127 KVM_XEN_HVM_SET_ATTR
5200 --------------------------
5202 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5205 :Parameters: struct kvm_xen_hvm_attr
5206 :Returns: 0 on success, < 0 on error
5210 struct kvm_xen_hvm_attr {
5225 KVM_XEN_ATTR_TYPE_LONG_MODE
5226 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
5227 determines the layout of the shared info pages exposed to the VM.
5229 KVM_XEN_ATTR_TYPE_SHARED_INFO
5230 Sets the guest physical frame number at which the Xen "shared info"
5231 page resides. Note that although Xen places vcpu_info for the first
5232 32 vCPUs in the shared_info page, KVM does not automatically do so
5233 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
5234 explicitly even when the vcpu_info for a given vCPU resides at the
5235 "default" location in the shared_info page. This is because KVM is
5236 not aware of the Xen CPU id which is used as the index into the
5237 vcpu_info[] array, so cannot know the correct default location.
5239 Note that the shared info page may be constantly written to by KVM;
5240 it contains the event channel bitmap used to deliver interrupts to
5241 a Xen guest, amongst other things. It is exempt from dirty tracking
5242 mechanisms — KVM will not explicitly mark the page as dirty each
5243 time an event channel interrupt is delivered to the guest! Thus,
5244 userspace should always assume that the designated GFN is dirty if
5245 any vCPU has been running or any event channel interrupts can be
5246 routed to the guest.
5248 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
5249 Sets the exception vector used to deliver Xen event channel upcalls.
5251 4.127 KVM_XEN_HVM_GET_ATTR
5252 --------------------------
5254 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5257 :Parameters: struct kvm_xen_hvm_attr
5258 :Returns: 0 on success, < 0 on error
5260 Allows Xen VM attributes to be read. For the structure and types,
5261 see KVM_XEN_HVM_SET_ATTR above.
5263 4.128 KVM_XEN_VCPU_SET_ATTR
5264 ---------------------------
5266 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5269 :Parameters: struct kvm_xen_vcpu_attr
5270 :Returns: 0 on success, < 0 on error
5274 struct kvm_xen_vcpu_attr {
5282 __u64 state_entry_time;
5284 __u64 time_runnable;
5293 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
5294 Sets the guest physical address of the vcpu_info for a given vCPU.
5295 As with the shared_info page for the VM, the corresponding page may be
5296 dirtied at any time if event channel interrupt delivery is enabled, so
5297 userspace should always assume that the page is dirty without relying
5300 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
5301 Sets the guest physical address of an additional pvclock structure
5302 for a given vCPU. This is typically used for guest vsyscall support.
5304 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5305 Sets the guest physical address of the vcpu_runstate_info for a given
5306 vCPU. This is how a Xen guest tracks CPU state such as steal time.
5308 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5309 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5310 the given vCPU from the .u.runstate.state member of the structure.
5311 KVM automatically accounts running and runnable time but blocked
5312 and offline states are only entered explicitly.
5314 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5315 Sets all fields of the vCPU runstate data from the .u.runstate member
5316 of the structure, including the current runstate. The state_entry_time
5317 must equal the sum of the other four times.
5319 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5320 This *adds* the contents of the .u.runstate members of the structure
5321 to the corresponding members of the given vCPU's runstate data, thus
5322 permitting atomic adjustments to the runstate times. The adjustment
5323 to the state_entry_time must equal the sum of the adjustments to the
5324 other four times. The state field must be set to -1, or to a valid
5325 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5326 or RUNSTATE_offline) to set the current accounted state as of the
5327 adjusted state_entry_time.
5329 4.129 KVM_XEN_VCPU_GET_ATTR
5330 ---------------------------
5332 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5335 :Parameters: struct kvm_xen_vcpu_attr
5336 :Returns: 0 on success, < 0 on error
5338 Allows Xen vCPU attributes to be read. For the structure and types,
5339 see KVM_XEN_VCPU_SET_ATTR above.
5341 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5342 with the KVM_XEN_VCPU_GET_ATTR ioctl.
5344 4.130 KVM_ARM_MTE_COPY_TAGS
5345 ---------------------------
5347 :Capability: KVM_CAP_ARM_MTE
5348 :Architectures: arm64
5350 :Parameters: struct kvm_arm_copy_mte_tags
5351 :Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5352 arguments, -EFAULT if memory cannot be accessed).
5356 struct kvm_arm_copy_mte_tags {
5364 Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5365 ``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr``
5366 field must point to a buffer which the tags will be copied to or from.
5368 ``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5369 ``KVM_ARM_TAGS_FROM_GUEST``.
5371 The size of the buffer to store the tags is ``(length / 16)`` bytes
5372 (granules in MTE are 16 bytes long). Each byte contains a single tag
5373 value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5374 ``PTRACE_POKEMTETAGS``.
5376 If an error occurs before any data is copied then a negative error code is
5377 returned. If some tags have been copied before an error occurs then the number
5378 of bytes successfully copied is returned. If the call completes successfully
5379 then ``length`` is returned.
5381 4.131 KVM_GET_SREGS2
5382 --------------------
5384 :Capability: KVM_CAP_SREGS2
5387 :Parameters: struct kvm_sregs2 (out)
5388 :Returns: 0 on success, -1 on error
5390 Reads special registers from the vcpu.
5391 This ioctl (when supported) replaces the KVM_GET_SREGS.
5396 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5397 struct kvm_segment cs, ds, es, fs, gs, ss;
5398 struct kvm_segment tr, ldt;
5399 struct kvm_dtable gdt, idt;
5400 __u64 cr0, cr2, cr3, cr4, cr8;
5407 flags values for ``kvm_sregs2``:
5409 ``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5411 Indicates thats the struct contain valid PDPTR values.
5414 4.132 KVM_SET_SREGS2
5415 --------------------
5417 :Capability: KVM_CAP_SREGS2
5420 :Parameters: struct kvm_sregs2 (in)
5421 :Returns: 0 on success, -1 on error
5423 Writes special registers into the vcpu.
5424 See KVM_GET_SREGS2 for the data structures.
5425 This ioctl (when supported) replaces the KVM_SET_SREGS.
5427 4.133 KVM_GET_STATS_FD
5428 ----------------------
5430 :Capability: KVM_CAP_STATS_BINARY_FD
5432 :Type: vm ioctl, vcpu ioctl
5434 :Returns: statistics file descriptor on success, < 0 on error
5438 ====== ======================================================
5439 ENOMEM if the fd could not be created due to lack of memory
5440 EMFILE if the number of opened files exceeds the limit
5441 ====== ======================================================
5443 The returned file descriptor can be used to read VM/vCPU statistics data in
5444 binary format. The data in the file descriptor consists of four blocks
5445 organized as follows:
5457 Apart from the header starting at offset 0, please be aware that it is
5458 not guaranteed that the four blocks are adjacent or in the above order;
5459 the offsets of the id, descriptors and data blocks are found in the
5460 header. However, all four blocks are aligned to 64 bit offsets in the
5461 file and they do not overlap.
5463 All blocks except the data block are immutable. Userspace can read them
5464 only one time after retrieving the file descriptor, and then use ``pread`` or
5465 ``lseek`` to read the statistics repeatedly.
5467 All data is in system endianness.
5469 The format of the header is as follows::
5471 struct kvm_stats_header {
5480 The ``flags`` field is not used at the moment. It is always read as 0.
5482 The ``name_size`` field is the size (in byte) of the statistics name string
5483 (including trailing '\0') which is contained in the "id string" block and
5484 appended at the end of every descriptor.
5486 The ``num_desc`` field is the number of descriptors that are included in the
5487 descriptor block. (The actual number of values in the data block may be
5488 larger, since each descriptor may comprise more than one value).
5490 The ``id_offset`` field is the offset of the id string from the start of the
5491 file indicated by the file descriptor. It is a multiple of 8.
5493 The ``desc_offset`` field is the offset of the Descriptors block from the start
5494 of the file indicated by the file descriptor. It is a multiple of 8.
5496 The ``data_offset`` field is the offset of the Stats Data block from the start
5497 of the file indicated by the file descriptor. It is a multiple of 8.
5499 The id string block contains a string which identifies the file descriptor on
5500 which KVM_GET_STATS_FD was invoked. The size of the block, including the
5501 trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5503 The descriptors block is only needed to be read once for the lifetime of the
5504 file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5505 by a string of size ``name_size``.
5508 #define KVM_STATS_TYPE_SHIFT 0
5509 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT)
5510 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT)
5511 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT)
5512 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT)
5513 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT)
5514 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT)
5515 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST
5517 #define KVM_STATS_UNIT_SHIFT 4
5518 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT)
5519 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT)
5520 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT)
5521 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT)
5522 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT)
5523 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_CYCLES
5525 #define KVM_STATS_BASE_SHIFT 8
5526 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT)
5527 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT)
5528 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT)
5529 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2
5531 struct kvm_stats_desc {
5540 The ``flags`` field contains the type and unit of the statistics data described
5541 by this descriptor. Its endianness is CPU native.
5542 The following flags are supported:
5544 Bits 0-3 of ``flags`` encode the type:
5546 * ``KVM_STATS_TYPE_CUMULATIVE``
5547 The statistics reports a cumulative count. The value of data can only be increased.
5548 Most of the counters used in KVM are of this type.
5549 The corresponding ``size`` field for this type is always 1.
5550 All cumulative statistics data are read/write.
5551 * ``KVM_STATS_TYPE_INSTANT``
5552 The statistics reports an instantaneous value. Its value can be increased or
5553 decreased. This type is usually used as a measurement of some resources,
5554 like the number of dirty pages, the number of large pages, etc.
5555 All instant statistics are read only.
5556 The corresponding ``size`` field for this type is always 1.
5557 * ``KVM_STATS_TYPE_PEAK``
5558 The statistics data reports a peak value, for example the maximum number
5559 of items in a hash table bucket, the longest time waited and so on.
5560 The value of data can only be increased.
5561 The corresponding ``size`` field for this type is always 1.
5562 * ``KVM_STATS_TYPE_LINEAR_HIST``
5563 The statistic is reported as a linear histogram. The number of
5564 buckets is specified by the ``size`` field. The size of buckets is specified
5565 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``)
5566 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last
5567 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity
5568 value.) The bucket value indicates how many samples fell in the bucket's range.
5569 * ``KVM_STATS_TYPE_LOG_HIST``
5570 The statistic is reported as a logarithmic histogram. The number of
5571 buckets is specified by the ``size`` field. The range of the first bucket is
5572 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF).
5573 Otherwise, The Nth bucket (1 < N < ``size``) covers
5574 [pow(2, N-2), pow(2, N-1)). The bucket value indicates how many samples fell
5575 in the bucket's range.
5577 Bits 4-7 of ``flags`` encode the unit:
5579 * ``KVM_STATS_UNIT_NONE``
5580 There is no unit for the value of statistics data. This usually means that
5581 the value is a simple counter of an event.
5582 * ``KVM_STATS_UNIT_BYTES``
5583 It indicates that the statistics data is used to measure memory size, in the
5584 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
5585 determined by the ``exponent`` field in the descriptor.
5586 * ``KVM_STATS_UNIT_SECONDS``
5587 It indicates that the statistics data is used to measure time or latency.
5588 * ``KVM_STATS_UNIT_CYCLES``
5589 It indicates that the statistics data is used to measure CPU clock cycles.
5591 Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
5594 * ``KVM_STATS_BASE_POW10``
5595 The scale is based on power of 10. It is used for measurement of time and
5596 CPU clock cycles. For example, an exponent of -9 can be used with
5597 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
5598 * ``KVM_STATS_BASE_POW2``
5599 The scale is based on power of 2. It is used for measurement of memory size.
5600 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
5601 express that the unit is MiB.
5603 The ``size`` field is the number of values of this statistics data. Its
5604 value is usually 1 for most of simple statistics. 1 means it contains an
5605 unsigned 64bit data.
5607 The ``offset`` field is the offset from the start of Data Block to the start of
5608 the corresponding statistics data.
5610 The ``bucket_size`` field is used as a parameter for histogram statistics data.
5611 It is only used by linear histogram statistics data, specifying the size of a
5614 The ``name`` field is the name string of the statistics data. The name string
5615 starts at the end of ``struct kvm_stats_desc``. The maximum length including
5616 the trailing ``'\0'``, is indicated by ``name_size`` in the header.
5618 The Stats Data block contains an array of 64-bit values in the same order
5619 as the descriptors in Descriptors block.
5621 4.134 KVM_GET_XSAVE2
5622 --------------------
5624 :Capability: KVM_CAP_XSAVE2
5627 :Parameters: struct kvm_xsave (out)
5628 :Returns: 0 on success, -1 on error
5638 This ioctl would copy current vcpu's xsave struct to the userspace. It
5639 copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2)
5640 when invoked on the vm file descriptor. The size value returned by
5641 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
5642 Currently, it is only greater than 4096 if a dynamic feature has been
5643 enabled with ``arch_prctl()``, but this may change in the future.
5645 The offsets of the state save areas in struct kvm_xsave follow the contents
5646 of CPUID leaf 0xD on the host.
5649 5. The kvm_run structure
5650 ========================
5652 Application code obtains a pointer to the kvm_run structure by
5653 mmap()ing a vcpu fd. From that point, application code can control
5654 execution by changing fields in kvm_run prior to calling the KVM_RUN
5655 ioctl, and obtain information about the reason KVM_RUN returned by
5656 looking up structure members.
5662 __u8 request_interrupt_window;
5664 Request that KVM_RUN return when it becomes possible to inject external
5665 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
5669 __u8 immediate_exit;
5671 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
5672 exits immediately, returning -EINTR. In the common scenario where a
5673 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
5674 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
5675 Rather than blocking the signal outside KVM_RUN, userspace can set up
5676 a signal handler that sets run->immediate_exit to a non-zero value.
5678 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
5687 When KVM_RUN has returned successfully (return value 0), this informs
5688 application code why KVM_RUN has returned. Allowable values for this
5689 field are detailed below.
5693 __u8 ready_for_interrupt_injection;
5695 If request_interrupt_window has been specified, this field indicates
5696 an interrupt can be injected now with KVM_INTERRUPT.
5702 The value of the current interrupt flag. Only valid if in-kernel
5703 local APIC is not used.
5709 More architecture-specific flags detailing state of the VCPU that may
5710 affect the device's behavior. Current defined flags::
5712 /* x86, set if the VCPU is in system management mode */
5713 #define KVM_RUN_X86_SMM (1 << 0)
5714 /* x86, set if bus lock detected in VM */
5715 #define KVM_RUN_BUS_LOCK (1 << 1)
5716 /* arm64, set for KVM_EXIT_DEBUG */
5717 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0)
5721 /* in (pre_kvm_run), out (post_kvm_run) */
5724 The value of the cr8 register. Only valid if in-kernel local APIC is
5725 not used. Both input and output.
5731 The value of the APIC BASE msr. Only valid if in-kernel local
5732 APIC is not used. Both input and output.
5737 /* KVM_EXIT_UNKNOWN */
5739 __u64 hardware_exit_reason;
5742 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
5743 reasons. Further architecture-specific information is available in
5744 hardware_exit_reason.
5748 /* KVM_EXIT_FAIL_ENTRY */
5750 __u64 hardware_entry_failure_reason;
5751 __u32 cpu; /* if KVM_LAST_CPU */
5754 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
5755 to unknown reasons. Further architecture-specific information is
5756 available in hardware_entry_failure_reason.
5760 /* KVM_EXIT_EXCEPTION */
5772 #define KVM_EXIT_IO_IN 0
5773 #define KVM_EXIT_IO_OUT 1
5775 __u8 size; /* bytes */
5778 __u64 data_offset; /* relative to kvm_run start */
5781 If exit_reason is KVM_EXIT_IO, then the vcpu has
5782 executed a port I/O instruction which could not be satisfied by kvm.
5783 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
5784 where kvm expects application code to place the data for the next
5785 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
5789 /* KVM_EXIT_DEBUG */
5791 struct kvm_debug_exit_arch arch;
5794 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
5795 for which architecture specific information is returned.
5807 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
5808 executed a memory-mapped I/O instruction which could not be satisfied
5809 by kvm. The 'data' member contains the written data if 'is_write' is
5810 true, and should be filled by application code otherwise.
5812 The 'data' member contains, in its first 'len' bytes, the value as it would
5813 appear if the VCPU performed a load or store of the appropriate width directly
5818 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
5819 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5820 operations are complete (and guest state is consistent) only after userspace
5821 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5822 incomplete operations and then check for pending signals.
5824 The pending state of the operation is not preserved in state which is
5825 visible to userspace, thus userspace should ensure that the operation is
5826 completed before performing a live migration. Userspace can re-enter the
5827 guest with an unmasked signal pending or with the immediate_exit field set
5828 to complete pending operations without allowing any further instructions
5833 /* KVM_EXIT_HYPERCALL */
5842 Unused. This was once used for 'hypercall to userspace'. To implement
5843 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5845 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5849 /* KVM_EXIT_TPR_ACCESS */
5856 To be documented (KVM_TPR_ACCESS_REPORTING).
5860 /* KVM_EXIT_S390_SIEIC */
5863 __u64 mask; /* psw upper half */
5864 __u64 addr; /* psw lower half */
5873 /* KVM_EXIT_S390_RESET */
5874 #define KVM_S390_RESET_POR 1
5875 #define KVM_S390_RESET_CLEAR 2
5876 #define KVM_S390_RESET_SUBSYSTEM 4
5877 #define KVM_S390_RESET_CPU_INIT 8
5878 #define KVM_S390_RESET_IPL 16
5879 __u64 s390_reset_flags;
5885 /* KVM_EXIT_S390_UCONTROL */
5887 __u64 trans_exc_code;
5891 s390 specific. A page fault has occurred for a user controlled virtual
5892 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5893 resolved by the kernel.
5894 The program code and the translation exception code that were placed
5895 in the cpu's lowcore are presented here as defined by the z Architecture
5896 Principles of Operation Book in the Chapter for Dynamic Address Translation
5908 Deprecated - was used for 440 KVM.
5917 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5918 hypercalls and exit with this exit struct that contains all the guest gprs.
5920 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5921 Userspace can now handle the hypercall and when it's done modify the gprs as
5922 necessary. Upon guest entry all guest GPRs will then be replaced by the values
5927 /* KVM_EXIT_PAPR_HCALL */
5934 This is used on 64-bit PowerPC when emulating a pSeries partition,
5935 e.g. with the 'pseries' machine type in qemu. It occurs when the
5936 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
5937 contains the hypercall number (from the guest R3), and 'args' contains
5938 the arguments (from the guest R4 - R12). Userspace should put the
5939 return code in 'ret' and any extra returned values in args[].
5940 The possible hypercalls are defined in the Power Architecture Platform
5941 Requirements (PAPR) document available from www.power.org (free
5942 developer registration required to access it).
5946 /* KVM_EXIT_S390_TSCH */
5948 __u16 subchannel_id;
5949 __u16 subchannel_nr;
5956 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5957 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5958 interrupt for the target subchannel has been dequeued and subchannel_id,
5959 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5960 interrupt. ipb is needed for instruction parameter decoding.
5969 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5970 interrupt acknowledge path to the core. When the core successfully
5971 delivers an interrupt, it automatically populates the EPR register with
5972 the interrupt vector number and acknowledges the interrupt inside
5973 the interrupt controller.
5975 In case the interrupt controller lives in user space, we need to do
5976 the interrupt acknowledge cycle through it to fetch the next to be
5977 delivered interrupt vector using this exit.
5979 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5980 external interrupt has just been delivered into the guest. User space
5981 should put the acknowledged interrupt vector into the 'epr' field.
5985 /* KVM_EXIT_SYSTEM_EVENT */
5987 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
5988 #define KVM_SYSTEM_EVENT_RESET 2
5989 #define KVM_SYSTEM_EVENT_CRASH 3
5995 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5996 a system-level event using some architecture specific mechanism (hypercall
5997 or some special instruction). In case of ARM64, this is triggered using
5998 HVC instruction based PSCI call from the vcpu.
6000 The 'type' field describes the system-level event type.
6001 Valid values for 'type' are:
6003 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
6004 VM. Userspace is not obliged to honour this, and if it does honour
6005 this does not need to destroy the VM synchronously (ie it may call
6006 KVM_RUN again before shutdown finally occurs).
6007 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
6008 As with SHUTDOWN, userspace can choose to ignore the request, or
6009 to schedule the reset to occur in the future and may call KVM_RUN again.
6010 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
6011 has requested a crash condition maintenance. Userspace can choose
6012 to ignore the request, or to gather VM memory core dump and/or
6013 reset/shutdown of the VM.
6015 If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain
6016 architecture specific information for the system-level event. Only
6017 the first `ndata` items (possibly zero) of the data array are valid.
6019 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if
6020 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI
6023 - for RISC-V, data[0] is set to the value of the second argument of the
6024 ``sbi_system_reset`` call.
6026 Previous versions of Linux defined a `flags` member in this struct. The
6027 field is now aliased to `data[0]`. Userspace can assume that it is only
6028 written if ndata is greater than 0.
6032 /* KVM_EXIT_IOAPIC_EOI */
6037 Indicates that the VCPU's in-kernel local APIC received an EOI for a
6038 level-triggered IOAPIC interrupt. This exit only triggers when the
6039 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
6040 the userspace IOAPIC should process the EOI and retrigger the interrupt if
6041 it is still asserted. Vector is the LAPIC interrupt vector for which the
6046 struct kvm_hyperv_exit {
6047 #define KVM_EXIT_HYPERV_SYNIC 1
6048 #define KVM_EXIT_HYPERV_HCALL 2
6049 #define KVM_EXIT_HYPERV_SYNDBG 3
6076 /* KVM_EXIT_HYPERV */
6077 struct kvm_hyperv_exit hyperv;
6079 Indicates that the VCPU exits into userspace to process some tasks
6080 related to Hyper-V emulation.
6082 Valid values for 'type' are:
6084 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
6086 Hyper-V SynIC state change. Notification is used to remap SynIC
6087 event/message pages and to enable/disable SynIC messages/events processing
6090 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
6092 Hyper-V Synthetic debugger state change. Notification is used to either update
6093 the pending_page location or to send a control command (send the buffer located
6094 in send_page or recv a buffer to recv_page).
6098 /* KVM_EXIT_ARM_NISV */
6104 Used on arm64 systems. If a guest accesses memory not in a memslot,
6105 KVM will typically return to userspace and ask it to do MMIO emulation on its
6106 behalf. However, for certain classes of instructions, no instruction decode
6107 (direction, length of memory access) is provided, and fetching and decoding
6108 the instruction from the VM is overly complicated to live in the kernel.
6110 Historically, when this situation occurred, KVM would print a warning and kill
6111 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
6112 trying to do I/O, which just couldn't be emulated, and the warning message was
6113 phrased accordingly. However, what happened more often was that a guest bug
6114 caused access outside the guest memory areas which should lead to a more
6115 meaningful warning message and an external abort in the guest, if the access
6116 did not fall within an I/O window.
6118 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
6119 this capability at VM creation. Once this is done, these types of errors will
6120 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
6121 the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field.
6122 Userspace can either fix up the access if it's actually an I/O access by
6123 decoding the instruction from guest memory (if it's very brave) and continue
6124 executing the guest, or it can decide to suspend, dump, or restart the guest.
6126 Note that KVM does not skip the faulting instruction as it does for
6127 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
6128 if it decides to decode and emulate the instruction.
6132 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
6134 __u8 error; /* user -> kernel */
6136 __u32 reason; /* kernel -> user */
6137 __u32 index; /* kernel -> user */
6138 __u64 data; /* kernel <-> user */
6141 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
6142 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
6143 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
6146 The "reason" field specifies why the MSR trap occurred. User space will only
6147 receive MSR exit traps when a particular reason was requested during through
6148 ENABLE_CAP. Currently valid exit reasons are:
6150 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
6151 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
6152 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
6154 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
6155 wants to read. To respond to this request with a successful read, user space
6156 writes the respective data into the "data" field and must continue guest
6157 execution to ensure the read data is transferred into guest register state.
6159 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
6160 the "error" field. This will inject a #GP into the guest when the VCPU is
6163 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
6164 wants to write. Once finished processing the event, user space must continue
6165 vCPU execution. If the MSR write was unsuccessful, user space also sets the
6166 "error" field to "1".
6171 struct kvm_xen_exit {
6172 #define KVM_EXIT_XEN_HCALL 1
6185 struct kvm_hyperv_exit xen;
6187 Indicates that the VCPU exits into userspace to process some tasks
6188 related to Xen emulation.
6190 Valid values for 'type' are:
6192 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
6193 Userspace is expected to place the hypercall result into the appropriate
6194 field before invoking KVM_RUN again.
6198 /* KVM_EXIT_RISCV_SBI */
6200 unsigned long extension_id;
6201 unsigned long function_id;
6202 unsigned long args[6];
6203 unsigned long ret[2];
6206 If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has
6207 done a SBI call which is not handled by KVM RISC-V kernel module. The details
6208 of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The
6209 'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the
6210 'function_id' field represents function ID of given SBI extension. The 'args'
6211 array field of 'riscv_sbi' represents parameters for the SBI call and 'ret'
6212 array field represents return values. The userspace should update the return
6213 values of SBI call before resuming the VCPU. For more details on RISC-V SBI
6214 spec refer, https://github.com/riscv/riscv-sbi-doc.
6218 /* Fix the size of the union. */
6223 * shared registers between kvm and userspace.
6224 * kvm_valid_regs specifies the register classes set by the host
6225 * kvm_dirty_regs specified the register classes dirtied by userspace
6226 * struct kvm_sync_regs is architecture specific, as well as the
6227 * bits for kvm_valid_regs and kvm_dirty_regs
6229 __u64 kvm_valid_regs;
6230 __u64 kvm_dirty_regs;
6232 struct kvm_sync_regs regs;
6233 char padding[SYNC_REGS_SIZE_BYTES];
6236 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
6237 certain guest registers without having to call SET/GET_*REGS. Thus we can
6238 avoid some system call overhead if userspace has to handle the exit.
6239 Userspace can query the validity of the structure by checking
6240 kvm_valid_regs for specific bits. These bits are architecture specific
6241 and usually define the validity of a groups of registers. (e.g. one bit
6242 for general purpose registers)
6244 Please note that the kernel is allowed to use the kvm_run structure as the
6245 primary storage for certain register types. Therefore, the kernel may use the
6246 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
6254 6. Capabilities that can be enabled on vCPUs
6255 ============================================
6257 There are certain capabilities that change the behavior of the virtual CPU or
6258 the virtual machine when enabled. To enable them, please see section 4.37.
6259 Below you can find a list of capabilities and what their effect on the vCPU or
6260 the virtual machine is when enabling them.
6262 The following information is provided along with the description:
6265 which instruction set architectures provide this ioctl.
6266 x86 includes both i386 and x86_64.
6269 whether this is a per-vcpu or per-vm capability.
6272 what parameters are accepted by the capability.
6275 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6276 are not detailed, but errors with specific meanings are.
6285 :Returns: 0 on success; -1 on error
6287 This capability enables interception of OSI hypercalls that otherwise would
6288 be treated as normal system calls to be injected into the guest. OSI hypercalls
6289 were invented by Mac-on-Linux to have a standardized communication mechanism
6290 between the guest and the host.
6292 When this capability is enabled, KVM_EXIT_OSI can occur.
6295 6.2 KVM_CAP_PPC_PAPR
6296 --------------------
6301 :Returns: 0 on success; -1 on error
6303 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
6304 done using the hypercall instruction "sc 1".
6306 It also sets the guest privilege level to "supervisor" mode. Usually the guest
6307 runs in "hypervisor" privilege mode with a few missing features.
6309 In addition to the above, it changes the semantics of SDR1. In this mode, the
6310 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
6311 HTAB invisible to the guest.
6313 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
6321 :Parameters: args[0] is the address of a struct kvm_config_tlb
6322 :Returns: 0 on success; -1 on error
6326 struct kvm_config_tlb {
6333 Configures the virtual CPU's TLB array, establishing a shared memory area
6334 between userspace and KVM. The "params" and "array" fields are userspace
6335 addresses of mmu-type-specific data structures. The "array_len" field is an
6336 safety mechanism, and should be set to the size in bytes of the memory that
6337 userspace has reserved for the array. It must be at least the size dictated
6338 by "mmu_type" and "params".
6340 While KVM_RUN is active, the shared region is under control of KVM. Its
6341 contents are undefined, and any modification by userspace results in
6342 boundedly undefined behavior.
6344 On return from KVM_RUN, the shared region will reflect the current state of
6345 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
6346 to tell KVM which entries have been changed, prior to calling KVM_RUN again
6349 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
6351 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
6352 - The "array" field points to an array of type "struct
6353 kvm_book3e_206_tlb_entry".
6354 - The array consists of all entries in the first TLB, followed by all
6355 entries in the second TLB.
6356 - Within a TLB, entries are ordered first by increasing set number. Within a
6357 set, entries are ordered by way (increasing ESEL).
6358 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
6359 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
6360 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
6361 hardware ignores this value for TLB0.
6363 6.4 KVM_CAP_S390_CSS_SUPPORT
6364 ----------------------------
6366 :Architectures: s390
6369 :Returns: 0 on success; -1 on error
6371 This capability enables support for handling of channel I/O instructions.
6373 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
6374 handled in-kernel, while the other I/O instructions are passed to userspace.
6376 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
6377 SUBCHANNEL intercepts.
6379 Note that even though this capability is enabled per-vcpu, the complete
6380 virtual machine is affected.
6387 :Parameters: args[0] defines whether the proxy facility is active
6388 :Returns: 0 on success; -1 on error
6390 This capability enables or disables the delivery of interrupts through the
6391 external proxy facility.
6393 When enabled (args[0] != 0), every time the guest gets an external interrupt
6394 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
6395 to receive the topmost interrupt vector.
6397 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
6399 When this capability is enabled, KVM_EXIT_EPR can occur.
6401 6.6 KVM_CAP_IRQ_MPIC
6402 --------------------
6405 :Parameters: args[0] is the MPIC device fd;
6406 args[1] is the MPIC CPU number for this vcpu
6408 This capability connects the vcpu to an in-kernel MPIC device.
6410 6.7 KVM_CAP_IRQ_XICS
6411 --------------------
6415 :Parameters: args[0] is the XICS device fd;
6416 args[1] is the XICS CPU number (server ID) for this vcpu
6418 This capability connects the vcpu to an in-kernel XICS device.
6420 6.8 KVM_CAP_S390_IRQCHIP
6421 ------------------------
6423 :Architectures: s390
6427 This capability enables the in-kernel irqchip for s390. Please refer to
6428 "4.24 KVM_CREATE_IRQCHIP" for details.
6430 6.9 KVM_CAP_MIPS_FPU
6431 --------------------
6433 :Architectures: mips
6435 :Parameters: args[0] is reserved for future use (should be 0).
6437 This capability allows the use of the host Floating Point Unit by the guest. It
6438 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
6439 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
6440 accessed (depending on the current guest FPU register mode), and the Status.FR,
6441 Config5.FRE bits are accessible via the KVM API and also from the guest,
6442 depending on them being supported by the FPU.
6444 6.10 KVM_CAP_MIPS_MSA
6445 ---------------------
6447 :Architectures: mips
6449 :Parameters: args[0] is reserved for future use (should be 0).
6451 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
6452 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
6453 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
6454 registers can be accessed, and the Config5.MSAEn bit is accessible via the
6455 KVM API and also from the guest.
6457 6.74 KVM_CAP_SYNC_REGS
6458 ----------------------
6460 :Architectures: s390, x86
6461 :Target: s390: always enabled, x86: vcpu
6463 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
6465 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
6467 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
6468 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
6469 without having to call SET/GET_*REGS". This reduces overhead by eliminating
6470 repeated ioctl calls for setting and/or getting register values. This is
6471 particularly important when userspace is making synchronous guest state
6472 modifications, e.g. when emulating and/or intercepting instructions in
6475 For s390 specifics, please refer to the source code.
6479 - the register sets to be copied out to kvm_run are selectable
6480 by userspace (rather that all sets being copied out for every exit).
6481 - vcpu_events are available in addition to regs and sregs.
6483 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
6484 function as an input bit-array field set by userspace to indicate the
6485 specific register sets to be copied out on the next exit.
6487 To indicate when userspace has modified values that should be copied into
6488 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
6489 This is done using the same bitflags as for the 'kvm_valid_regs' field.
6490 If the dirty bit is not set, then the register set values will not be copied
6491 into the vCPU even if they've been modified.
6493 Unused bitfields in the bitarrays must be set to zero.
6497 struct kvm_sync_regs {
6498 struct kvm_regs regs;
6499 struct kvm_sregs sregs;
6500 struct kvm_vcpu_events events;
6503 6.75 KVM_CAP_PPC_IRQ_XIVE
6504 -------------------------
6508 :Parameters: args[0] is the XIVE device fd;
6509 args[1] is the XIVE CPU number (server ID) for this vcpu
6511 This capability connects the vcpu to an in-kernel XIVE device.
6513 7. Capabilities that can be enabled on VMs
6514 ==========================================
6516 There are certain capabilities that change the behavior of the virtual
6517 machine when enabled. To enable them, please see section 4.37. Below
6518 you can find a list of capabilities and what their effect on the VM
6519 is when enabling them.
6521 The following information is provided along with the description:
6524 which instruction set architectures provide this ioctl.
6525 x86 includes both i386 and x86_64.
6528 what parameters are accepted by the capability.
6531 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6532 are not detailed, but errors with specific meanings are.
6535 7.1 KVM_CAP_PPC_ENABLE_HCALL
6536 ----------------------------
6539 :Parameters: args[0] is the sPAPR hcall number;
6540 args[1] is 0 to disable, 1 to enable in-kernel handling
6542 This capability controls whether individual sPAPR hypercalls (hcalls)
6543 get handled by the kernel or not. Enabling or disabling in-kernel
6544 handling of an hcall is effective across the VM. On creation, an
6545 initial set of hcalls are enabled for in-kernel handling, which
6546 consists of those hcalls for which in-kernel handlers were implemented
6547 before this capability was implemented. If disabled, the kernel will
6548 not to attempt to handle the hcall, but will always exit to userspace
6549 to handle it. Note that it may not make sense to enable some and
6550 disable others of a group of related hcalls, but KVM does not prevent
6551 userspace from doing that.
6553 If the hcall number specified is not one that has an in-kernel
6554 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
6557 7.2 KVM_CAP_S390_USER_SIGP
6558 --------------------------
6560 :Architectures: s390
6563 This capability controls which SIGP orders will be handled completely in user
6564 space. With this capability enabled, all fast orders will be handled completely
6571 - CONDITIONAL EMERGENCY SIGNAL
6573 All other orders will be handled completely in user space.
6575 Only privileged operation exceptions will be checked for in the kernel (or even
6576 in the hardware prior to interception). If this capability is not enabled, the
6577 old way of handling SIGP orders is used (partially in kernel and user space).
6579 7.3 KVM_CAP_S390_VECTOR_REGISTERS
6580 ---------------------------------
6582 :Architectures: s390
6584 :Returns: 0 on success, negative value on error
6586 Allows use of the vector registers introduced with z13 processor, and
6587 provides for the synchronization between host and user space. Will
6588 return -EINVAL if the machine does not support vectors.
6590 7.4 KVM_CAP_S390_USER_STSI
6591 --------------------------
6593 :Architectures: s390
6596 This capability allows post-handlers for the STSI instruction. After
6597 initial handling in the kernel, KVM exits to user space with
6598 KVM_EXIT_S390_STSI to allow user space to insert further data.
6600 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
6612 @addr - guest address of STSI SYSIB
6616 @ar - access register number
6618 KVM handlers should exit to userspace with rc = -EREMOTE.
6620 7.5 KVM_CAP_SPLIT_IRQCHIP
6621 -------------------------
6624 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
6625 :Returns: 0 on success, -1 on error
6627 Create a local apic for each processor in the kernel. This can be used
6628 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
6629 IOAPIC and PIC (and also the PIT, even though this has to be enabled
6632 This capability also enables in kernel routing of interrupt requests;
6633 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
6634 used in the IRQ routing table. The first args[0] MSI routes are reserved
6635 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
6636 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
6638 Fails if VCPU has already been created, or if the irqchip is already in the
6639 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
6644 :Architectures: s390
6647 Allows use of runtime-instrumentation introduced with zEC12 processor.
6648 Will return -EINVAL if the machine does not support runtime-instrumentation.
6649 Will return -EBUSY if a VCPU has already been created.
6651 7.7 KVM_CAP_X2APIC_API
6652 ----------------------
6655 :Parameters: args[0] - features that should be enabled
6656 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
6658 Valid feature flags in args[0] are::
6660 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
6661 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
6663 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
6664 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
6665 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
6666 respective sections.
6668 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
6669 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
6670 as a broadcast even in x2APIC mode in order to support physical x2APIC
6671 without interrupt remapping. This is undesirable in logical mode,
6672 where 0xff represents CPUs 0-7 in cluster 0.
6674 7.8 KVM_CAP_S390_USER_INSTR0
6675 ----------------------------
6677 :Architectures: s390
6680 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
6681 be intercepted and forwarded to user space. User space can use this
6682 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
6683 not inject an operating exception for these instructions, user space has
6684 to take care of that.
6686 This capability can be enabled dynamically even if VCPUs were already
6687 created and are running.
6692 :Architectures: s390
6694 :Returns: 0 on success; -EINVAL if the machine does not support
6695 guarded storage; -EBUSY if a VCPU has already been created.
6697 Allows use of guarded storage for the KVM guest.
6699 7.10 KVM_CAP_S390_AIS
6700 ---------------------
6702 :Architectures: s390
6705 Allow use of adapter-interruption suppression.
6706 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
6708 7.11 KVM_CAP_PPC_SMT
6709 --------------------
6712 :Parameters: vsmt_mode, flags
6714 Enabling this capability on a VM provides userspace with a way to set
6715 the desired virtual SMT mode (i.e. the number of virtual CPUs per
6716 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
6717 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
6718 the number of threads per subcore for the host. Currently flags must
6719 be 0. A successful call to enable this capability will result in
6720 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
6721 subsequently queried for the VM. This capability is only supported by
6722 HV KVM, and can only be set before any VCPUs have been created.
6723 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
6724 modes are available.
6726 7.12 KVM_CAP_PPC_FWNMI
6727 ----------------------
6732 With this capability a machine check exception in the guest address
6733 space will cause KVM to exit the guest with NMI exit reason. This
6734 enables QEMU to build error log and branch to guest kernel registered
6735 machine check handling routine. Without this capability KVM will
6736 branch to guests' 0x200 interrupt vector.
6738 7.13 KVM_CAP_X86_DISABLE_EXITS
6739 ------------------------------
6742 :Parameters: args[0] defines which exits are disabled
6743 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
6745 Valid bits in args[0] are::
6747 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
6748 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
6749 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
6750 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
6752 Enabling this capability on a VM provides userspace with a way to no
6753 longer intercept some instructions for improved latency in some
6754 workloads, and is suggested when vCPUs are associated to dedicated
6755 physical CPUs. More bits can be added in the future; userspace can
6756 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
6759 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
6761 7.14 KVM_CAP_S390_HPAGE_1M
6762 --------------------------
6764 :Architectures: s390
6766 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
6767 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
6770 With this capability the KVM support for memory backing with 1m pages
6771 through hugetlbfs can be enabled for a VM. After the capability is
6772 enabled, cmma can't be enabled anymore and pfmfi and the storage key
6773 interpretation are disabled. If cmma has already been enabled or the
6774 hpage module parameter is not set to 1, -EINVAL is returned.
6776 While it is generally possible to create a huge page backed VM without
6777 this capability, the VM will not be able to run.
6779 7.15 KVM_CAP_MSR_PLATFORM_INFO
6780 ------------------------------
6783 :Parameters: args[0] whether feature should be enabled or not
6785 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
6786 a #GP would be raised when the guest tries to access. Currently, this
6787 capability does not enable write permissions of this MSR for the guest.
6789 7.16 KVM_CAP_PPC_NESTED_HV
6790 --------------------------
6794 :Returns: 0 on success, -EINVAL when the implementation doesn't support
6795 nested-HV virtualization.
6797 HV-KVM on POWER9 and later systems allows for "nested-HV"
6798 virtualization, which provides a way for a guest VM to run guests that
6799 can run using the CPU's supervisor mode (privileged non-hypervisor
6800 state). Enabling this capability on a VM depends on the CPU having
6801 the necessary functionality and on the facility being enabled with a
6802 kvm-hv module parameter.
6804 7.17 KVM_CAP_EXCEPTION_PAYLOAD
6805 ------------------------------
6808 :Parameters: args[0] whether feature should be enabled or not
6810 With this capability enabled, CR2 will not be modified prior to the
6811 emulated VM-exit when L1 intercepts a #PF exception that occurs in
6812 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
6813 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
6814 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
6815 #DB) exception for L2, exception.has_payload will be set and the
6816 faulting address (or the new DR6 bits*) will be reported in the
6817 exception_payload field. Similarly, when userspace injects a #PF (or
6818 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
6819 exception.has_payload and to put the faulting address - or the new DR6
6820 bits\ [#]_ - in the exception_payload field.
6822 This capability also enables exception.pending in struct
6823 kvm_vcpu_events, which allows userspace to distinguish between pending
6824 and injected exceptions.
6827 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
6830 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
6832 :Architectures: x86, arm64, mips
6833 :Parameters: args[0] whether feature should be enabled or not
6837 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
6838 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
6840 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
6841 automatically clear and write-protect all pages that are returned as dirty.
6842 Rather, userspace will have to do this operation separately using
6843 KVM_CLEAR_DIRTY_LOG.
6845 At the cost of a slightly more complicated operation, this provides better
6846 scalability and responsiveness for two reasons. First,
6847 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
6848 than requiring to sync a full memslot; this ensures that KVM does not
6849 take spinlocks for an extended period of time. Second, in some cases a
6850 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
6851 userspace actually using the data in the page. Pages can be modified
6852 during this time, which is inefficient for both the guest and userspace:
6853 the guest will incur a higher penalty due to write protection faults,
6854 while userspace can see false reports of dirty pages. Manual reprotection
6855 helps reducing this time, improving guest performance and reducing the
6856 number of dirty log false positives.
6858 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
6859 will be initialized to 1 when created. This also improves performance because
6860 dirty logging can be enabled gradually in small chunks on the first call
6861 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
6862 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
6863 x86 and arm64 for now).
6865 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
6866 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
6867 it hard or impossible to use it correctly. The availability of
6868 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
6869 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
6871 7.19 KVM_CAP_PPC_SECURE_GUEST
6872 ------------------------------
6876 This capability indicates that KVM is running on a host that has
6877 ultravisor firmware and thus can support a secure guest. On such a
6878 system, a guest can ask the ultravisor to make it a secure guest,
6879 one whose memory is inaccessible to the host except for pages which
6880 are explicitly requested to be shared with the host. The ultravisor
6881 notifies KVM when a guest requests to become a secure guest, and KVM
6882 has the opportunity to veto the transition.
6884 If present, this capability can be enabled for a VM, meaning that KVM
6885 will allow the transition to secure guest mode. Otherwise KVM will
6886 veto the transition.
6888 7.20 KVM_CAP_HALT_POLL
6889 ----------------------
6893 :Parameters: args[0] is the maximum poll time in nanoseconds
6894 :Returns: 0 on success; -1 on error
6896 This capability overrides the kvm module parameter halt_poll_ns for the
6899 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6900 scheduling during guest halts. The maximum time a VCPU can spend polling is
6901 controlled by the kvm module parameter halt_poll_ns. This capability allows
6902 the maximum halt time to specified on a per-VM basis, effectively overriding
6903 the module parameter for the target VM.
6905 7.21 KVM_CAP_X86_USER_SPACE_MSR
6906 -------------------------------
6910 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6911 :Returns: 0 on success; -1 on error
6913 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6916 When a guest requests to read or write an MSR, KVM may not implement all MSRs
6917 that are relevant to a respective system. It also does not differentiate by
6920 To allow more fine grained control over MSR handling, user space may enable
6921 this capability. With it enabled, MSR accesses that match the mask specified in
6922 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6923 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6924 can then handle to implement model specific MSR handling and/or user notifications
6925 to inform a user that an MSR was not handled.
6927 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
6928 -------------------------------
6932 :Parameters: args[0] defines the policy used when bus locks detected in guest
6933 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
6935 Valid bits in args[0] are::
6937 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
6938 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
6940 Enabling this capability on a VM provides userspace with a way to select
6941 a policy to handle the bus locks detected in guest. Userspace can obtain
6942 the supported modes from the result of KVM_CHECK_EXTENSION and define it
6943 through the KVM_ENABLE_CAP.
6945 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
6946 currently and mutually exclusive with each other. More bits can be added in
6949 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
6950 so that no additional actions are needed. This is the default mode.
6952 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
6953 in VM. KVM just exits to userspace when handling them. Userspace can enforce
6954 its own throttling or other policy based mitigations.
6956 This capability is aimed to address the thread that VM can exploit bus locks to
6957 degree the performance of the whole system. Once the userspace enable this
6958 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
6959 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
6960 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
6961 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
6962 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
6964 7.23 KVM_CAP_PPC_DAWR1
6965 ----------------------
6969 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
6971 This capability can be used to check / enable 2nd DAWR feature provided
6972 by POWER10 processor.
6975 7.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
6976 -------------------------------------
6978 Architectures: x86 SEV enabled
6980 Parameters: args[0] is the fd of the source vm
6981 Returns: 0 on success; ENOTTY on error
6983 This capability enables userspace to copy encryption context from the vm
6984 indicated by the fd to the vm this is called on.
6986 This is intended to support in-guest workloads scheduled by the host. This
6987 allows the in-guest workload to maintain its own NPTs and keeps the two vms
6988 from accidentally clobbering each other with interrupts and the like (separate
6991 7.25 KVM_CAP_SGX_ATTRIBUTE
6992 --------------------------
6996 :Parameters: args[0] is a file handle of a SGX attribute file in securityfs
6997 :Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
6998 attribute is not supported by KVM.
7000 KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
7001 more priveleged enclave attributes. args[0] must hold a file handle to a valid
7002 SGX attribute file corresponding to an attribute that is supported/restricted
7003 by KVM (currently only PROVISIONKEY).
7005 The SGX subsystem restricts access to a subset of enclave attributes to provide
7006 additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
7007 is restricted to deter malware from using the PROVISIONKEY to obtain a stable
7008 system fingerprint. To prevent userspace from circumventing such restrictions
7009 by running an enclave in a VM, KVM prevents access to privileged attributes by
7012 See Documentation/x86/sgx.rst for more details.
7014 7.26 KVM_CAP_PPC_RPT_INVALIDATE
7015 -------------------------------
7017 :Capability: KVM_CAP_PPC_RPT_INVALIDATE
7021 This capability indicates that the kernel is capable of handling
7022 H_RPT_INVALIDATE hcall.
7024 In order to enable the use of H_RPT_INVALIDATE in the guest,
7025 user space might have to advertise it for the guest. For example,
7026 IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
7027 present in the "ibm,hypertas-functions" device-tree property.
7029 This capability is enabled for hypervisors on platforms like POWER9
7030 that support radix MMU.
7032 7.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
7033 --------------------------------------
7036 :Parameters: args[0] whether the feature should be enabled or not
7038 When this capability is enabled, an emulation failure will result in an exit
7039 to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
7040 to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
7041 to 15 instruction bytes for any exit to userspace resulting from an emulation
7042 failure. When these exits to userspace occur use the emulation_failure struct
7043 instead of the internal struct. They both have the same layout, but the
7044 emulation_failure struct matches the content better. It also explicitly
7045 defines the 'flags' field which is used to describe the fields in the struct
7046 that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
7047 set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
7050 7.28 KVM_CAP_ARM_MTE
7051 --------------------
7053 :Architectures: arm64
7056 This capability indicates that KVM (and the hardware) supports exposing the
7057 Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
7058 VMM before creating any VCPUs to allow the guest access. Note that MTE is only
7059 available to a guest running in AArch64 mode and enabling this capability will
7060 cause attempts to create AArch32 VCPUs to fail.
7062 When enabled the guest is able to access tags associated with any memory given
7063 to the guest. KVM will ensure that the tags are maintained during swap or
7064 hibernation of the host; however the VMM needs to manually save/restore the
7065 tags as appropriate if the VM is migrated.
7067 When this capability is enabled all memory in memslots must be mapped as
7068 not-shareable (no MAP_SHARED), attempts to create a memslot with a
7069 MAP_SHARED mmap will result in an -EINVAL return.
7071 When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
7072 perform a bulk copy of tags to/from the guest.
7074 7.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM
7075 -------------------------------------
7077 Architectures: x86 SEV enabled
7079 Parameters: args[0] is the fd of the source vm
7080 Returns: 0 on success
7082 This capability enables userspace to migrate the encryption context from the VM
7083 indicated by the fd to the VM this is called on.
7085 This is intended to support intra-host migration of VMs between userspace VMMs,
7086 upgrading the VMM process without interrupting the guest.
7088 7.30 KVM_CAP_PPC_AIL_MODE_3
7089 -------------------------------
7091 :Capability: KVM_CAP_PPC_AIL_MODE_3
7095 This capability indicates that the kernel supports the mode 3 setting for the
7096 "Address Translation Mode on Interrupt" aka "Alternate Interrupt Location"
7097 resource that is controlled with the H_SET_MODE hypercall.
7099 This capability allows a guest kernel to use a better-performance mode for
7100 handling interrupts and system calls.
7102 7.31 KVM_CAP_DISABLE_QUIRKS2
7103 ----------------------------
7105 :Capability: KVM_CAP_DISABLE_QUIRKS2
7106 :Parameters: args[0] - set of KVM quirks to disable
7110 This capability, if enabled, will cause KVM to disable some behavior
7113 Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
7114 quirks that can be disabled in KVM.
7116 The argument to KVM_ENABLE_CAP for this capability is a bitmask of
7117 quirks to disable, and must be a subset of the bitmask returned by
7118 KVM_CHECK_EXTENSION.
7120 The valid bits in cap.args[0] are:
7122 =================================== ============================================
7123 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT
7124 LINT0 register is 0x700 (APIC_MODE_EXTINT).
7125 When this quirk is disabled, the reset value
7126 is 0x10000 (APIC_LVT_MASKED).
7128 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW.
7129 When this quirk is disabled, KVM does not
7130 change the value of CR0.CD and CR0.NW.
7132 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is
7133 available even when configured for x2APIC
7134 mode. When this quirk is disabled, KVM
7135 disables the MMIO LAPIC interface if the
7136 LAPIC is in x2APIC mode.
7138 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before
7139 exiting to userspace for an OUT instruction
7140 to port 0x7e. When this quirk is disabled,
7141 KVM does not pre-increment %rip before
7142 exiting to userspace.
7144 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets
7145 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if
7146 IA32_MISC_ENABLE[bit 18] (MWAIT) is set.
7147 Additionally, when this quirk is disabled,
7148 KVM clears CPUID.01H:ECX[bit 3] if
7149 IA32_MISC_ENABLE[bit 18] is cleared.
7150 =================================== ============================================
7152 8. Other capabilities.
7153 ======================
7155 This section lists capabilities that give information about other
7156 features of the KVM implementation.
7158 8.1 KVM_CAP_PPC_HWRNG
7159 ---------------------
7163 This capability, if KVM_CHECK_EXTENSION indicates that it is
7164 available, means that the kernel has an implementation of the
7165 H_RANDOM hypercall backed by a hardware random-number generator.
7166 If present, the kernel H_RANDOM handler can be enabled for guest use
7167 with the KVM_CAP_PPC_ENABLE_HCALL capability.
7169 8.2 KVM_CAP_HYPERV_SYNIC
7170 ------------------------
7174 This capability, if KVM_CHECK_EXTENSION indicates that it is
7175 available, means that the kernel has an implementation of the
7176 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
7177 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
7179 In order to use SynIC, it has to be activated by setting this
7180 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
7181 will disable the use of APIC hardware virtualization even if supported
7182 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
7184 8.3 KVM_CAP_PPC_RADIX_MMU
7185 -------------------------
7189 This capability, if KVM_CHECK_EXTENSION indicates that it is
7190 available, means that the kernel can support guests using the
7191 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
7194 8.4 KVM_CAP_PPC_HASH_MMU_V3
7195 ---------------------------
7199 This capability, if KVM_CHECK_EXTENSION indicates that it is
7200 available, means that the kernel can support guests using the
7201 hashed page table MMU defined in Power ISA V3.00 (as implemented in
7202 the POWER9 processor), including in-memory segment tables.
7207 :Architectures: mips
7209 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7210 it is available, means that full hardware assisted virtualization capabilities
7211 of the hardware are available for use through KVM. An appropriate
7212 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
7215 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7216 available, it means that the VM is using full hardware assisted virtualization
7217 capabilities of the hardware. This is useful to check after creating a VM with
7218 KVM_VM_MIPS_DEFAULT.
7220 The value returned by KVM_CHECK_EXTENSION should be compared against known
7221 values (see below). All other values are reserved. This is to allow for the
7222 possibility of other hardware assisted virtualization implementations which
7223 may be incompatible with the MIPS VZ ASE.
7225 == ==========================================================================
7226 0 The trap & emulate implementation is in use to run guest code in user
7227 mode. Guest virtual memory segments are rearranged to fit the guest in the
7228 user mode address space.
7230 1 The MIPS VZ ASE is in use, providing full hardware assisted
7231 virtualization, including standard guest virtual memory segments.
7232 == ==========================================================================
7237 :Architectures: mips
7239 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7240 it is available, means that the trap & emulate implementation is available to
7241 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
7242 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
7243 to KVM_CREATE_VM to create a VM which utilises it.
7245 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7246 available, it means that the VM is using trap & emulate.
7248 8.7 KVM_CAP_MIPS_64BIT
7249 ----------------------
7251 :Architectures: mips
7253 This capability indicates the supported architecture type of the guest, i.e. the
7254 supported register and address width.
7256 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
7257 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
7258 be checked specifically against known values (see below). All other values are
7261 == ========================================================================
7262 0 MIPS32 or microMIPS32.
7263 Both registers and addresses are 32-bits wide.
7264 It will only be possible to run 32-bit guest code.
7266 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
7267 Registers are 64-bits wide, but addresses are 32-bits wide.
7268 64-bit guest code may run but cannot access MIPS64 memory segments.
7269 It will also be possible to run 32-bit guest code.
7271 2 MIPS64 or microMIPS64 with access to all address segments.
7272 Both registers and addresses are 64-bits wide.
7273 It will be possible to run 64-bit or 32-bit guest code.
7274 == ========================================================================
7276 8.9 KVM_CAP_ARM_USER_IRQ
7277 ------------------------
7279 :Architectures: arm64
7281 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
7282 that if userspace creates a VM without an in-kernel interrupt controller, it
7283 will be notified of changes to the output level of in-kernel emulated devices,
7284 which can generate virtual interrupts, presented to the VM.
7285 For such VMs, on every return to userspace, the kernel
7286 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
7287 output level of the device.
7289 Whenever kvm detects a change in the device output level, kvm guarantees at
7290 least one return to userspace before running the VM. This exit could either
7291 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
7292 userspace can always sample the device output level and re-compute the state of
7293 the userspace interrupt controller. Userspace should always check the state
7294 of run->s.regs.device_irq_level on every kvm exit.
7295 The value in run->s.regs.device_irq_level can represent both level and edge
7296 triggered interrupt signals, depending on the device. Edge triggered interrupt
7297 signals will exit to userspace with the bit in run->s.regs.device_irq_level
7298 set exactly once per edge signal.
7300 The field run->s.regs.device_irq_level is available independent of
7301 run->kvm_valid_regs or run->kvm_dirty_regs bits.
7303 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
7304 number larger than 0 indicating the version of this capability is implemented
7305 and thereby which bits in run->s.regs.device_irq_level can signal values.
7307 Currently the following bits are defined for the device_irq_level bitmap::
7309 KVM_CAP_ARM_USER_IRQ >= 1:
7311 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
7312 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
7313 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
7315 Future versions of kvm may implement additional events. These will get
7316 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
7319 8.10 KVM_CAP_PPC_SMT_POSSIBLE
7320 -----------------------------
7324 Querying this capability returns a bitmap indicating the possible
7325 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
7326 (counting from the right) is set, then a virtual SMT mode of 2^N is
7329 8.11 KVM_CAP_HYPERV_SYNIC2
7330 --------------------------
7334 This capability enables a newer version of Hyper-V Synthetic interrupt
7335 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
7336 doesn't clear SynIC message and event flags pages when they are enabled by
7337 writing to the respective MSRs.
7339 8.12 KVM_CAP_HYPERV_VP_INDEX
7340 ----------------------------
7344 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
7345 value is used to denote the target vcpu for a SynIC interrupt. For
7346 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
7347 capability is absent, userspace can still query this msr's value.
7349 8.13 KVM_CAP_S390_AIS_MIGRATION
7350 -------------------------------
7352 :Architectures: s390
7355 This capability indicates if the flic device will be able to get/set the
7356 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
7357 to discover this without having to create a flic device.
7359 8.14 KVM_CAP_S390_PSW
7360 ---------------------
7362 :Architectures: s390
7364 This capability indicates that the PSW is exposed via the kvm_run structure.
7366 8.15 KVM_CAP_S390_GMAP
7367 ----------------------
7369 :Architectures: s390
7371 This capability indicates that the user space memory used as guest mapping can
7372 be anywhere in the user memory address space, as long as the memory slots are
7373 aligned and sized to a segment (1MB) boundary.
7375 8.16 KVM_CAP_S390_COW
7376 ---------------------
7378 :Architectures: s390
7380 This capability indicates that the user space memory used as guest mapping can
7381 use copy-on-write semantics as well as dirty pages tracking via read-only page
7384 8.17 KVM_CAP_S390_BPB
7385 ---------------------
7387 :Architectures: s390
7389 This capability indicates that kvm will implement the interfaces to handle
7390 reset, migration and nested KVM for branch prediction blocking. The stfle
7391 facility 82 should not be provided to the guest without this capability.
7393 8.18 KVM_CAP_HYPERV_TLBFLUSH
7394 ----------------------------
7398 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
7400 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
7401 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
7403 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
7404 ----------------------------------
7406 :Architectures: arm64
7408 This capability indicates that userspace can specify (via the
7409 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
7410 takes a virtual SError interrupt exception.
7411 If KVM advertises this capability, userspace can only specify the ISS field for
7412 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
7413 CPU when the exception is taken. If this virtual SError is taken to EL1 using
7414 AArch64, this value will be reported in the ISS field of ESR_ELx.
7416 See KVM_CAP_VCPU_EVENTS for more details.
7418 8.20 KVM_CAP_HYPERV_SEND_IPI
7419 ----------------------------
7423 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
7425 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
7427 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
7428 -----------------------------------
7432 This capability indicates that KVM running on top of Hyper-V hypervisor
7433 enables Direct TLB flush for its guests meaning that TLB flush
7434 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
7435 Due to the different ABI for hypercall parameters between Hyper-V and
7436 KVM, enabling this capability effectively disables all hypercall
7437 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
7438 flush hypercalls by Hyper-V) so userspace should disable KVM identification
7439 in CPUID and only exposes Hyper-V identification. In this case, guest
7440 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
7442 8.22 KVM_CAP_S390_VCPU_RESETS
7443 -----------------------------
7445 :Architectures: s390
7447 This capability indicates that the KVM_S390_NORMAL_RESET and
7448 KVM_S390_CLEAR_RESET ioctls are available.
7450 8.23 KVM_CAP_S390_PROTECTED
7451 ---------------------------
7453 :Architectures: s390
7455 This capability indicates that the Ultravisor has been initialized and
7456 KVM can therefore start protected VMs.
7457 This capability governs the KVM_S390_PV_COMMAND ioctl and the
7458 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
7459 guests when the state change is invalid.
7461 8.24 KVM_CAP_STEAL_TIME
7462 -----------------------
7464 :Architectures: arm64, x86
7466 This capability indicates that KVM supports steal time accounting.
7467 When steal time accounting is supported it may be enabled with
7468 architecture-specific interfaces. This capability and the architecture-
7469 specific interfaces must be consistent, i.e. if one says the feature
7470 is supported, than the other should as well and vice versa. For arm64
7471 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
7472 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
7474 8.25 KVM_CAP_S390_DIAG318
7475 -------------------------
7477 :Architectures: s390
7479 This capability enables a guest to set information about its control program
7480 (i.e. guest kernel type and version). The information is helpful during
7481 system/firmware service events, providing additional data about the guest
7482 environments running on the machine.
7484 The information is associated with the DIAGNOSE 0x318 instruction, which sets
7485 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
7486 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
7487 environment the control program is running in (e.g. Linux, z/VM...), and the
7488 CPVC is used for information specific to OS (e.g. Linux version, Linux
7491 If this capability is available, then the CPNC and CPVC can be synchronized
7492 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
7494 8.26 KVM_CAP_X86_USER_SPACE_MSR
7495 -------------------------------
7499 This capability indicates that KVM supports deflection of MSR reads and
7500 writes to user space. It can be enabled on a VM level. If enabled, MSR
7501 accesses that would usually trigger a #GP by KVM into the guest will
7502 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
7503 KVM_EXIT_X86_WRMSR exit notifications.
7505 8.27 KVM_CAP_X86_MSR_FILTER
7506 ---------------------------
7510 This capability indicates that KVM supports that accesses to user defined MSRs
7511 may be rejected. With this capability exposed, KVM exports new VM ioctl
7512 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
7513 ranges that KVM should reject access to.
7515 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
7516 trap and emulate MSRs that are outside of the scope of KVM as well as
7517 limit the attack surface on KVM's MSR emulation code.
7519 8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
7520 -------------------------------------
7524 When enabled, KVM will disable paravirtual features provided to the
7525 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
7526 (0x40000001). Otherwise, a guest may use the paravirtual features
7527 regardless of what has actually been exposed through the CPUID leaf.
7529 8.29 KVM_CAP_DIRTY_LOG_RING
7530 ---------------------------
7533 :Parameters: args[0] - size of the dirty log ring
7535 KVM is capable of tracking dirty memory using ring buffers that are
7536 mmaped into userspace; there is one dirty ring per vcpu.
7538 The dirty ring is available to userspace as an array of
7539 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
7541 struct kvm_dirty_gfn {
7543 __u32 slot; /* as_id | slot_id */
7547 The following values are defined for the flags field to define the
7548 current state of the entry::
7550 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
7551 #define KVM_DIRTY_GFN_F_RESET BIT(1)
7552 #define KVM_DIRTY_GFN_F_MASK 0x3
7554 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
7555 ioctl to enable this capability for the new guest and set the size of
7556 the rings. Enabling the capability is only allowed before creating any
7557 vCPU, and the size of the ring must be a power of two. The larger the
7558 ring buffer, the less likely the ring is full and the VM is forced to
7559 exit to userspace. The optimal size depends on the workload, but it is
7560 recommended that it be at least 64 KiB (4096 entries).
7562 Just like for dirty page bitmaps, the buffer tracks writes to
7563 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
7564 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
7565 with the flag set, userspace can start harvesting dirty pages from the
7568 An entry in the ring buffer can be unused (flag bits ``00``),
7569 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
7570 state machine for the entry is as follows::
7572 dirtied harvested reset
7573 00 -----------> 01 -------------> 1X -------+
7576 +------------------------------------------+
7578 To harvest the dirty pages, userspace accesses the mmaped ring buffer
7579 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
7580 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
7581 The userspace should harvest this GFN and mark the flags from state
7582 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
7583 to show that this GFN is harvested and waiting for a reset), and move
7584 on to the next GFN. The userspace should continue to do this until the
7585 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
7586 all the dirty GFNs that were available.
7588 It's not necessary for userspace to harvest the all dirty GFNs at once.
7589 However it must collect the dirty GFNs in sequence, i.e., the userspace
7590 program cannot skip one dirty GFN to collect the one next to it.
7592 After processing one or more entries in the ring buffer, userspace
7593 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
7594 it, so that the kernel will reprotect those collected GFNs.
7595 Therefore, the ioctl must be called *before* reading the content of
7598 The dirty ring can get full. When it happens, the KVM_RUN of the
7599 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
7601 The dirty ring interface has a major difference comparing to the
7602 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
7603 userspace, it's still possible that the kernel has not yet flushed the
7604 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
7605 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
7606 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
7607 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
7609 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
7610 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
7611 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
7612 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
7613 machine will switch to ring-buffer dirty page tracking and further
7614 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
7616 8.30 KVM_CAP_XEN_HVM
7617 --------------------
7621 This capability indicates the features that Xen supports for hosting Xen
7622 PVHVM guests. Valid flags are::
7624 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
7625 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
7626 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
7627 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2)
7628 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 3)
7630 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
7631 ioctl is available, for the guest to set its hypercall page.
7633 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
7634 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
7635 contents, to request that KVM generate hypercall page content automatically
7636 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
7638 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
7639 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
7640 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
7641 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
7644 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
7645 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
7646 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
7648 The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries
7649 of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority
7650 field set to indicate 2 level event channel delivery.
7652 8.31 KVM_CAP_PPC_MULTITCE
7653 -------------------------
7655 :Capability: KVM_CAP_PPC_MULTITCE
7659 This capability means the kernel is capable of handling hypercalls
7660 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
7661 space. This significantly accelerates DMA operations for PPC KVM guests.
7662 User space should expect that its handlers for these hypercalls
7663 are not going to be called if user space previously registered LIOBN
7664 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
7666 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
7667 user space might have to advertise it for the guest. For example,
7668 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
7669 present in the "ibm,hypertas-functions" device-tree property.
7671 The hypercalls mentioned above may or may not be processed successfully
7672 in the kernel based fast path. If they can not be handled by the kernel,
7673 they will get passed on to user space. So user space still has to have
7674 an implementation for these despite the in kernel acceleration.
7676 This capability is always enabled.
7678 8.32 KVM_CAP_PTP_KVM
7679 --------------------
7681 :Architectures: arm64
7683 This capability indicates that the KVM virtual PTP service is
7684 supported in the host. A VMM can check whether the service is
7685 available to the guest on migration.
7687 8.33 KVM_CAP_HYPERV_ENFORCE_CPUID
7688 ---------------------------------
7692 When enabled, KVM will disable emulated Hyper-V features provided to the
7693 guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
7694 currently implmented Hyper-V features are provided unconditionally when
7695 Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
7698 8.34 KVM_CAP_EXIT_HYPERCALL
7699 ---------------------------
7701 :Capability: KVM_CAP_EXIT_HYPERCALL
7705 This capability, if enabled, will cause KVM to exit to userspace
7706 with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
7708 Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
7709 of hypercalls that can be configured to exit to userspace.
7710 Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
7712 The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
7713 of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace
7714 the hypercalls whose corresponding bit is in the argument, and return
7715 ENOSYS for the others.
7717 8.35 KVM_CAP_PMU_CAPABILITY
7718 ---------------------------
7720 :Capability KVM_CAP_PMU_CAPABILITY
7723 :Parameters: arg[0] is bitmask of PMU virtualization capabilities.
7724 :Returns 0 on success, -EINVAL when arg[0] contains invalid bits
7726 This capability alters PMU virtualization in KVM.
7728 Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
7729 PMU virtualization capabilities that can be adjusted on a VM.
7731 The argument to KVM_ENABLE_CAP is also a bitmask and selects specific
7732 PMU virtualization capabilities to be applied to the VM. This can
7733 only be invoked on a VM prior to the creation of VCPUs.
7735 At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting
7736 this capability will disable PMU virtualization for that VM. Usermode
7737 should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
7739 9. Known KVM API problems
7740 =========================
7742 In some cases, KVM's API has some inconsistencies or common pitfalls
7743 that userspace need to be aware of. This section details some of
7746 Most of them are architecture specific, so the section is split by
7752 ``KVM_GET_SUPPORTED_CPUID`` issues
7753 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7755 In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible
7756 to take its result and pass it directly to ``KVM_SET_CPUID2``. This section
7757 documents some cases in which that requires some care.
7762 CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``,
7763 but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or
7764 ``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of
7767 The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature.
7769 CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``.
7770 It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel
7771 has enabled in-kernel emulation of the local APIC.
7773 Obsolete ioctls and capabilities
7774 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7776 KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually
7777 available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if
7780 Ordering of KVM_GET_*/KVM_SET_* ioctls
7781 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^