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 certain flags are returned from the KVM_CAP_XEN_HVM check, they may
986 be set in the flags field of this ioctl:
988 The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate
989 the contents of the hypercall page automatically; hypercalls will be
990 intercepted and passed to userspace through KVM_EXIT_XEN. In this
991 ase, all of the blob size and address fields must be zero.
993 The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace
994 will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event
995 channel interrupts rather than manipulating the guest's shared_info
996 structures directly. This, in turn, may allow KVM to enable features
997 such as intercepting the SCHEDOP_poll hypercall to accelerate PV
998 spinlock operation for the guest. Userspace may still use the ioctl
999 to deliver events if it was advertised, even if userspace does not
1000 send this indication that it will always do so
1002 No other flags are currently valid in the struct kvm_xen_hvm_config.
1007 :Capability: KVM_CAP_ADJUST_CLOCK
1010 :Parameters: struct kvm_clock_data (out)
1011 :Returns: 0 on success, -1 on error
1013 Gets the current timestamp of kvmclock as seen by the current guest. In
1014 conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
1017 When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
1018 set of bits that KVM can return in struct kvm_clock_data's flag member.
1020 The following flags are defined:
1022 KVM_CLOCK_TSC_STABLE
1023 If set, the returned value is the exact kvmclock
1024 value seen by all VCPUs at the instant when KVM_GET_CLOCK was called.
1025 If clear, the returned value is simply CLOCK_MONOTONIC plus a constant
1026 offset; the offset can be modified with KVM_SET_CLOCK. KVM will try
1027 to make all VCPUs follow this clock, but the exact value read by each
1028 VCPU could differ, because the host TSC is not stable.
1031 If set, the `realtime` field in the kvm_clock_data
1032 structure is populated with the value of the host's real time
1033 clocksource at the instant when KVM_GET_CLOCK was called. If clear,
1034 the `realtime` field does not contain a value.
1037 If set, the `host_tsc` field in the kvm_clock_data
1038 structure is populated with the value of the host's timestamp counter (TSC)
1039 at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field
1040 does not contain a value.
1044 struct kvm_clock_data {
1045 __u64 clock; /* kvmclock current value */
1057 :Capability: KVM_CAP_ADJUST_CLOCK
1060 :Parameters: struct kvm_clock_data (in)
1061 :Returns: 0 on success, -1 on error
1063 Sets the current timestamp of kvmclock to the value specified in its parameter.
1064 In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1067 The following flags can be passed:
1070 If set, KVM will compare the value of the `realtime` field
1071 with the value of the host's real time clocksource at the instant when
1072 KVM_SET_CLOCK was called. The difference in elapsed time is added to the final
1073 kvmclock value that will be provided to guests.
1075 Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored.
1079 struct kvm_clock_data {
1080 __u64 clock; /* kvmclock current value */
1089 4.31 KVM_GET_VCPU_EVENTS
1090 ------------------------
1092 :Capability: KVM_CAP_VCPU_EVENTS
1093 :Extended by: KVM_CAP_INTR_SHADOW
1094 :Architectures: x86, arm64
1096 :Parameters: struct kvm_vcpu_event (out)
1097 :Returns: 0 on success, -1 on error
1102 Gets currently pending exceptions, interrupts, and NMIs as well as related
1107 struct kvm_vcpu_events {
1111 __u8 has_error_code;
1132 __u8 smm_inside_nmi;
1136 __u8 exception_has_payload;
1137 __u64 exception_payload;
1140 The following bits are defined in the flags field:
1142 - KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1143 interrupt.shadow contains a valid state.
1145 - KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1148 - KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1149 exception_has_payload, exception_payload, and exception.pending
1150 fields contain a valid state. This bit will be set whenever
1151 KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1156 If the guest accesses a device that is being emulated by the host kernel in
1157 such a way that a real device would generate a physical SError, KVM may make
1158 a virtual SError pending for that VCPU. This system error interrupt remains
1159 pending until the guest takes the exception by unmasking PSTATE.A.
1161 Running the VCPU may cause it to take a pending SError, or make an access that
1162 causes an SError to become pending. The event's description is only valid while
1163 the VPCU is not running.
1165 This API provides a way to read and write the pending 'event' state that is not
1166 visible to the guest. To save, restore or migrate a VCPU the struct representing
1167 the state can be read then written using this GET/SET API, along with the other
1168 guest-visible registers. It is not possible to 'cancel' an SError that has been
1171 A device being emulated in user-space may also wish to generate an SError. To do
1172 this the events structure can be populated by user-space. The current state
1173 should be read first, to ensure no existing SError is pending. If an existing
1174 SError is pending, the architecture's 'Multiple SError interrupts' rules should
1175 be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1176 Serviceability (RAS) Specification").
1178 SError exceptions always have an ESR value. Some CPUs have the ability to
1179 specify what the virtual SError's ESR value should be. These systems will
1180 advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1181 always have a non-zero value when read, and the agent making an SError pending
1182 should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1183 the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1184 with exception.has_esr as zero, KVM will choose an ESR.
1186 Specifying exception.has_esr on a system that does not support it will return
1187 -EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1188 will return -EINVAL.
1190 It is not possible to read back a pending external abort (injected via
1191 KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1192 directly to the virtual CPU).
1196 struct kvm_vcpu_events {
1198 __u8 serror_pending;
1199 __u8 serror_has_esr;
1200 __u8 ext_dabt_pending;
1201 /* Align it to 8 bytes */
1208 4.32 KVM_SET_VCPU_EVENTS
1209 ------------------------
1211 :Capability: KVM_CAP_VCPU_EVENTS
1212 :Extended by: KVM_CAP_INTR_SHADOW
1213 :Architectures: x86, arm64
1215 :Parameters: struct kvm_vcpu_event (in)
1216 :Returns: 0 on success, -1 on error
1221 Set pending exceptions, interrupts, and NMIs as well as related states of the
1224 See KVM_GET_VCPU_EVENTS for the data structure.
1226 Fields that may be modified asynchronously by running VCPUs can be excluded
1227 from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1228 smi.pending. Keep the corresponding bits in the flags field cleared to
1229 suppress overwriting the current in-kernel state. The bits are:
1231 =============================== ==================================
1232 KVM_VCPUEVENT_VALID_NMI_PENDING transfer nmi.pending to the kernel
1233 KVM_VCPUEVENT_VALID_SIPI_VECTOR transfer sipi_vector
1234 KVM_VCPUEVENT_VALID_SMM transfer the smi sub-struct.
1235 =============================== ==================================
1237 If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1238 the flags field to signal that interrupt.shadow contains a valid state and
1239 shall be written into the VCPU.
1241 KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1243 If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1244 can be set in the flags field to signal that the
1245 exception_has_payload, exception_payload, and exception.pending fields
1246 contain a valid state and shall be written into the VCPU.
1251 User space may need to inject several types of events to the guest.
1253 Set the pending SError exception state for this VCPU. It is not possible to
1254 'cancel' an Serror that has been made pending.
1256 If the guest performed an access to I/O memory which could not be handled by
1257 userspace, for example because of missing instruction syndrome decode
1258 information or because there is no device mapped at the accessed IPA, then
1259 userspace can ask the kernel to inject an external abort using the address
1260 from the exiting fault on the VCPU. It is a programming error to set
1261 ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1262 KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1263 KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1264 how userspace reports accesses for the above cases to guests, across different
1265 userspace implementations. Nevertheless, userspace can still emulate all Arm
1266 exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1268 See KVM_GET_VCPU_EVENTS for the data structure.
1271 4.33 KVM_GET_DEBUGREGS
1272 ----------------------
1274 :Capability: KVM_CAP_DEBUGREGS
1277 :Parameters: struct kvm_debugregs (out)
1278 :Returns: 0 on success, -1 on error
1280 Reads debug registers from the vcpu.
1284 struct kvm_debugregs {
1293 4.34 KVM_SET_DEBUGREGS
1294 ----------------------
1296 :Capability: KVM_CAP_DEBUGREGS
1299 :Parameters: struct kvm_debugregs (in)
1300 :Returns: 0 on success, -1 on error
1302 Writes debug registers into the vcpu.
1304 See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1305 yet and must be cleared on entry.
1308 4.35 KVM_SET_USER_MEMORY_REGION
1309 -------------------------------
1311 :Capability: KVM_CAP_USER_MEMORY
1314 :Parameters: struct kvm_userspace_memory_region (in)
1315 :Returns: 0 on success, -1 on error
1319 struct kvm_userspace_memory_region {
1322 __u64 guest_phys_addr;
1323 __u64 memory_size; /* bytes */
1324 __u64 userspace_addr; /* start of the userspace allocated memory */
1327 /* for kvm_memory_region::flags */
1328 #define KVM_MEM_LOG_DIRTY_PAGES (1UL << 0)
1329 #define KVM_MEM_READONLY (1UL << 1)
1331 This ioctl allows the user to create, modify or delete a guest physical
1332 memory slot. Bits 0-15 of "slot" specify the slot id and this value
1333 should be less than the maximum number of user memory slots supported per
1334 VM. The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1335 Slots may not overlap in guest physical address space.
1337 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1338 specifies the address space which is being modified. They must be
1339 less than the value that KVM_CHECK_EXTENSION returns for the
1340 KVM_CAP_MULTI_ADDRESS_SPACE capability. Slots in separate address spaces
1341 are unrelated; the restriction on overlapping slots only applies within
1344 Deleting a slot is done by passing zero for memory_size. When changing
1345 an existing slot, it may be moved in the guest physical memory space,
1346 or its flags may be modified, but it may not be resized.
1348 Memory for the region is taken starting at the address denoted by the
1349 field userspace_addr, which must point at user addressable memory for
1350 the entire memory slot size. Any object may back this memory, including
1351 anonymous memory, ordinary files, and hugetlbfs.
1353 On architectures that support a form of address tagging, userspace_addr must
1354 be an untagged address.
1356 It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1357 be identical. This allows large pages in the guest to be backed by large
1360 The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1361 KVM_MEM_READONLY. The former can be set to instruct KVM to keep track of
1362 writes to memory within the slot. See KVM_GET_DIRTY_LOG ioctl to know how to
1363 use it. The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1364 to make a new slot read-only. In this case, writes to this memory will be
1365 posted to userspace as KVM_EXIT_MMIO exits.
1367 When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1368 the memory region are automatically reflected into the guest. For example, an
1369 mmap() that affects the region will be made visible immediately. Another
1370 example is madvise(MADV_DROP).
1372 It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1373 The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1374 allocation and is deprecated.
1377 4.36 KVM_SET_TSS_ADDR
1378 ---------------------
1380 :Capability: KVM_CAP_SET_TSS_ADDR
1383 :Parameters: unsigned long tss_address (in)
1384 :Returns: 0 on success, -1 on error
1386 This ioctl defines the physical address of a three-page region in the guest
1387 physical address space. The region must be within the first 4GB of the
1388 guest physical address space and must not conflict with any memory slot
1389 or any mmio address. The guest may malfunction if it accesses this memory
1392 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1393 because of a quirk in the virtualization implementation (see the internals
1394 documentation when it pops into existence).
1400 :Capability: KVM_CAP_ENABLE_CAP
1401 :Architectures: mips, ppc, s390, x86
1403 :Parameters: struct kvm_enable_cap (in)
1404 :Returns: 0 on success; -1 on error
1406 :Capability: KVM_CAP_ENABLE_CAP_VM
1409 :Parameters: struct kvm_enable_cap (in)
1410 :Returns: 0 on success; -1 on error
1414 Not all extensions are enabled by default. Using this ioctl the application
1415 can enable an extension, making it available to the guest.
1417 On systems that do not support this ioctl, it always fails. On systems that
1418 do support it, it only works for extensions that are supported for enablement.
1420 To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1425 struct kvm_enable_cap {
1429 The capability that is supposed to get enabled.
1435 A bitfield indicating future enhancements. Has to be 0 for now.
1441 Arguments for enabling a feature. If a feature needs initial values to
1442 function properly, this is the place to put them.
1449 The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1450 for vm-wide capabilities.
1452 4.38 KVM_GET_MP_STATE
1453 ---------------------
1455 :Capability: KVM_CAP_MP_STATE
1456 :Architectures: x86, s390, arm64, riscv
1458 :Parameters: struct kvm_mp_state (out)
1459 :Returns: 0 on success; -1 on error
1463 struct kvm_mp_state {
1467 Returns the vcpu's current "multiprocessing state" (though also valid on
1468 uniprocessor guests).
1470 Possible values are:
1472 ========================== ===============================================
1473 KVM_MP_STATE_RUNNABLE the vcpu is currently running
1475 KVM_MP_STATE_UNINITIALIZED the vcpu is an application processor (AP)
1476 which has not yet received an INIT signal [x86]
1477 KVM_MP_STATE_INIT_RECEIVED the vcpu has received an INIT signal, and is
1478 now ready for a SIPI [x86]
1479 KVM_MP_STATE_HALTED the vcpu has executed a HLT instruction and
1480 is waiting for an interrupt [x86]
1481 KVM_MP_STATE_SIPI_RECEIVED the vcpu has just received a SIPI (vector
1482 accessible via KVM_GET_VCPU_EVENTS) [x86]
1483 KVM_MP_STATE_STOPPED the vcpu is stopped [s390,arm64,riscv]
1484 KVM_MP_STATE_CHECK_STOP the vcpu is in a special error state [s390]
1485 KVM_MP_STATE_OPERATING the vcpu is operating (running or halted)
1487 KVM_MP_STATE_LOAD the vcpu is in a special load/startup state
1489 KVM_MP_STATE_SUSPENDED the vcpu is in a suspend state and is waiting
1490 for a wakeup event [arm64]
1491 ========================== ===============================================
1493 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1494 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1495 these architectures.
1500 If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the
1501 architectural execution of a WFI instruction.
1503 If a wakeup event is recognized, KVM will exit to userspace with a
1504 KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If
1505 userspace wants to honor the wakeup, it must set the vCPU's MP state to
1506 KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup
1507 event in subsequent calls to KVM_RUN.
1511 If userspace intends to keep the vCPU in a SUSPENDED state, it is
1512 strongly recommended that userspace take action to suppress the
1513 wakeup event (such as masking an interrupt). Otherwise, subsequent
1514 calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP
1515 event and inadvertently waste CPU cycles.
1517 Additionally, if userspace takes action to suppress a wakeup event,
1518 it is strongly recommended that it also restores the vCPU to its
1519 original state when the vCPU is made RUNNABLE again. For example,
1520 if userspace masked a pending interrupt to suppress the wakeup,
1521 the interrupt should be unmasked before returning control to the
1527 The only states that are valid are KVM_MP_STATE_STOPPED and
1528 KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1530 4.39 KVM_SET_MP_STATE
1531 ---------------------
1533 :Capability: KVM_CAP_MP_STATE
1534 :Architectures: x86, s390, arm64, riscv
1536 :Parameters: struct kvm_mp_state (in)
1537 :Returns: 0 on success; -1 on error
1539 Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1542 On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1543 in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1544 these architectures.
1549 The only states that are valid are KVM_MP_STATE_STOPPED and
1550 KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1552 4.40 KVM_SET_IDENTITY_MAP_ADDR
1553 ------------------------------
1555 :Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1558 :Parameters: unsigned long identity (in)
1559 :Returns: 0 on success, -1 on error
1561 This ioctl defines the physical address of a one-page region in the guest
1562 physical address space. The region must be within the first 4GB of the
1563 guest physical address space and must not conflict with any memory slot
1564 or any mmio address. The guest may malfunction if it accesses this memory
1567 Setting the address to 0 will result in resetting the address to its default
1570 This ioctl is required on Intel-based hosts. This is needed on Intel hardware
1571 because of a quirk in the virtualization implementation (see the internals
1572 documentation when it pops into existence).
1574 Fails if any VCPU has already been created.
1576 4.41 KVM_SET_BOOT_CPU_ID
1577 ------------------------
1579 :Capability: KVM_CAP_SET_BOOT_CPU_ID
1582 :Parameters: unsigned long vcpu_id
1583 :Returns: 0 on success, -1 on error
1585 Define which vcpu is the Bootstrap Processor (BSP). Values are the same
1586 as the vcpu id in KVM_CREATE_VCPU. If this ioctl is not called, the default
1587 is vcpu 0. This ioctl has to be called before vcpu creation,
1588 otherwise it will return EBUSY error.
1594 :Capability: KVM_CAP_XSAVE
1597 :Parameters: struct kvm_xsave (out)
1598 :Returns: 0 on success, -1 on error
1608 This ioctl would copy current vcpu's xsave struct to the userspace.
1614 :Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2
1617 :Parameters: struct kvm_xsave (in)
1618 :Returns: 0 on success, -1 on error
1628 This ioctl would copy userspace's xsave struct to the kernel. It copies
1629 as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2),
1630 when invoked on the vm file descriptor. The size value returned by
1631 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
1632 Currently, it is only greater than 4096 if a dynamic feature has been
1633 enabled with ``arch_prctl()``, but this may change in the future.
1635 The offsets of the state save areas in struct kvm_xsave follow the
1636 contents of CPUID leaf 0xD on the host.
1642 :Capability: KVM_CAP_XCRS
1645 :Parameters: struct kvm_xcrs (out)
1646 :Returns: 0 on success, -1 on error
1659 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1663 This ioctl would copy current vcpu's xcrs to the userspace.
1669 :Capability: KVM_CAP_XCRS
1672 :Parameters: struct kvm_xcrs (in)
1673 :Returns: 0 on success, -1 on error
1686 struct kvm_xcr xcrs[KVM_MAX_XCRS];
1690 This ioctl would set vcpu's xcr to the value userspace specified.
1693 4.46 KVM_GET_SUPPORTED_CPUID
1694 ----------------------------
1696 :Capability: KVM_CAP_EXT_CPUID
1699 :Parameters: struct kvm_cpuid2 (in/out)
1700 :Returns: 0 on success, -1 on error
1707 struct kvm_cpuid_entry2 entries[0];
1710 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
1711 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
1712 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
1714 struct kvm_cpuid_entry2 {
1725 This ioctl returns x86 cpuid features which are supported by both the
1726 hardware and kvm in its default configuration. Userspace can use the
1727 information returned by this ioctl to construct cpuid information (for
1728 KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1729 userspace capabilities, and with user requirements (for example, the
1730 user may wish to constrain cpuid to emulate older hardware, or for
1731 feature consistency across a cluster).
1733 Dynamically-enabled feature bits need to be requested with
1734 ``arch_prctl()`` before calling this ioctl. Feature bits that have not
1735 been requested are excluded from the result.
1737 Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1738 expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1739 its default configuration. If userspace enables such capabilities, it
1740 is responsible for modifying the results of this ioctl appropriately.
1742 Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1743 with the 'nent' field indicating the number of entries in the variable-size
1744 array 'entries'. If the number of entries is too low to describe the cpu
1745 capabilities, an error (E2BIG) is returned. If the number is too high,
1746 the 'nent' field is adjusted and an error (ENOMEM) is returned. If the
1747 number is just right, the 'nent' field is adjusted to the number of valid
1748 entries in the 'entries' array, which is then filled.
1750 The entries returned are the host cpuid as returned by the cpuid instruction,
1751 with unknown or unsupported features masked out. Some features (for example,
1752 x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1753 emulate them efficiently. The fields in each entry are defined as follows:
1756 the eax value used to obtain the entry
1759 the ecx value used to obtain the entry (for entries that are
1763 an OR of zero or more of the following:
1765 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1766 if the index field is valid
1769 the values returned by the cpuid instruction for
1770 this function/index combination
1772 The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1773 as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1774 support. Instead it is reported via::
1776 ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1778 if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1779 feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1782 4.47 KVM_PPC_GET_PVINFO
1783 -----------------------
1785 :Capability: KVM_CAP_PPC_GET_PVINFO
1788 :Parameters: struct kvm_ppc_pvinfo (out)
1789 :Returns: 0 on success, !0 on error
1793 struct kvm_ppc_pvinfo {
1799 This ioctl fetches PV specific information that need to be passed to the guest
1800 using the device tree or other means from vm context.
1802 The hcall array defines 4 instructions that make up a hypercall.
1804 If any additional field gets added to this structure later on, a bit for that
1805 additional piece of information will be set in the flags bitmap.
1807 The flags bitmap is defined as::
1809 /* the host supports the ePAPR idle hcall
1810 #define KVM_PPC_PVINFO_FLAGS_EV_IDLE (1<<0)
1812 4.52 KVM_SET_GSI_ROUTING
1813 ------------------------
1815 :Capability: KVM_CAP_IRQ_ROUTING
1816 :Architectures: x86 s390 arm64
1818 :Parameters: struct kvm_irq_routing (in)
1819 :Returns: 0 on success, -1 on error
1821 Sets the GSI routing table entries, overwriting any previously set entries.
1823 On arm64, GSI routing has the following limitation:
1825 - GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1829 struct kvm_irq_routing {
1832 struct kvm_irq_routing_entry entries[0];
1835 No flags are specified so far, the corresponding field must be set to zero.
1839 struct kvm_irq_routing_entry {
1845 struct kvm_irq_routing_irqchip irqchip;
1846 struct kvm_irq_routing_msi msi;
1847 struct kvm_irq_routing_s390_adapter adapter;
1848 struct kvm_irq_routing_hv_sint hv_sint;
1849 struct kvm_irq_routing_xen_evtchn xen_evtchn;
1854 /* gsi routing entry types */
1855 #define KVM_IRQ_ROUTING_IRQCHIP 1
1856 #define KVM_IRQ_ROUTING_MSI 2
1857 #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1858 #define KVM_IRQ_ROUTING_HV_SINT 4
1859 #define KVM_IRQ_ROUTING_XEN_EVTCHN 5
1863 - KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1864 type, specifies that the devid field contains a valid value. The per-VM
1865 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1866 the device ID. If this capability is not available, userspace should
1867 never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1872 struct kvm_irq_routing_irqchip {
1877 struct kvm_irq_routing_msi {
1887 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1888 for the device that wrote the MSI message. For PCI, this is usually a
1889 BFD identifier in the lower 16 bits.
1891 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1892 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
1893 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
1894 address_hi must be zero.
1898 struct kvm_irq_routing_s390_adapter {
1902 __u32 summary_offset;
1906 struct kvm_irq_routing_hv_sint {
1911 struct kvm_irq_routing_xen_evtchn {
1918 When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit
1919 in its indication of supported features, routing to Xen event channels
1920 is supported. Although the priority field is present, only the value
1921 KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by
1922 2 level event channels. FIFO event channel support may be added in
1926 4.55 KVM_SET_TSC_KHZ
1927 --------------------
1929 :Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL
1931 :Type: vcpu ioctl / vm ioctl
1932 :Parameters: virtual tsc_khz
1933 :Returns: 0 on success, -1 on error
1935 Specifies the tsc frequency for the virtual machine. The unit of the
1938 If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also
1939 be used as a vm ioctl to set the initial tsc frequency of subsequently
1942 4.56 KVM_GET_TSC_KHZ
1943 --------------------
1945 :Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL
1947 :Type: vcpu ioctl / vm ioctl
1949 :Returns: virtual tsc-khz on success, negative value on error
1951 Returns the tsc frequency of the guest. The unit of the return value is
1952 KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1959 :Capability: KVM_CAP_IRQCHIP
1962 :Parameters: struct kvm_lapic_state (out)
1963 :Returns: 0 on success, -1 on error
1967 #define KVM_APIC_REG_SIZE 0x400
1968 struct kvm_lapic_state {
1969 char regs[KVM_APIC_REG_SIZE];
1972 Reads the Local APIC registers and copies them into the input argument. The
1973 data format and layout are the same as documented in the architecture manual.
1975 If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1976 enabled, then the format of APIC_ID register depends on the APIC mode
1977 (reported by MSR_IA32_APICBASE) of its VCPU. x2APIC stores APIC ID in
1978 the APIC_ID register (bytes 32-35). xAPIC only allows an 8-bit APIC ID
1979 which is stored in bits 31-24 of the APIC register, or equivalently in
1980 byte 35 of struct kvm_lapic_state's regs field. KVM_GET_LAPIC must then
1981 be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1983 If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1984 always uses xAPIC format.
1990 :Capability: KVM_CAP_IRQCHIP
1993 :Parameters: struct kvm_lapic_state (in)
1994 :Returns: 0 on success, -1 on error
1998 #define KVM_APIC_REG_SIZE 0x400
1999 struct kvm_lapic_state {
2000 char regs[KVM_APIC_REG_SIZE];
2003 Copies the input argument into the Local APIC registers. The data format
2004 and layout are the same as documented in the architecture manual.
2006 The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
2007 regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
2008 See the note in KVM_GET_LAPIC.
2014 :Capability: KVM_CAP_IOEVENTFD
2017 :Parameters: struct kvm_ioeventfd (in)
2018 :Returns: 0 on success, !0 on error
2020 This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
2021 within the guest. A guest write in the registered address will signal the
2022 provided event instead of triggering an exit.
2026 struct kvm_ioeventfd {
2028 __u64 addr; /* legal pio/mmio address */
2029 __u32 len; /* 0, 1, 2, 4, or 8 bytes */
2035 For the special case of virtio-ccw devices on s390, the ioevent is matched
2036 to a subchannel/virtqueue tuple instead.
2038 The following flags are defined::
2040 #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
2041 #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio)
2042 #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign)
2043 #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2044 (1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2046 If datamatch flag is set, the event will be signaled only if the written value
2047 to the registered address is equal to datamatch in struct kvm_ioeventfd.
2049 For virtio-ccw devices, addr contains the subchannel id and datamatch the
2052 With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2053 the kernel will ignore the length of guest write and may get a faster vmexit.
2054 The speedup may only apply to specific architectures, but the ioeventfd will
2060 :Capability: KVM_CAP_SW_TLB
2063 :Parameters: struct kvm_dirty_tlb (in)
2064 :Returns: 0 on success, -1 on error
2068 struct kvm_dirty_tlb {
2073 This must be called whenever userspace has changed an entry in the shared
2074 TLB, prior to calling KVM_RUN on the associated vcpu.
2076 The "bitmap" field is the userspace address of an array. This array
2077 consists of a number of bits, equal to the total number of TLB entries as
2078 determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
2079 nearest multiple of 64.
2081 Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2084 The array is little-endian: the bit 0 is the least significant bit of the
2085 first byte, bit 8 is the least significant bit of the second byte, etc.
2086 This avoids any complications with differing word sizes.
2088 The "num_dirty" field is a performance hint for KVM to determine whether it
2089 should skip processing the bitmap and just invalidate everything. It must
2090 be set to the number of set bits in the bitmap.
2093 4.62 KVM_CREATE_SPAPR_TCE
2094 -------------------------
2096 :Capability: KVM_CAP_SPAPR_TCE
2097 :Architectures: powerpc
2099 :Parameters: struct kvm_create_spapr_tce (in)
2100 :Returns: file descriptor for manipulating the created TCE table
2102 This creates a virtual TCE (translation control entry) table, which
2103 is an IOMMU for PAPR-style virtual I/O. It is used to translate
2104 logical addresses used in virtual I/O into guest physical addresses,
2105 and provides a scatter/gather capability for PAPR virtual I/O.
2109 /* for KVM_CAP_SPAPR_TCE */
2110 struct kvm_create_spapr_tce {
2115 The liobn field gives the logical IO bus number for which to create a
2116 TCE table. The window_size field specifies the size of the DMA window
2117 which this TCE table will translate - the table will contain one 64
2118 bit TCE entry for every 4kiB of the DMA window.
2120 When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2121 table has been created using this ioctl(), the kernel will handle it
2122 in real mode, updating the TCE table. H_PUT_TCE calls for other
2123 liobns will cause a vm exit and must be handled by userspace.
2125 The return value is a file descriptor which can be passed to mmap(2)
2126 to map the created TCE table into userspace. This lets userspace read
2127 the entries written by kernel-handled H_PUT_TCE calls, and also lets
2128 userspace update the TCE table directly which is useful in some
2132 4.63 KVM_ALLOCATE_RMA
2133 ---------------------
2135 :Capability: KVM_CAP_PPC_RMA
2136 :Architectures: powerpc
2138 :Parameters: struct kvm_allocate_rma (out)
2139 :Returns: file descriptor for mapping the allocated RMA
2141 This allocates a Real Mode Area (RMA) from the pool allocated at boot
2142 time by the kernel. An RMA is a physically-contiguous, aligned region
2143 of memory used on older POWER processors to provide the memory which
2144 will be accessed by real-mode (MMU off) accesses in a KVM guest.
2145 POWER processors support a set of sizes for the RMA that usually
2146 includes 64MB, 128MB, 256MB and some larger powers of two.
2150 /* for KVM_ALLOCATE_RMA */
2151 struct kvm_allocate_rma {
2155 The return value is a file descriptor which can be passed to mmap(2)
2156 to map the allocated RMA into userspace. The mapped area can then be
2157 passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2158 RMA for a virtual machine. The size of the RMA in bytes (which is
2159 fixed at host kernel boot time) is returned in the rma_size field of
2160 the argument structure.
2162 The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2163 is supported; 2 if the processor requires all virtual machines to have
2164 an RMA, or 1 if the processor can use an RMA but doesn't require it,
2165 because it supports the Virtual RMA (VRMA) facility.
2171 :Capability: KVM_CAP_USER_NMI
2175 :Returns: 0 on success, -1 on error
2177 Queues an NMI on the thread's vcpu. Note this is well defined only
2178 when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2179 between the virtual cpu core and virtual local APIC. After KVM_CREATE_IRQCHIP
2180 has been called, this interface is completely emulated within the kernel.
2182 To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2183 following algorithm:
2186 - read the local APIC's state (KVM_GET_LAPIC)
2187 - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2188 - if so, issue KVM_NMI
2191 Some guests configure the LINT1 NMI input to cause a panic, aiding in
2195 4.65 KVM_S390_UCAS_MAP
2196 ----------------------
2198 :Capability: KVM_CAP_S390_UCONTROL
2199 :Architectures: s390
2201 :Parameters: struct kvm_s390_ucas_mapping (in)
2202 :Returns: 0 in case of success
2204 The parameter is defined like this::
2206 struct kvm_s390_ucas_mapping {
2212 This ioctl maps the memory at "user_addr" with the length "length" to
2213 the vcpu's address space starting at "vcpu_addr". All parameters need to
2214 be aligned by 1 megabyte.
2217 4.66 KVM_S390_UCAS_UNMAP
2218 ------------------------
2220 :Capability: KVM_CAP_S390_UCONTROL
2221 :Architectures: s390
2223 :Parameters: struct kvm_s390_ucas_mapping (in)
2224 :Returns: 0 in case of success
2226 The parameter is defined like this::
2228 struct kvm_s390_ucas_mapping {
2234 This ioctl unmaps the memory in the vcpu's address space starting at
2235 "vcpu_addr" with the length "length". The field "user_addr" is ignored.
2236 All parameters need to be aligned by 1 megabyte.
2239 4.67 KVM_S390_VCPU_FAULT
2240 ------------------------
2242 :Capability: KVM_CAP_S390_UCONTROL
2243 :Architectures: s390
2245 :Parameters: vcpu absolute address (in)
2246 :Returns: 0 in case of success
2248 This call creates a page table entry on the virtual cpu's address space
2249 (for user controlled virtual machines) or the virtual machine's address
2250 space (for regular virtual machines). This only works for minor faults,
2251 thus it's recommended to access subject memory page via the user page
2252 table upfront. This is useful to handle validity intercepts for user
2253 controlled virtual machines to fault in the virtual cpu's lowcore pages
2254 prior to calling the KVM_RUN ioctl.
2257 4.68 KVM_SET_ONE_REG
2258 --------------------
2260 :Capability: KVM_CAP_ONE_REG
2263 :Parameters: struct kvm_one_reg (in)
2264 :Returns: 0 on success, negative value on failure
2268 ====== ============================================================
2269 ENOENT no such register
2270 EINVAL invalid register ID, or no such register or used with VMs in
2271 protected virtualization mode on s390
2272 EPERM (arm64) register access not allowed before vcpu finalization
2273 ====== ============================================================
2275 (These error codes are indicative only: do not rely on a specific error
2276 code being returned in a specific situation.)
2280 struct kvm_one_reg {
2285 Using this ioctl, a single vcpu register can be set to a specific value
2286 defined by user space with the passed in struct kvm_one_reg, where id
2287 refers to the register identifier as described below and addr is a pointer
2288 to a variable with the respective size. There can be architecture agnostic
2289 and architecture specific registers. Each have their own range of operation
2290 and their own constants and width. To keep track of the implemented
2291 registers, find a list below:
2293 ======= =============================== ============
2294 Arch Register Width (bits)
2295 ======= =============================== ============
2296 PPC KVM_REG_PPC_HIOR 64
2297 PPC KVM_REG_PPC_IAC1 64
2298 PPC KVM_REG_PPC_IAC2 64
2299 PPC KVM_REG_PPC_IAC3 64
2300 PPC KVM_REG_PPC_IAC4 64
2301 PPC KVM_REG_PPC_DAC1 64
2302 PPC KVM_REG_PPC_DAC2 64
2303 PPC KVM_REG_PPC_DABR 64
2304 PPC KVM_REG_PPC_DSCR 64
2305 PPC KVM_REG_PPC_PURR 64
2306 PPC KVM_REG_PPC_SPURR 64
2307 PPC KVM_REG_PPC_DAR 64
2308 PPC KVM_REG_PPC_DSISR 32
2309 PPC KVM_REG_PPC_AMR 64
2310 PPC KVM_REG_PPC_UAMOR 64
2311 PPC KVM_REG_PPC_MMCR0 64
2312 PPC KVM_REG_PPC_MMCR1 64
2313 PPC KVM_REG_PPC_MMCRA 64
2314 PPC KVM_REG_PPC_MMCR2 64
2315 PPC KVM_REG_PPC_MMCRS 64
2316 PPC KVM_REG_PPC_MMCR3 64
2317 PPC KVM_REG_PPC_SIAR 64
2318 PPC KVM_REG_PPC_SDAR 64
2319 PPC KVM_REG_PPC_SIER 64
2320 PPC KVM_REG_PPC_SIER2 64
2321 PPC KVM_REG_PPC_SIER3 64
2322 PPC KVM_REG_PPC_PMC1 32
2323 PPC KVM_REG_PPC_PMC2 32
2324 PPC KVM_REG_PPC_PMC3 32
2325 PPC KVM_REG_PPC_PMC4 32
2326 PPC KVM_REG_PPC_PMC5 32
2327 PPC KVM_REG_PPC_PMC6 32
2328 PPC KVM_REG_PPC_PMC7 32
2329 PPC KVM_REG_PPC_PMC8 32
2330 PPC KVM_REG_PPC_FPR0 64
2332 PPC KVM_REG_PPC_FPR31 64
2333 PPC KVM_REG_PPC_VR0 128
2335 PPC KVM_REG_PPC_VR31 128
2336 PPC KVM_REG_PPC_VSR0 128
2338 PPC KVM_REG_PPC_VSR31 128
2339 PPC KVM_REG_PPC_FPSCR 64
2340 PPC KVM_REG_PPC_VSCR 32
2341 PPC KVM_REG_PPC_VPA_ADDR 64
2342 PPC KVM_REG_PPC_VPA_SLB 128
2343 PPC KVM_REG_PPC_VPA_DTL 128
2344 PPC KVM_REG_PPC_EPCR 32
2345 PPC KVM_REG_PPC_EPR 32
2346 PPC KVM_REG_PPC_TCR 32
2347 PPC KVM_REG_PPC_TSR 32
2348 PPC KVM_REG_PPC_OR_TSR 32
2349 PPC KVM_REG_PPC_CLEAR_TSR 32
2350 PPC KVM_REG_PPC_MAS0 32
2351 PPC KVM_REG_PPC_MAS1 32
2352 PPC KVM_REG_PPC_MAS2 64
2353 PPC KVM_REG_PPC_MAS7_3 64
2354 PPC KVM_REG_PPC_MAS4 32
2355 PPC KVM_REG_PPC_MAS6 32
2356 PPC KVM_REG_PPC_MMUCFG 32
2357 PPC KVM_REG_PPC_TLB0CFG 32
2358 PPC KVM_REG_PPC_TLB1CFG 32
2359 PPC KVM_REG_PPC_TLB2CFG 32
2360 PPC KVM_REG_PPC_TLB3CFG 32
2361 PPC KVM_REG_PPC_TLB0PS 32
2362 PPC KVM_REG_PPC_TLB1PS 32
2363 PPC KVM_REG_PPC_TLB2PS 32
2364 PPC KVM_REG_PPC_TLB3PS 32
2365 PPC KVM_REG_PPC_EPTCFG 32
2366 PPC KVM_REG_PPC_ICP_STATE 64
2367 PPC KVM_REG_PPC_VP_STATE 128
2368 PPC KVM_REG_PPC_TB_OFFSET 64
2369 PPC KVM_REG_PPC_SPMC1 32
2370 PPC KVM_REG_PPC_SPMC2 32
2371 PPC KVM_REG_PPC_IAMR 64
2372 PPC KVM_REG_PPC_TFHAR 64
2373 PPC KVM_REG_PPC_TFIAR 64
2374 PPC KVM_REG_PPC_TEXASR 64
2375 PPC KVM_REG_PPC_FSCR 64
2376 PPC KVM_REG_PPC_PSPB 32
2377 PPC KVM_REG_PPC_EBBHR 64
2378 PPC KVM_REG_PPC_EBBRR 64
2379 PPC KVM_REG_PPC_BESCR 64
2380 PPC KVM_REG_PPC_TAR 64
2381 PPC KVM_REG_PPC_DPDES 64
2382 PPC KVM_REG_PPC_DAWR 64
2383 PPC KVM_REG_PPC_DAWRX 64
2384 PPC KVM_REG_PPC_CIABR 64
2385 PPC KVM_REG_PPC_IC 64
2386 PPC KVM_REG_PPC_VTB 64
2387 PPC KVM_REG_PPC_CSIGR 64
2388 PPC KVM_REG_PPC_TACR 64
2389 PPC KVM_REG_PPC_TCSCR 64
2390 PPC KVM_REG_PPC_PID 64
2391 PPC KVM_REG_PPC_ACOP 64
2392 PPC KVM_REG_PPC_VRSAVE 32
2393 PPC KVM_REG_PPC_LPCR 32
2394 PPC KVM_REG_PPC_LPCR_64 64
2395 PPC KVM_REG_PPC_PPR 64
2396 PPC KVM_REG_PPC_ARCH_COMPAT 32
2397 PPC KVM_REG_PPC_DABRX 32
2398 PPC KVM_REG_PPC_WORT 64
2399 PPC KVM_REG_PPC_SPRG9 64
2400 PPC KVM_REG_PPC_DBSR 32
2401 PPC KVM_REG_PPC_TIDR 64
2402 PPC KVM_REG_PPC_PSSCR 64
2403 PPC KVM_REG_PPC_DEC_EXPIRY 64
2404 PPC KVM_REG_PPC_PTCR 64
2405 PPC KVM_REG_PPC_DAWR1 64
2406 PPC KVM_REG_PPC_DAWRX1 64
2407 PPC KVM_REG_PPC_TM_GPR0 64
2409 PPC KVM_REG_PPC_TM_GPR31 64
2410 PPC KVM_REG_PPC_TM_VSR0 128
2412 PPC KVM_REG_PPC_TM_VSR63 128
2413 PPC KVM_REG_PPC_TM_CR 64
2414 PPC KVM_REG_PPC_TM_LR 64
2415 PPC KVM_REG_PPC_TM_CTR 64
2416 PPC KVM_REG_PPC_TM_FPSCR 64
2417 PPC KVM_REG_PPC_TM_AMR 64
2418 PPC KVM_REG_PPC_TM_PPR 64
2419 PPC KVM_REG_PPC_TM_VRSAVE 64
2420 PPC KVM_REG_PPC_TM_VSCR 32
2421 PPC KVM_REG_PPC_TM_DSCR 64
2422 PPC KVM_REG_PPC_TM_TAR 64
2423 PPC KVM_REG_PPC_TM_XER 64
2425 MIPS KVM_REG_MIPS_R0 64
2427 MIPS KVM_REG_MIPS_R31 64
2428 MIPS KVM_REG_MIPS_HI 64
2429 MIPS KVM_REG_MIPS_LO 64
2430 MIPS KVM_REG_MIPS_PC 64
2431 MIPS KVM_REG_MIPS_CP0_INDEX 32
2432 MIPS KVM_REG_MIPS_CP0_ENTRYLO0 64
2433 MIPS KVM_REG_MIPS_CP0_ENTRYLO1 64
2434 MIPS KVM_REG_MIPS_CP0_CONTEXT 64
2435 MIPS KVM_REG_MIPS_CP0_CONTEXTCONFIG 32
2436 MIPS KVM_REG_MIPS_CP0_USERLOCAL 64
2437 MIPS KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2438 MIPS KVM_REG_MIPS_CP0_PAGEMASK 32
2439 MIPS KVM_REG_MIPS_CP0_PAGEGRAIN 32
2440 MIPS KVM_REG_MIPS_CP0_SEGCTL0 64
2441 MIPS KVM_REG_MIPS_CP0_SEGCTL1 64
2442 MIPS KVM_REG_MIPS_CP0_SEGCTL2 64
2443 MIPS KVM_REG_MIPS_CP0_PWBASE 64
2444 MIPS KVM_REG_MIPS_CP0_PWFIELD 64
2445 MIPS KVM_REG_MIPS_CP0_PWSIZE 64
2446 MIPS KVM_REG_MIPS_CP0_WIRED 32
2447 MIPS KVM_REG_MIPS_CP0_PWCTL 32
2448 MIPS KVM_REG_MIPS_CP0_HWRENA 32
2449 MIPS KVM_REG_MIPS_CP0_BADVADDR 64
2450 MIPS KVM_REG_MIPS_CP0_BADINSTR 32
2451 MIPS KVM_REG_MIPS_CP0_BADINSTRP 32
2452 MIPS KVM_REG_MIPS_CP0_COUNT 32
2453 MIPS KVM_REG_MIPS_CP0_ENTRYHI 64
2454 MIPS KVM_REG_MIPS_CP0_COMPARE 32
2455 MIPS KVM_REG_MIPS_CP0_STATUS 32
2456 MIPS KVM_REG_MIPS_CP0_INTCTL 32
2457 MIPS KVM_REG_MIPS_CP0_CAUSE 32
2458 MIPS KVM_REG_MIPS_CP0_EPC 64
2459 MIPS KVM_REG_MIPS_CP0_PRID 32
2460 MIPS KVM_REG_MIPS_CP0_EBASE 64
2461 MIPS KVM_REG_MIPS_CP0_CONFIG 32
2462 MIPS KVM_REG_MIPS_CP0_CONFIG1 32
2463 MIPS KVM_REG_MIPS_CP0_CONFIG2 32
2464 MIPS KVM_REG_MIPS_CP0_CONFIG3 32
2465 MIPS KVM_REG_MIPS_CP0_CONFIG4 32
2466 MIPS KVM_REG_MIPS_CP0_CONFIG5 32
2467 MIPS KVM_REG_MIPS_CP0_CONFIG7 32
2468 MIPS KVM_REG_MIPS_CP0_XCONTEXT 64
2469 MIPS KVM_REG_MIPS_CP0_ERROREPC 64
2470 MIPS KVM_REG_MIPS_CP0_KSCRATCH1 64
2471 MIPS KVM_REG_MIPS_CP0_KSCRATCH2 64
2472 MIPS KVM_REG_MIPS_CP0_KSCRATCH3 64
2473 MIPS KVM_REG_MIPS_CP0_KSCRATCH4 64
2474 MIPS KVM_REG_MIPS_CP0_KSCRATCH5 64
2475 MIPS KVM_REG_MIPS_CP0_KSCRATCH6 64
2476 MIPS KVM_REG_MIPS_CP0_MAAR(0..63) 64
2477 MIPS KVM_REG_MIPS_COUNT_CTL 64
2478 MIPS KVM_REG_MIPS_COUNT_RESUME 64
2479 MIPS KVM_REG_MIPS_COUNT_HZ 64
2480 MIPS KVM_REG_MIPS_FPR_32(0..31) 32
2481 MIPS KVM_REG_MIPS_FPR_64(0..31) 64
2482 MIPS KVM_REG_MIPS_VEC_128(0..31) 128
2483 MIPS KVM_REG_MIPS_FCR_IR 32
2484 MIPS KVM_REG_MIPS_FCR_CSR 32
2485 MIPS KVM_REG_MIPS_MSA_IR 32
2486 MIPS KVM_REG_MIPS_MSA_CSR 32
2487 ======= =============================== ============
2489 ARM registers are mapped using the lower 32 bits. The upper 16 of that
2490 is the register group type, or coprocessor number:
2492 ARM core registers have the following id bit patterns::
2494 0x4020 0000 0010 <index into the kvm_regs struct:16>
2496 ARM 32-bit CP15 registers have the following id bit patterns::
2498 0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2500 ARM 64-bit CP15 registers have the following id bit patterns::
2502 0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2504 ARM CCSIDR registers are demultiplexed by CSSELR value::
2506 0x4020 0000 0011 00 <csselr:8>
2508 ARM 32-bit VFP control registers have the following id bit patterns::
2510 0x4020 0000 0012 1 <regno:12>
2512 ARM 64-bit FP registers have the following id bit patterns::
2514 0x4030 0000 0012 0 <regno:12>
2516 ARM firmware pseudo-registers have the following bit pattern::
2518 0x4030 0000 0014 <regno:16>
2521 arm64 registers are mapped using the lower 32 bits. The upper 16 of
2522 that is the register group type, or coprocessor number:
2524 arm64 core/FP-SIMD registers have the following id bit patterns. Note
2525 that the size of the access is variable, as the kvm_regs structure
2526 contains elements ranging from 32 to 128 bits. The index is a 32bit
2527 value in the kvm_regs structure seen as a 32bit array::
2529 0x60x0 0000 0010 <index into the kvm_regs struct:16>
2533 ======================= ========= ===== =======================================
2534 Encoding Register Bits kvm_regs member
2535 ======================= ========= ===== =======================================
2536 0x6030 0000 0010 0000 X0 64 regs.regs[0]
2537 0x6030 0000 0010 0002 X1 64 regs.regs[1]
2539 0x6030 0000 0010 003c X30 64 regs.regs[30]
2540 0x6030 0000 0010 003e SP 64 regs.sp
2541 0x6030 0000 0010 0040 PC 64 regs.pc
2542 0x6030 0000 0010 0042 PSTATE 64 regs.pstate
2543 0x6030 0000 0010 0044 SP_EL1 64 sp_el1
2544 0x6030 0000 0010 0046 ELR_EL1 64 elr_el1
2545 0x6030 0000 0010 0048 SPSR_EL1 64 spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2546 0x6030 0000 0010 004a SPSR_ABT 64 spsr[KVM_SPSR_ABT]
2547 0x6030 0000 0010 004c SPSR_UND 64 spsr[KVM_SPSR_UND]
2548 0x6030 0000 0010 004e SPSR_IRQ 64 spsr[KVM_SPSR_IRQ]
2549 0x6060 0000 0010 0050 SPSR_FIQ 64 spsr[KVM_SPSR_FIQ]
2550 0x6040 0000 0010 0054 V0 128 fp_regs.vregs[0] [1]_
2551 0x6040 0000 0010 0058 V1 128 fp_regs.vregs[1] [1]_
2553 0x6040 0000 0010 00d0 V31 128 fp_regs.vregs[31] [1]_
2554 0x6020 0000 0010 00d4 FPSR 32 fp_regs.fpsr
2555 0x6020 0000 0010 00d5 FPCR 32 fp_regs.fpcr
2556 ======================= ========= ===== =======================================
2558 .. [1] These encodings are not accepted for SVE-enabled vcpus. See
2561 The equivalent register content can be accessed via bits [127:0] of
2562 the corresponding SVE Zn registers instead for vcpus that have SVE
2563 enabled (see below).
2565 arm64 CCSIDR registers are demultiplexed by CSSELR value::
2567 0x6020 0000 0011 00 <csselr:8>
2569 arm64 system registers have the following id bit patterns::
2571 0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2575 Two system register IDs do not follow the specified pattern. These
2576 are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2577 system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively. These
2578 two had their values accidentally swapped, which means TIMER_CVAL is
2579 derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2580 derived from the register encoding for CNTV_CVAL_EL0. As this is
2581 API, it must remain this way.
2583 arm64 firmware pseudo-registers have the following bit pattern::
2585 0x6030 0000 0014 <regno:16>
2587 arm64 SVE registers have the following bit patterns::
2589 0x6080 0000 0015 00 <n:5> <slice:5> Zn bits[2048*slice + 2047 : 2048*slice]
2590 0x6050 0000 0015 04 <n:4> <slice:5> Pn bits[256*slice + 255 : 256*slice]
2591 0x6050 0000 0015 060 <slice:5> FFR bits[256*slice + 255 : 256*slice]
2592 0x6060 0000 0015 ffff KVM_REG_ARM64_SVE_VLS pseudo-register
2594 Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2595 ENOENT. max_vq is the vcpu's maximum supported vector length in 128-bit
2596 quadwords: see [2]_ below.
2598 These registers are only accessible on vcpus for which SVE is enabled.
2599 See KVM_ARM_VCPU_INIT for details.
2601 In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2602 accessible until the vcpu's SVE configuration has been finalized
2603 using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE). See KVM_ARM_VCPU_INIT
2604 and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2606 KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2607 lengths supported by the vcpu to be discovered and configured by
2608 userspace. When transferred to or from user memory via KVM_GET_ONE_REG
2609 or KVM_SET_ONE_REG, the value of this register is of type
2610 __u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2613 __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2615 if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2616 ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2617 ((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2618 /* Vector length vq * 16 bytes supported */
2620 /* Vector length vq * 16 bytes not supported */
2622 .. [2] The maximum value vq for which the above condition is true is
2623 max_vq. This is the maximum vector length available to the guest on
2624 this vcpu, and determines which register slices are visible through
2625 this ioctl interface.
2627 (See Documentation/arm64/sve.rst for an explanation of the "vq"
2630 KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2631 KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2634 Userspace may subsequently modify it if desired until the vcpu's SVE
2635 configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2637 Apart from simply removing all vector lengths from the host set that
2638 exceed some value, support for arbitrarily chosen sets of vector lengths
2639 is hardware-dependent and may not be available. Attempting to configure
2640 an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2643 After the vcpu's SVE configuration is finalized, further attempts to
2644 write this register will fail with EPERM.
2646 arm64 bitmap feature firmware pseudo-registers have the following bit pattern::
2648 0x6030 0000 0016 <regno:16>
2650 The bitmap feature firmware registers exposes the hypercall services that
2651 are available for userspace to configure. The set bits corresponds to the
2652 services that are available for the guests to access. By default, KVM
2653 sets all the supported bits during VM initialization. The userspace can
2654 discover the available services via KVM_GET_ONE_REG, and write back the
2655 bitmap corresponding to the features that it wishes guests to see via
2658 Note: These registers are immutable once any of the vCPUs of the VM has
2659 run at least once. A KVM_SET_ONE_REG in such a scenario will return
2660 a -EBUSY to userspace.
2662 (See Documentation/virt/kvm/arm/hypercalls.rst for more details.)
2665 MIPS registers are mapped using the lower 32 bits. The upper 16 of that is
2666 the register group type:
2668 MIPS core registers (see above) have the following id bit patterns::
2670 0x7030 0000 0000 <reg:16>
2672 MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2673 patterns depending on whether they're 32-bit or 64-bit registers::
2675 0x7020 0000 0001 00 <reg:5> <sel:3> (32-bit)
2676 0x7030 0000 0001 00 <reg:5> <sel:3> (64-bit)
2678 Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2679 versions of the EntryLo registers regardless of the word size of the host
2680 hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2681 with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2682 the PFNX field starting at bit 30.
2684 MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2687 0x7030 0000 0001 01 <reg:8>
2689 MIPS KVM control registers (see above) have the following id bit patterns::
2691 0x7030 0000 0002 <reg:16>
2693 MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2694 id bit patterns depending on the size of the register being accessed. They are
2695 always accessed according to the current guest FPU mode (Status.FR and
2696 Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2697 if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2698 registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2699 overlap the FPU registers::
2701 0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2702 0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2703 0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2705 MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2706 following id bit patterns::
2708 0x7020 0000 0003 01 <0:3> <reg:5>
2710 MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2711 following id bit patterns::
2713 0x7020 0000 0003 02 <0:3> <reg:5>
2715 RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of
2716 that is the register group type.
2718 RISC-V config registers are meant for configuring a Guest VCPU and it has
2719 the following id bit patterns::
2721 0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host)
2722 0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host)
2724 Following are the RISC-V config registers:
2726 ======================= ========= =============================================
2727 Encoding Register Description
2728 ======================= ========= =============================================
2729 0x80x0 0000 0100 0000 isa ISA feature bitmap of Guest VCPU
2730 ======================= ========= =============================================
2732 The isa config register can be read anytime but can only be written before
2733 a Guest VCPU runs. It will have ISA feature bits matching underlying host
2736 RISC-V core registers represent the general excution state of a Guest VCPU
2737 and it has the following id bit patterns::
2739 0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host)
2740 0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host)
2742 Following are the RISC-V core registers:
2744 ======================= ========= =============================================
2745 Encoding Register Description
2746 ======================= ========= =============================================
2747 0x80x0 0000 0200 0000 regs.pc Program counter
2748 0x80x0 0000 0200 0001 regs.ra Return address
2749 0x80x0 0000 0200 0002 regs.sp Stack pointer
2750 0x80x0 0000 0200 0003 regs.gp Global pointer
2751 0x80x0 0000 0200 0004 regs.tp Task pointer
2752 0x80x0 0000 0200 0005 regs.t0 Caller saved register 0
2753 0x80x0 0000 0200 0006 regs.t1 Caller saved register 1
2754 0x80x0 0000 0200 0007 regs.t2 Caller saved register 2
2755 0x80x0 0000 0200 0008 regs.s0 Callee saved register 0
2756 0x80x0 0000 0200 0009 regs.s1 Callee saved register 1
2757 0x80x0 0000 0200 000a regs.a0 Function argument (or return value) 0
2758 0x80x0 0000 0200 000b regs.a1 Function argument (or return value) 1
2759 0x80x0 0000 0200 000c regs.a2 Function argument 2
2760 0x80x0 0000 0200 000d regs.a3 Function argument 3
2761 0x80x0 0000 0200 000e regs.a4 Function argument 4
2762 0x80x0 0000 0200 000f regs.a5 Function argument 5
2763 0x80x0 0000 0200 0010 regs.a6 Function argument 6
2764 0x80x0 0000 0200 0011 regs.a7 Function argument 7
2765 0x80x0 0000 0200 0012 regs.s2 Callee saved register 2
2766 0x80x0 0000 0200 0013 regs.s3 Callee saved register 3
2767 0x80x0 0000 0200 0014 regs.s4 Callee saved register 4
2768 0x80x0 0000 0200 0015 regs.s5 Callee saved register 5
2769 0x80x0 0000 0200 0016 regs.s6 Callee saved register 6
2770 0x80x0 0000 0200 0017 regs.s7 Callee saved register 7
2771 0x80x0 0000 0200 0018 regs.s8 Callee saved register 8
2772 0x80x0 0000 0200 0019 regs.s9 Callee saved register 9
2773 0x80x0 0000 0200 001a regs.s10 Callee saved register 10
2774 0x80x0 0000 0200 001b regs.s11 Callee saved register 11
2775 0x80x0 0000 0200 001c regs.t3 Caller saved register 3
2776 0x80x0 0000 0200 001d regs.t4 Caller saved register 4
2777 0x80x0 0000 0200 001e regs.t5 Caller saved register 5
2778 0x80x0 0000 0200 001f regs.t6 Caller saved register 6
2779 0x80x0 0000 0200 0020 mode Privilege mode (1 = S-mode or 0 = U-mode)
2780 ======================= ========= =============================================
2782 RISC-V csr registers represent the supervisor mode control/status registers
2783 of a Guest VCPU and it has the following id bit patterns::
2785 0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host)
2786 0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host)
2788 Following are the RISC-V csr registers:
2790 ======================= ========= =============================================
2791 Encoding Register Description
2792 ======================= ========= =============================================
2793 0x80x0 0000 0300 0000 sstatus Supervisor status
2794 0x80x0 0000 0300 0001 sie Supervisor interrupt enable
2795 0x80x0 0000 0300 0002 stvec Supervisor trap vector base
2796 0x80x0 0000 0300 0003 sscratch Supervisor scratch register
2797 0x80x0 0000 0300 0004 sepc Supervisor exception program counter
2798 0x80x0 0000 0300 0005 scause Supervisor trap cause
2799 0x80x0 0000 0300 0006 stval Supervisor bad address or instruction
2800 0x80x0 0000 0300 0007 sip Supervisor interrupt pending
2801 0x80x0 0000 0300 0008 satp Supervisor address translation and protection
2802 ======================= ========= =============================================
2804 RISC-V timer registers represent the timer state of a Guest VCPU and it has
2805 the following id bit patterns::
2807 0x8030 0000 04 <index into the kvm_riscv_timer struct:24>
2809 Following are the RISC-V timer registers:
2811 ======================= ========= =============================================
2812 Encoding Register Description
2813 ======================= ========= =============================================
2814 0x8030 0000 0400 0000 frequency Time base frequency (read-only)
2815 0x8030 0000 0400 0001 time Time value visible to Guest
2816 0x8030 0000 0400 0002 compare Time compare programmed by Guest
2817 0x8030 0000 0400 0003 state Time compare state (1 = ON or 0 = OFF)
2818 ======================= ========= =============================================
2820 RISC-V F-extension registers represent the single precision floating point
2821 state of a Guest VCPU and it has the following id bit patterns::
2823 0x8020 0000 05 <index into the __riscv_f_ext_state struct:24>
2825 Following are the RISC-V F-extension registers:
2827 ======================= ========= =============================================
2828 Encoding Register Description
2829 ======================= ========= =============================================
2830 0x8020 0000 0500 0000 f[0] Floating point register 0
2832 0x8020 0000 0500 001f f[31] Floating point register 31
2833 0x8020 0000 0500 0020 fcsr Floating point control and status register
2834 ======================= ========= =============================================
2836 RISC-V D-extension registers represent the double precision floating point
2837 state of a Guest VCPU and it has the following id bit patterns::
2839 0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr)
2840 0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr)
2842 Following are the RISC-V D-extension registers:
2844 ======================= ========= =============================================
2845 Encoding Register Description
2846 ======================= ========= =============================================
2847 0x8030 0000 0600 0000 f[0] Floating point register 0
2849 0x8030 0000 0600 001f f[31] Floating point register 31
2850 0x8020 0000 0600 0020 fcsr Floating point control and status register
2851 ======================= ========= =============================================
2854 4.69 KVM_GET_ONE_REG
2855 --------------------
2857 :Capability: KVM_CAP_ONE_REG
2860 :Parameters: struct kvm_one_reg (in and out)
2861 :Returns: 0 on success, negative value on failure
2865 ======== ============================================================
2866 ENOENT no such register
2867 EINVAL invalid register ID, or no such register or used with VMs in
2868 protected virtualization mode on s390
2869 EPERM (arm64) register access not allowed before vcpu finalization
2870 ======== ============================================================
2872 (These error codes are indicative only: do not rely on a specific error
2873 code being returned in a specific situation.)
2875 This ioctl allows to receive the value of a single register implemented
2876 in a vcpu. The register to read is indicated by the "id" field of the
2877 kvm_one_reg struct passed in. On success, the register value can be found
2878 at the memory location pointed to by "addr".
2880 The list of registers accessible using this interface is identical to the
2884 4.70 KVM_KVMCLOCK_CTRL
2885 ----------------------
2887 :Capability: KVM_CAP_KVMCLOCK_CTRL
2888 :Architectures: Any that implement pvclocks (currently x86 only)
2891 :Returns: 0 on success, -1 on error
2893 This ioctl sets a flag accessible to the guest indicating that the specified
2894 vCPU has been paused by the host userspace.
2896 The host will set a flag in the pvclock structure that is checked from the
2897 soft lockup watchdog. The flag is part of the pvclock structure that is
2898 shared between guest and host, specifically the second bit of the flags
2899 field of the pvclock_vcpu_time_info structure. It will be set exclusively by
2900 the host and read/cleared exclusively by the guest. The guest operation of
2901 checking and clearing the flag must be an atomic operation so
2902 load-link/store-conditional, or equivalent must be used. There are two cases
2903 where the guest will clear the flag: when the soft lockup watchdog timer resets
2904 itself or when a soft lockup is detected. This ioctl can be called any time
2905 after pausing the vcpu, but before it is resumed.
2911 :Capability: KVM_CAP_SIGNAL_MSI
2912 :Architectures: x86 arm64
2914 :Parameters: struct kvm_msi (in)
2915 :Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2917 Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2932 KVM_MSI_VALID_DEVID: devid contains a valid value. The per-VM
2933 KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2934 the device ID. If this capability is not available, userspace
2935 should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2937 If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2938 for the device that wrote the MSI message. For PCI, this is usually a
2939 BFD identifier in the lower 16 bits.
2941 On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2942 feature of KVM_CAP_X2APIC_API capability is enabled. If it is enabled,
2943 address_hi bits 31-8 provide bits 31-8 of the destination id. Bits 7-0 of
2944 address_hi must be zero.
2947 4.71 KVM_CREATE_PIT2
2948 --------------------
2950 :Capability: KVM_CAP_PIT2
2953 :Parameters: struct kvm_pit_config (in)
2954 :Returns: 0 on success, -1 on error
2956 Creates an in-kernel device model for the i8254 PIT. This call is only valid
2957 after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2958 parameters have to be passed::
2960 struct kvm_pit_config {
2967 #define KVM_PIT_SPEAKER_DUMMY 1 /* emulate speaker port stub */
2969 PIT timer interrupts may use a per-VM kernel thread for injection. If it
2970 exists, this thread will have a name of the following pattern::
2972 kvm-pit/<owner-process-pid>
2974 When running a guest with elevated priorities, the scheduling parameters of
2975 this thread may have to be adjusted accordingly.
2977 This IOCTL replaces the obsolete KVM_CREATE_PIT.
2983 :Capability: KVM_CAP_PIT_STATE2
2986 :Parameters: struct kvm_pit_state2 (out)
2987 :Returns: 0 on success, -1 on error
2989 Retrieves the state of the in-kernel PIT model. Only valid after
2990 KVM_CREATE_PIT2. The state is returned in the following structure::
2992 struct kvm_pit_state2 {
2993 struct kvm_pit_channel_state channels[3];
3000 /* disable PIT in HPET legacy mode */
3001 #define KVM_PIT_FLAGS_HPET_LEGACY 0x00000001
3003 This IOCTL replaces the obsolete KVM_GET_PIT.
3009 :Capability: KVM_CAP_PIT_STATE2
3012 :Parameters: struct kvm_pit_state2 (in)
3013 :Returns: 0 on success, -1 on error
3015 Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
3016 See KVM_GET_PIT2 for details on struct kvm_pit_state2.
3018 This IOCTL replaces the obsolete KVM_SET_PIT.
3021 4.74 KVM_PPC_GET_SMMU_INFO
3022 --------------------------
3024 :Capability: KVM_CAP_PPC_GET_SMMU_INFO
3025 :Architectures: powerpc
3028 :Returns: 0 on success, -1 on error
3030 This populates and returns a structure describing the features of
3031 the "Server" class MMU emulation supported by KVM.
3032 This can in turn be used by userspace to generate the appropriate
3033 device-tree properties for the guest operating system.
3035 The structure contains some global information, followed by an
3036 array of supported segment page sizes::
3038 struct kvm_ppc_smmu_info {
3042 struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
3045 The supported flags are:
3047 - KVM_PPC_PAGE_SIZES_REAL:
3048 When that flag is set, guest page sizes must "fit" the backing
3049 store page sizes. When not set, any page size in the list can
3050 be used regardless of how they are backed by userspace.
3052 - KVM_PPC_1T_SEGMENTS
3053 The emulated MMU supports 1T segments in addition to the
3057 This flag indicates that HPT guests are not supported by KVM,
3058 thus all guests must use radix MMU mode.
3060 The "slb_size" field indicates how many SLB entries are supported
3062 The "sps" array contains 8 entries indicating the supported base
3063 page sizes for a segment in increasing order. Each entry is defined
3066 struct kvm_ppc_one_seg_page_size {
3067 __u32 page_shift; /* Base page shift of segment (or 0) */
3068 __u32 slb_enc; /* SLB encoding for BookS */
3069 struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3072 An entry with a "page_shift" of 0 is unused. Because the array is
3073 organized in increasing order, a lookup can stop when encoutering
3076 The "slb_enc" field provides the encoding to use in the SLB for the
3077 page size. The bits are in positions such as the value can directly
3078 be OR'ed into the "vsid" argument of the slbmte instruction.
3080 The "enc" array is a list which for each of those segment base page
3081 size provides the list of supported actual page sizes (which can be
3082 only larger or equal to the base page size), along with the
3083 corresponding encoding in the hash PTE. Similarly, the array is
3084 8 entries sorted by increasing sizes and an entry with a "0" shift
3085 is an empty entry and a terminator::
3087 struct kvm_ppc_one_page_size {
3088 __u32 page_shift; /* Page shift (or 0) */
3089 __u32 pte_enc; /* Encoding in the HPTE (>>12) */
3092 The "pte_enc" field provides a value that can OR'ed into the hash
3093 PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3094 into the hash PTE second double word).
3099 :Capability: KVM_CAP_IRQFD
3100 :Architectures: x86 s390 arm64
3102 :Parameters: struct kvm_irqfd (in)
3103 :Returns: 0 on success, -1 on error
3105 Allows setting an eventfd to directly trigger a guest interrupt.
3106 kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3107 kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
3108 an event is triggered on the eventfd, an interrupt is injected into
3109 the guest using the specified gsi pin. The irqfd is removed using
3110 the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3113 With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3114 mechanism allowing emulation of level-triggered, irqfd-based
3115 interrupts. When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3116 additional eventfd in the kvm_irqfd.resamplefd field. When operating
3117 in resample mode, posting of an interrupt through kvm_irq.fd asserts
3118 the specified gsi in the irqchip. When the irqchip is resampled, such
3119 as from an EOI, the gsi is de-asserted and the user is notified via
3120 kvm_irqfd.resamplefd. It is the user's responsibility to re-queue
3121 the interrupt if the device making use of it still requires service.
3122 Note that closing the resamplefd is not sufficient to disable the
3123 irqfd. The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3124 and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3126 On arm64, gsi routing being supported, the following can happen:
3128 - in case no routing entry is associated to this gsi, injection fails
3129 - in case the gsi is associated to an irqchip routing entry,
3130 irqchip.pin + 32 corresponds to the injected SPI ID.
3131 - in case the gsi is associated to an MSI routing entry, the MSI
3132 message and device ID are translated into an LPI (support restricted
3133 to GICv3 ITS in-kernel emulation).
3135 4.76 KVM_PPC_ALLOCATE_HTAB
3136 --------------------------
3138 :Capability: KVM_CAP_PPC_ALLOC_HTAB
3139 :Architectures: powerpc
3141 :Parameters: Pointer to u32 containing hash table order (in/out)
3142 :Returns: 0 on success, -1 on error
3144 This requests the host kernel to allocate an MMU hash table for a
3145 guest using the PAPR paravirtualization interface. This only does
3146 anything if the kernel is configured to use the Book 3S HV style of
3147 virtualization. Otherwise the capability doesn't exist and the ioctl
3148 returns an ENOTTY error. The rest of this description assumes Book 3S
3151 There must be no vcpus running when this ioctl is called; if there
3152 are, it will do nothing and return an EBUSY error.
3154 The parameter is a pointer to a 32-bit unsigned integer variable
3155 containing the order (log base 2) of the desired size of the hash
3156 table, which must be between 18 and 46. On successful return from the
3157 ioctl, the value will not be changed by the kernel.
3159 If no hash table has been allocated when any vcpu is asked to run
3160 (with the KVM_RUN ioctl), the host kernel will allocate a
3161 default-sized hash table (16 MB).
3163 If this ioctl is called when a hash table has already been allocated,
3164 with a different order from the existing hash table, the existing hash
3165 table will be freed and a new one allocated. If this is ioctl is
3166 called when a hash table has already been allocated of the same order
3167 as specified, the kernel will clear out the existing hash table (zero
3168 all HPTEs). In either case, if the guest is using the virtualized
3169 real-mode area (VRMA) facility, the kernel will re-create the VMRA
3170 HPTEs on the next KVM_RUN of any vcpu.
3172 4.77 KVM_S390_INTERRUPT
3173 -----------------------
3176 :Architectures: s390
3177 :Type: vm ioctl, vcpu ioctl
3178 :Parameters: struct kvm_s390_interrupt (in)
3179 :Returns: 0 on success, -1 on error
3181 Allows to inject an interrupt to the guest. Interrupts can be floating
3182 (vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3184 Interrupt parameters are passed via kvm_s390_interrupt::
3186 struct kvm_s390_interrupt {
3192 type can be one of the following:
3194 KVM_S390_SIGP_STOP (vcpu)
3195 - sigp stop; optional flags in parm
3196 KVM_S390_PROGRAM_INT (vcpu)
3197 - program check; code in parm
3198 KVM_S390_SIGP_SET_PREFIX (vcpu)
3199 - sigp set prefix; prefix address in parm
3200 KVM_S390_RESTART (vcpu)
3202 KVM_S390_INT_CLOCK_COMP (vcpu)
3203 - clock comparator interrupt
3204 KVM_S390_INT_CPU_TIMER (vcpu)
3205 - CPU timer interrupt
3206 KVM_S390_INT_VIRTIO (vm)
3207 - virtio external interrupt; external interrupt
3208 parameters in parm and parm64
3209 KVM_S390_INT_SERVICE (vm)
3210 - sclp external interrupt; sclp parameter in parm
3211 KVM_S390_INT_EMERGENCY (vcpu)
3212 - sigp emergency; source cpu in parm
3213 KVM_S390_INT_EXTERNAL_CALL (vcpu)
3214 - sigp external call; source cpu in parm
3215 KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3216 - compound value to indicate an
3217 I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3218 I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3219 interruption subclass)
3220 KVM_S390_MCHK (vm, vcpu)
3221 - machine check interrupt; cr 14 bits in parm, machine check interrupt
3222 code in parm64 (note that machine checks needing further payload are not
3223 supported by this ioctl)
3225 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3227 4.78 KVM_PPC_GET_HTAB_FD
3228 ------------------------
3230 :Capability: KVM_CAP_PPC_HTAB_FD
3231 :Architectures: powerpc
3233 :Parameters: Pointer to struct kvm_get_htab_fd (in)
3234 :Returns: file descriptor number (>= 0) on success, -1 on error
3236 This returns a file descriptor that can be used either to read out the
3237 entries in the guest's hashed page table (HPT), or to write entries to
3238 initialize the HPT. The returned fd can only be written to if the
3239 KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3240 can only be read if that bit is clear. The argument struct looks like
3243 /* For KVM_PPC_GET_HTAB_FD */
3244 struct kvm_get_htab_fd {
3250 /* Values for kvm_get_htab_fd.flags */
3251 #define KVM_GET_HTAB_BOLTED_ONLY ((__u64)0x1)
3252 #define KVM_GET_HTAB_WRITE ((__u64)0x2)
3254 The 'start_index' field gives the index in the HPT of the entry at
3255 which to start reading. It is ignored when writing.
3257 Reads on the fd will initially supply information about all
3258 "interesting" HPT entries. Interesting entries are those with the
3259 bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3260 all entries. When the end of the HPT is reached, the read() will
3261 return. If read() is called again on the fd, it will start again from
3262 the beginning of the HPT, but will only return HPT entries that have
3263 changed since they were last read.
3265 Data read or written is structured as a header (8 bytes) followed by a
3266 series of valid HPT entries (16 bytes) each. The header indicates how
3267 many valid HPT entries there are and how many invalid entries follow
3268 the valid entries. The invalid entries are not represented explicitly
3269 in the stream. The header format is::
3271 struct kvm_get_htab_header {
3277 Writes to the fd create HPT entries starting at the index given in the
3278 header; first 'n_valid' valid entries with contents from the data
3279 written, then 'n_invalid' invalid entries, invalidating any previously
3280 valid entries found.
3282 4.79 KVM_CREATE_DEVICE
3283 ----------------------
3285 :Capability: KVM_CAP_DEVICE_CTRL
3287 :Parameters: struct kvm_create_device (in/out)
3288 :Returns: 0 on success, -1 on error
3292 ====== =======================================================
3293 ENODEV The device type is unknown or unsupported
3294 EEXIST Device already created, and this type of device may not
3295 be instantiated multiple times
3296 ====== =======================================================
3298 Other error conditions may be defined by individual device types or
3299 have their standard meanings.
3301 Creates an emulated device in the kernel. The file descriptor returned
3302 in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3304 If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3305 device type is supported (not necessarily whether it can be created
3308 Individual devices should not define flags. Attributes should be used
3309 for specifying any behavior that is not implied by the device type
3314 struct kvm_create_device {
3315 __u32 type; /* in: KVM_DEV_TYPE_xxx */
3316 __u32 fd; /* out: device handle */
3317 __u32 flags; /* in: KVM_CREATE_DEVICE_xxx */
3320 4.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3321 --------------------------------------------
3323 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3324 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3325 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set)
3326 :Type: device ioctl, vm ioctl, vcpu ioctl
3327 :Parameters: struct kvm_device_attr
3328 :Returns: 0 on success, -1 on error
3332 ===== =============================================================
3333 ENXIO The group or attribute is unknown/unsupported for this device
3334 or hardware support is missing.
3335 EPERM The attribute cannot (currently) be accessed this way
3336 (e.g. read-only attribute, or attribute that only makes
3337 sense when the device is in a different state)
3338 ===== =============================================================
3340 Other error conditions may be defined by individual device types.
3342 Gets/sets a specified piece of device configuration and/or state. The
3343 semantics are device-specific. See individual device documentation in
3344 the "devices" directory. As with ONE_REG, the size of the data
3345 transferred is defined by the particular attribute.
3349 struct kvm_device_attr {
3350 __u32 flags; /* no flags currently defined */
3351 __u32 group; /* device-defined */
3352 __u64 attr; /* group-defined */
3353 __u64 addr; /* userspace address of attr data */
3356 4.81 KVM_HAS_DEVICE_ATTR
3357 ------------------------
3359 :Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3360 KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3361 KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device
3362 :Type: device ioctl, vm ioctl, vcpu ioctl
3363 :Parameters: struct kvm_device_attr
3364 :Returns: 0 on success, -1 on error
3368 ===== =============================================================
3369 ENXIO The group or attribute is unknown/unsupported for this device
3370 or hardware support is missing.
3371 ===== =============================================================
3373 Tests whether a device supports a particular attribute. A successful
3374 return indicates the attribute is implemented. It does not necessarily
3375 indicate that the attribute can be read or written in the device's
3376 current state. "addr" is ignored.
3378 4.82 KVM_ARM_VCPU_INIT
3379 ----------------------
3382 :Architectures: arm64
3384 :Parameters: struct kvm_vcpu_init (in)
3385 :Returns: 0 on success; -1 on error
3389 ====== =================================================================
3390 EINVAL the target is unknown, or the combination of features is invalid.
3391 ENOENT a features bit specified is unknown.
3392 ====== =================================================================
3394 This tells KVM what type of CPU to present to the guest, and what
3395 optional features it should have. This will cause a reset of the cpu
3396 registers to their initial values. If this is not called, KVM_RUN will
3397 return ENOEXEC for that vcpu.
3399 The initial values are defined as:
3401 * AArch64: EL1h, D, A, I and F bits set. All other bits
3403 * AArch32: SVC, A, I and F bits set. All other bits are
3405 - General Purpose registers, including PC and SP: set to 0
3406 - FPSIMD/NEON registers: set to 0
3407 - SVE registers: set to 0
3408 - System registers: Reset to their architecturally defined
3409 values as for a warm reset to EL1 (resp. SVC)
3411 Note that because some registers reflect machine topology, all vcpus
3412 should be created before this ioctl is invoked.
3414 Userspace can call this function multiple times for a given vcpu, including
3415 after the vcpu has been run. This will reset the vcpu to its initial
3416 state. All calls to this function after the initial call must use the same
3417 target and same set of feature flags, otherwise EINVAL will be returned.
3421 - KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3422 Depends on KVM_CAP_ARM_PSCI. If not set, the CPU will be powered on
3423 and execute guest code when KVM_RUN is called.
3424 - KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3425 Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3426 - KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3427 backward compatible with v0.2) for the CPU.
3428 Depends on KVM_CAP_ARM_PSCI_0_2.
3429 - KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3430 Depends on KVM_CAP_ARM_PMU_V3.
3432 - KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3434 Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3435 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3436 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3437 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3440 - KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3442 Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3443 If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3444 both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3445 KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3448 - KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3449 Depends on KVM_CAP_ARM_SVE.
3450 Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3452 * After KVM_ARM_VCPU_INIT:
3454 - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3455 initial value of this pseudo-register indicates the best set of
3456 vector lengths possible for a vcpu on this host.
3458 * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3460 - KVM_RUN and KVM_GET_REG_LIST are not available;
3462 - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3463 the scalable archietctural SVE registers
3464 KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3465 KVM_REG_ARM64_SVE_FFR;
3467 - KVM_REG_ARM64_SVE_VLS may optionally be written using
3468 KVM_SET_ONE_REG, to modify the set of vector lengths available
3471 * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3473 - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3474 no longer be written using KVM_SET_ONE_REG.
3476 4.83 KVM_ARM_PREFERRED_TARGET
3477 -----------------------------
3480 :Architectures: arm64
3482 :Parameters: struct kvm_vcpu_init (out)
3483 :Returns: 0 on success; -1 on error
3487 ====== ==========================================
3488 ENODEV no preferred target available for the host
3489 ====== ==========================================
3491 This queries KVM for preferred CPU target type which can be emulated
3492 by KVM on underlying host.
3494 The ioctl returns struct kvm_vcpu_init instance containing information
3495 about preferred CPU target type and recommended features for it. The
3496 kvm_vcpu_init->features bitmap returned will have feature bits set if
3497 the preferred target recommends setting these features, but this is
3500 The information returned by this ioctl can be used to prepare an instance
3501 of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3502 VCPU matching underlying host.
3505 4.84 KVM_GET_REG_LIST
3506 ---------------------
3509 :Architectures: arm64, mips
3511 :Parameters: struct kvm_reg_list (in/out)
3512 :Returns: 0 on success; -1 on error
3516 ===== ==============================================================
3517 E2BIG the reg index list is too big to fit in the array specified by
3518 the user (the number required will be written into n).
3519 ===== ==============================================================
3523 struct kvm_reg_list {
3524 __u64 n; /* number of registers in reg[] */
3528 This ioctl returns the guest registers that are supported for the
3529 KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3532 4.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3533 -----------------------------------------
3535 :Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3536 :Architectures: arm64
3538 :Parameters: struct kvm_arm_device_address (in)
3539 :Returns: 0 on success, -1 on error
3543 ====== ============================================
3544 ENODEV The device id is unknown
3545 ENXIO Device not supported on current system
3546 EEXIST Address already set
3547 E2BIG Address outside guest physical address space
3548 EBUSY Address overlaps with other device range
3549 ====== ============================================
3553 struct kvm_arm_device_addr {
3558 Specify a device address in the guest's physical address space where guests
3559 can access emulated or directly exposed devices, which the host kernel needs
3560 to know about. The id field is an architecture specific identifier for a
3563 arm64 divides the id field into two parts, a device id and an
3564 address type id specific to the individual device::
3566 bits: | 63 ... 32 | 31 ... 16 | 15 ... 0 |
3567 field: | 0x00000000 | device id | addr type id |
3569 arm64 currently only require this when using the in-kernel GIC
3570 support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3571 as the device id. When setting the base address for the guest's
3572 mapping of the VGIC virtual CPU and distributor interface, the ioctl
3573 must be called after calling KVM_CREATE_IRQCHIP, but before calling
3574 KVM_RUN on any of the VCPUs. Calling this ioctl twice for any of the
3575 base addresses will return -EEXIST.
3577 Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3578 should be used instead.
3581 4.86 KVM_PPC_RTAS_DEFINE_TOKEN
3582 ------------------------------
3584 :Capability: KVM_CAP_PPC_RTAS
3587 :Parameters: struct kvm_rtas_token_args
3588 :Returns: 0 on success, -1 on error
3590 Defines a token value for a RTAS (Run Time Abstraction Services)
3591 service in order to allow it to be handled in the kernel. The
3592 argument struct gives the name of the service, which must be the name
3593 of a service that has a kernel-side implementation. If the token
3594 value is non-zero, it will be associated with that service, and
3595 subsequent RTAS calls by the guest specifying that token will be
3596 handled by the kernel. If the token value is 0, then any token
3597 associated with the service will be forgotten, and subsequent RTAS
3598 calls by the guest for that service will be passed to userspace to be
3601 4.87 KVM_SET_GUEST_DEBUG
3602 ------------------------
3604 :Capability: KVM_CAP_SET_GUEST_DEBUG
3605 :Architectures: x86, s390, ppc, arm64
3607 :Parameters: struct kvm_guest_debug (in)
3608 :Returns: 0 on success; -1 on error
3612 struct kvm_guest_debug {
3615 struct kvm_guest_debug_arch arch;
3618 Set up the processor specific debug registers and configure vcpu for
3619 handling guest debug events. There are two parts to the structure, the
3620 first a control bitfield indicates the type of debug events to handle
3621 when running. Common control bits are:
3623 - KVM_GUESTDBG_ENABLE: guest debugging is enabled
3624 - KVM_GUESTDBG_SINGLESTEP: the next run should single-step
3626 The top 16 bits of the control field are architecture specific control
3627 flags which can include the following:
3629 - KVM_GUESTDBG_USE_SW_BP: using software breakpoints [x86, arm64]
3630 - KVM_GUESTDBG_USE_HW_BP: using hardware breakpoints [x86, s390]
3631 - KVM_GUESTDBG_USE_HW: using hardware debug events [arm64]
3632 - KVM_GUESTDBG_INJECT_DB: inject DB type exception [x86]
3633 - KVM_GUESTDBG_INJECT_BP: inject BP type exception [x86]
3634 - KVM_GUESTDBG_EXIT_PENDING: trigger an immediate guest exit [s390]
3635 - KVM_GUESTDBG_BLOCKIRQ: avoid injecting interrupts/NMI/SMI [x86]
3637 For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3638 are enabled in memory so we need to ensure breakpoint exceptions are
3639 correctly trapped and the KVM run loop exits at the breakpoint and not
3640 running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3641 we need to ensure the guest vCPUs architecture specific registers are
3642 updated to the correct (supplied) values.
3644 The second part of the structure is architecture specific and
3645 typically contains a set of debug registers.
3647 For arm64 the number of debug registers is implementation defined and
3648 can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3649 KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3650 indicating the number of supported registers.
3652 For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3653 the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3655 Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3656 supported KVM_GUESTDBG_* bits in the control field.
3658 When debug events exit the main run loop with the reason
3659 KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3660 structure containing architecture specific debug information.
3662 4.88 KVM_GET_EMULATED_CPUID
3663 ---------------------------
3665 :Capability: KVM_CAP_EXT_EMUL_CPUID
3668 :Parameters: struct kvm_cpuid2 (in/out)
3669 :Returns: 0 on success, -1 on error
3676 struct kvm_cpuid_entry2 entries[0];
3679 The member 'flags' is used for passing flags from userspace.
3683 #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX BIT(0)
3684 #define KVM_CPUID_FLAG_STATEFUL_FUNC BIT(1) /* deprecated */
3685 #define KVM_CPUID_FLAG_STATE_READ_NEXT BIT(2) /* deprecated */
3687 struct kvm_cpuid_entry2 {
3698 This ioctl returns x86 cpuid features which are emulated by
3699 kvm.Userspace can use the information returned by this ioctl to query
3700 which features are emulated by kvm instead of being present natively.
3702 Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3703 structure with the 'nent' field indicating the number of entries in
3704 the variable-size array 'entries'. If the number of entries is too low
3705 to describe the cpu capabilities, an error (E2BIG) is returned. If the
3706 number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3707 is returned. If the number is just right, the 'nent' field is adjusted
3708 to the number of valid entries in the 'entries' array, which is then
3711 The entries returned are the set CPUID bits of the respective features
3712 which kvm emulates, as returned by the CPUID instruction, with unknown
3713 or unsupported feature bits cleared.
3715 Features like x2apic, for example, may not be present in the host cpu
3716 but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3717 emulated efficiently and thus not included here.
3719 The fields in each entry are defined as follows:
3722 the eax value used to obtain the entry
3724 the ecx value used to obtain the entry (for entries that are
3727 an OR of zero or more of the following:
3729 KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3730 if the index field is valid
3734 the values returned by the cpuid instruction for
3735 this function/index combination
3737 4.89 KVM_S390_MEM_OP
3738 --------------------
3740 :Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION
3741 :Architectures: s390
3742 :Type: vm ioctl, vcpu ioctl
3743 :Parameters: struct kvm_s390_mem_op (in)
3744 :Returns: = 0 on success,
3745 < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3746 > 0 if an exception occurred while walking the page tables
3748 Read or write data from/to the VM's memory.
3749 The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is
3752 Parameters are specified via the following structure::
3754 struct kvm_s390_mem_op {
3755 __u64 gaddr; /* the guest address */
3756 __u64 flags; /* flags */
3757 __u32 size; /* amount of bytes */
3758 __u32 op; /* type of operation */
3759 __u64 buf; /* buffer in userspace */
3762 __u8 ar; /* the access register number */
3763 __u8 key; /* access key, ignored if flag unset */
3765 __u32 sida_offset; /* offset into the sida */
3766 __u8 reserved[32]; /* ignored */
3770 The start address of the memory region has to be specified in the "gaddr"
3771 field, and the length of the region in the "size" field (which must not
3772 be 0). The maximum value for "size" can be obtained by checking the
3773 KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3774 userspace application where the read data should be written to for
3775 a read access, or where the data that should be written is stored for
3776 a write access. The "reserved" field is meant for future extensions.
3777 Reserved and unused values are ignored. Future extension that add members must
3778 introduce new flags.
3780 The type of operation is specified in the "op" field. Flags modifying
3781 their behavior can be set in the "flags" field. Undefined flag bits must
3784 Possible operations are:
3785 * ``KVM_S390_MEMOP_LOGICAL_READ``
3786 * ``KVM_S390_MEMOP_LOGICAL_WRITE``
3787 * ``KVM_S390_MEMOP_ABSOLUTE_READ``
3788 * ``KVM_S390_MEMOP_ABSOLUTE_WRITE``
3789 * ``KVM_S390_MEMOP_SIDA_READ``
3790 * ``KVM_S390_MEMOP_SIDA_WRITE``
3795 Access logical memory, i.e. translate the given guest address to an absolute
3796 address given the state of the VCPU and use the absolute address as target of
3797 the access. "ar" designates the access register number to be used; the valid
3799 Logical accesses are permitted for the VCPU ioctl only.
3800 Logical accesses are permitted for non-protected guests only.
3803 * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3804 * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION``
3805 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3807 The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the
3808 corresponding memory access would cause an access exception; however,
3809 no actual access to the data in memory at the destination is performed.
3810 In this case, "buf" is unused and can be NULL.
3812 In case an access exception occurred during the access (or would occur
3813 in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive
3814 error number indicating the type of exception. This exception is also
3815 raised directly at the corresponding VCPU if the flag
3816 KVM_S390_MEMOP_F_INJECT_EXCEPTION is set.
3817 On protection exceptions, unless specified otherwise, the injected
3818 translation-exception identifier (TEID) indicates suppression.
3820 If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key
3821 protection is also in effect and may cause exceptions if accesses are
3822 prohibited given the access key designated by "key"; the valid range is 0..15.
3823 KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION
3825 Since the accessed memory may span multiple pages and those pages might have
3826 different storage keys, it is possible that a protection exception occurs
3827 after memory has been modified. In this case, if the exception is injected,
3828 the TEID does not indicate suppression.
3830 Absolute read/write:
3831 ^^^^^^^^^^^^^^^^^^^^
3833 Access absolute memory. This operation is intended to be used with the
3834 KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing
3835 the checks required for storage key protection as one operation (as opposed to
3836 user space getting the storage keys, performing the checks, and accessing
3837 memory thereafter, which could lead to a delay between check and access).
3838 Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION
3840 Currently absolute accesses are not permitted for VCPU ioctls.
3841 Absolute accesses are permitted for non-protected guests only.
3844 * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3845 * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3847 The semantics of the flags are as for logical accesses.
3852 Access the secure instruction data area which contains memory operands necessary
3853 for instruction emulation for protected guests.
3854 SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available.
3855 SIDA accesses are permitted for the VCPU ioctl only.
3856 SIDA accesses are permitted for protected guests only.
3858 No flags are supported.
3860 4.90 KVM_S390_GET_SKEYS
3861 -----------------------
3863 :Capability: KVM_CAP_S390_SKEYS
3864 :Architectures: s390
3866 :Parameters: struct kvm_s390_skeys
3867 :Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage
3868 keys, negative value on error
3870 This ioctl is used to get guest storage key values on the s390
3871 architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3873 struct kvm_s390_skeys {
3876 __u64 skeydata_addr;
3881 The start_gfn field is the number of the first guest frame whose storage keys
3884 The count field is the number of consecutive frames (starting from start_gfn)
3885 whose storage keys to get. The count field must be at least 1 and the maximum
3886 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3887 will cause the ioctl to return -EINVAL.
3889 The skeydata_addr field is the address to a buffer large enough to hold count
3890 bytes. This buffer will be filled with storage key data by the ioctl.
3892 4.91 KVM_S390_SET_SKEYS
3893 -----------------------
3895 :Capability: KVM_CAP_S390_SKEYS
3896 :Architectures: s390
3898 :Parameters: struct kvm_s390_skeys
3899 :Returns: 0 on success, negative value on error
3901 This ioctl is used to set guest storage key values on the s390
3902 architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3903 See section on KVM_S390_GET_SKEYS for struct definition.
3905 The start_gfn field is the number of the first guest frame whose storage keys
3908 The count field is the number of consecutive frames (starting from start_gfn)
3909 whose storage keys to get. The count field must be at least 1 and the maximum
3910 allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3911 will cause the ioctl to return -EINVAL.
3913 The skeydata_addr field is the address to a buffer containing count bytes of
3914 storage keys. Each byte in the buffer will be set as the storage key for a
3915 single frame starting at start_gfn for count frames.
3917 Note: If any architecturally invalid key value is found in the given data then
3918 the ioctl will return -EINVAL.
3923 :Capability: KVM_CAP_S390_INJECT_IRQ
3924 :Architectures: s390
3926 :Parameters: struct kvm_s390_irq (in)
3927 :Returns: 0 on success, -1 on error
3932 ====== =================================================================
3933 EINVAL interrupt type is invalid
3934 type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3935 type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3936 than the maximum of VCPUs
3937 EBUSY type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3938 type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3939 type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3941 ====== =================================================================
3943 Allows to inject an interrupt to the guest.
3945 Using struct kvm_s390_irq as a parameter allows
3946 to inject additional payload which is not
3947 possible via KVM_S390_INTERRUPT.
3949 Interrupt parameters are passed via kvm_s390_irq::
3951 struct kvm_s390_irq {
3954 struct kvm_s390_io_info io;
3955 struct kvm_s390_ext_info ext;
3956 struct kvm_s390_pgm_info pgm;
3957 struct kvm_s390_emerg_info emerg;
3958 struct kvm_s390_extcall_info extcall;
3959 struct kvm_s390_prefix_info prefix;
3960 struct kvm_s390_stop_info stop;
3961 struct kvm_s390_mchk_info mchk;
3966 type can be one of the following:
3968 - KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3969 - KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3970 - KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3971 - KVM_S390_RESTART - restart; no parameters
3972 - KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3973 - KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3974 - KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3975 - KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3976 - KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3978 This is an asynchronous vcpu ioctl and can be invoked from any thread.
3980 4.94 KVM_S390_GET_IRQ_STATE
3981 ---------------------------
3983 :Capability: KVM_CAP_S390_IRQ_STATE
3984 :Architectures: s390
3986 :Parameters: struct kvm_s390_irq_state (out)
3987 :Returns: >= number of bytes copied into buffer,
3988 -EINVAL if buffer size is 0,
3989 -ENOBUFS if buffer size is too small to fit all pending interrupts,
3990 -EFAULT if the buffer address was invalid
3992 This ioctl allows userspace to retrieve the complete state of all currently
3993 pending interrupts in a single buffer. Use cases include migration
3994 and introspection. The parameter structure contains the address of a
3995 userspace buffer and its length::
3997 struct kvm_s390_irq_state {
3999 __u32 flags; /* will stay unused for compatibility reasons */
4001 __u32 reserved[4]; /* will stay unused for compatibility reasons */
4004 Userspace passes in the above struct and for each pending interrupt a
4005 struct kvm_s390_irq is copied to the provided buffer.
4007 The structure contains a flags and a reserved field for future extensions. As
4008 the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
4009 reserved, these fields can not be used in the future without breaking
4012 If -ENOBUFS is returned the buffer provided was too small and userspace
4013 may retry with a bigger buffer.
4015 4.95 KVM_S390_SET_IRQ_STATE
4016 ---------------------------
4018 :Capability: KVM_CAP_S390_IRQ_STATE
4019 :Architectures: s390
4021 :Parameters: struct kvm_s390_irq_state (in)
4022 :Returns: 0 on success,
4023 -EFAULT if the buffer address was invalid,
4024 -EINVAL for an invalid buffer length (see below),
4025 -EBUSY if there were already interrupts pending,
4026 errors occurring when actually injecting the
4027 interrupt. See KVM_S390_IRQ.
4029 This ioctl allows userspace to set the complete state of all cpu-local
4030 interrupts currently pending for the vcpu. It is intended for restoring
4031 interrupt state after a migration. The input parameter is a userspace buffer
4032 containing a struct kvm_s390_irq_state::
4034 struct kvm_s390_irq_state {
4036 __u32 flags; /* will stay unused for compatibility reasons */
4038 __u32 reserved[4]; /* will stay unused for compatibility reasons */
4041 The restrictions for flags and reserved apply as well.
4042 (see KVM_S390_GET_IRQ_STATE)
4044 The userspace memory referenced by buf contains a struct kvm_s390_irq
4045 for each interrupt to be injected into the guest.
4046 If one of the interrupts could not be injected for some reason the
4049 len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
4050 and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
4051 which is the maximum number of possibly pending cpu-local interrupts.
4056 :Capability: KVM_CAP_X86_SMM
4060 :Returns: 0 on success, -1 on error
4062 Queues an SMI on the thread's vcpu.
4064 4.97 KVM_X86_SET_MSR_FILTER
4065 ----------------------------
4067 :Capability: KVM_X86_SET_MSR_FILTER
4070 :Parameters: struct kvm_msr_filter
4071 :Returns: 0 on success, < 0 on error
4075 struct kvm_msr_filter_range {
4076 #define KVM_MSR_FILTER_READ (1 << 0)
4077 #define KVM_MSR_FILTER_WRITE (1 << 1)
4079 __u32 nmsrs; /* number of msrs in bitmap */
4080 __u32 base; /* MSR index the bitmap starts at */
4081 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4084 #define KVM_MSR_FILTER_MAX_RANGES 16
4085 struct kvm_msr_filter {
4086 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4087 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
4089 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4092 flags values for ``struct kvm_msr_filter_range``:
4094 ``KVM_MSR_FILTER_READ``
4096 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4097 indicates that a read should immediately fail, while a 1 indicates that
4098 a read for a particular MSR should be handled regardless of the default
4101 ``KVM_MSR_FILTER_WRITE``
4103 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4104 indicates that a write should immediately fail, while a 1 indicates that
4105 a write for a particular MSR should be handled regardless of the default
4108 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4110 Filter both read and write accesses to MSRs using the given bitmap. A 0
4111 in the bitmap indicates that both reads and writes should immediately fail,
4112 while a 1 indicates that reads and writes for a particular MSR are not
4113 filtered by this range.
4115 flags values for ``struct kvm_msr_filter``:
4117 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4119 If no filter range matches an MSR index that is getting accessed, KVM will
4120 fall back to allowing access to the MSR.
4122 ``KVM_MSR_FILTER_DEFAULT_DENY``
4124 If no filter range matches an MSR index that is getting accessed, KVM will
4125 fall back to rejecting access to the MSR. In this mode, all MSRs that should
4126 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4128 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4129 specify whether a certain MSR access should be explicitly filtered for or not.
4131 If this ioctl has never been invoked, MSR accesses are not guarded and the
4132 default KVM in-kernel emulation behavior is fully preserved.
4134 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4135 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4138 As soon as the filtering is in place, every MSR access is processed through
4139 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4140 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4141 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4145 MSR accesses coming from nested vmentry/vmexit are not filtered.
4146 This includes both writes to individual VMCS fields and reads/writes
4147 through the MSR lists pointed to by the VMCS.
4149 If a bit is within one of the defined ranges, read and write accesses are
4150 guarded by the bitmap's value for the MSR index if the kind of access
4151 is included in the ``struct kvm_msr_filter_range`` flags. If no range
4152 cover this particular access, the behavior is determined by the flags
4153 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4154 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4156 Each bitmap range specifies a range of MSRs to potentially allow access on.
4157 The range goes from MSR index [base .. base+nmsrs]. The flags field
4158 indicates whether reads, writes or both reads and writes are filtered
4159 by setting a 1 bit in the bitmap for the corresponding MSR index.
4161 If an MSR access is not permitted through the filtering, it generates a
4162 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4163 allows user space to deflect and potentially handle various MSR accesses
4166 If a vCPU is in running state while this ioctl is invoked, the vCPU may
4167 experience inconsistent filtering behavior on MSR accesses.
4169 4.98 KVM_CREATE_SPAPR_TCE_64
4170 ----------------------------
4172 :Capability: KVM_CAP_SPAPR_TCE_64
4173 :Architectures: powerpc
4175 :Parameters: struct kvm_create_spapr_tce_64 (in)
4176 :Returns: file descriptor for manipulating the created TCE table
4178 This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4179 windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4181 This capability uses extended struct in ioctl interface::
4183 /* for KVM_CAP_SPAPR_TCE_64 */
4184 struct kvm_create_spapr_tce_64 {
4188 __u64 offset; /* in pages */
4189 __u64 size; /* in pages */
4192 The aim of extension is to support an additional bigger DMA window with
4193 a variable page size.
4194 KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4195 a bus offset of the corresponding DMA window, @size and @offset are numbers
4198 @flags are not used at the moment.
4200 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4202 4.99 KVM_REINJECT_CONTROL
4203 -------------------------
4205 :Capability: KVM_CAP_REINJECT_CONTROL
4208 :Parameters: struct kvm_reinject_control (in)
4209 :Returns: 0 on success,
4210 -EFAULT if struct kvm_reinject_control cannot be read,
4211 -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4213 i8254 (PIT) has two modes, reinject and !reinject. The default is reinject,
4214 where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4215 vector(s) that i8254 injects. Reinject mode dequeues a tick and injects its
4216 interrupt whenever there isn't a pending interrupt from i8254.
4217 !reinject mode injects an interrupt as soon as a tick arrives.
4221 struct kvm_reinject_control {
4226 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4227 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4229 4.100 KVM_PPC_CONFIGURE_V3_MMU
4230 ------------------------------
4232 :Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
4235 :Parameters: struct kvm_ppc_mmuv3_cfg (in)
4236 :Returns: 0 on success,
4237 -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4238 -EINVAL if the configuration is invalid
4240 This ioctl controls whether the guest will use radix or HPT (hashed
4241 page table) translation, and sets the pointer to the process table for
4246 struct kvm_ppc_mmuv3_cfg {
4248 __u64 process_table;
4251 There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4252 KVM_PPC_MMUV3_GTSE. KVM_PPC_MMUV3_RADIX, if set, configures the guest
4253 to use radix tree translation, and if clear, to use HPT translation.
4254 KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4255 to be able to use the global TLB and SLB invalidation instructions;
4256 if clear, the guest may not use these instructions.
4258 The process_table field specifies the address and size of the guest
4259 process table, which is in the guest's space. This field is formatted
4260 as the second doubleword of the partition table entry, as defined in
4261 the Power ISA V3.00, Book III section 5.7.6.1.
4263 4.101 KVM_PPC_GET_RMMU_INFO
4264 ---------------------------
4266 :Capability: KVM_CAP_PPC_RADIX_MMU
4269 :Parameters: struct kvm_ppc_rmmu_info (out)
4270 :Returns: 0 on success,
4271 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4272 -EINVAL if no useful information can be returned
4274 This ioctl returns a structure containing two things: (a) a list
4275 containing supported radix tree geometries, and (b) a list that maps
4276 page sizes to put in the "AP" (actual page size) field for the tlbie
4277 (TLB invalidate entry) instruction.
4281 struct kvm_ppc_rmmu_info {
4282 struct kvm_ppc_radix_geom {
4287 __u32 ap_encodings[8];
4290 The geometries[] field gives up to 8 supported geometries for the
4291 radix page table, in terms of the log base 2 of the smallest page
4292 size, and the number of bits indexed at each level of the tree, from
4293 the PTE level up to the PGD level in that order. Any unused entries
4294 will have 0 in the page_shift field.
4296 The ap_encodings gives the supported page sizes and their AP field
4297 encodings, encoded with the AP value in the top 3 bits and the log
4298 base 2 of the page size in the bottom 6 bits.
4300 4.102 KVM_PPC_RESIZE_HPT_PREPARE
4301 --------------------------------
4303 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4304 :Architectures: powerpc
4306 :Parameters: struct kvm_ppc_resize_hpt (in)
4307 :Returns: 0 on successful completion,
4308 >0 if a new HPT is being prepared, the value is an estimated
4309 number of milliseconds until preparation is complete,
4310 -EFAULT if struct kvm_reinject_control cannot be read,
4311 -EINVAL if the supplied shift or flags are invalid,
4312 -ENOMEM if unable to allocate the new HPT,
4314 Used to implement the PAPR extension for runtime resizing of a guest's
4315 Hashed Page Table (HPT). Specifically this starts, stops or monitors
4316 the preparation of a new potential HPT for the guest, essentially
4317 implementing the H_RESIZE_HPT_PREPARE hypercall.
4321 struct kvm_ppc_resize_hpt {
4327 If called with shift > 0 when there is no pending HPT for the guest,
4328 this begins preparation of a new pending HPT of size 2^(shift) bytes.
4329 It then returns a positive integer with the estimated number of
4330 milliseconds until preparation is complete.
4332 If called when there is a pending HPT whose size does not match that
4333 requested in the parameters, discards the existing pending HPT and
4334 creates a new one as above.
4336 If called when there is a pending HPT of the size requested, will:
4338 * If preparation of the pending HPT is already complete, return 0
4339 * If preparation of the pending HPT has failed, return an error
4340 code, then discard the pending HPT.
4341 * If preparation of the pending HPT is still in progress, return an
4342 estimated number of milliseconds until preparation is complete.
4344 If called with shift == 0, discards any currently pending HPT and
4345 returns 0 (i.e. cancels any in-progress preparation).
4347 flags is reserved for future expansion, currently setting any bits in
4348 flags will result in an -EINVAL.
4350 Normally this will be called repeatedly with the same parameters until
4351 it returns <= 0. The first call will initiate preparation, subsequent
4352 ones will monitor preparation until it completes or fails.
4354 4.103 KVM_PPC_RESIZE_HPT_COMMIT
4355 -------------------------------
4357 :Capability: KVM_CAP_SPAPR_RESIZE_HPT
4358 :Architectures: powerpc
4360 :Parameters: struct kvm_ppc_resize_hpt (in)
4361 :Returns: 0 on successful completion,
4362 -EFAULT if struct kvm_reinject_control cannot be read,
4363 -EINVAL if the supplied shift or flags are invalid,
4364 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4365 have the requested size,
4366 -EBUSY if the pending HPT is not fully prepared,
4367 -ENOSPC if there was a hash collision when moving existing
4368 HPT entries to the new HPT,
4369 -EIO on other error conditions
4371 Used to implement the PAPR extension for runtime resizing of a guest's
4372 Hashed Page Table (HPT). Specifically this requests that the guest be
4373 transferred to working with the new HPT, essentially implementing the
4374 H_RESIZE_HPT_COMMIT hypercall.
4378 struct kvm_ppc_resize_hpt {
4384 This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4385 returned 0 with the same parameters. In other cases
4386 KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4387 -EBUSY, though others may be possible if the preparation was started,
4390 This will have undefined effects on the guest if it has not already
4391 placed itself in a quiescent state where no vcpu will make MMU enabled
4394 On succsful completion, the pending HPT will become the guest's active
4395 HPT and the previous HPT will be discarded.
4397 On failure, the guest will still be operating on its previous HPT.
4399 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4400 -----------------------------------
4402 :Capability: KVM_CAP_MCE
4405 :Parameters: u64 mce_cap (out)
4406 :Returns: 0 on success, -1 on error
4408 Returns supported MCE capabilities. The u64 mce_cap parameter
4409 has the same format as the MSR_IA32_MCG_CAP register. Supported
4410 capabilities will have the corresponding bits set.
4412 4.105 KVM_X86_SETUP_MCE
4413 -----------------------
4415 :Capability: KVM_CAP_MCE
4418 :Parameters: u64 mcg_cap (in)
4419 :Returns: 0 on success,
4420 -EFAULT if u64 mcg_cap cannot be read,
4421 -EINVAL if the requested number of banks is invalid,
4422 -EINVAL if requested MCE capability is not supported.
4424 Initializes MCE support for use. The u64 mcg_cap parameter
4425 has the same format as the MSR_IA32_MCG_CAP register and
4426 specifies which capabilities should be enabled. The maximum
4427 supported number of error-reporting banks can be retrieved when
4428 checking for KVM_CAP_MCE. The supported capabilities can be
4429 retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4431 4.106 KVM_X86_SET_MCE
4432 ---------------------
4434 :Capability: KVM_CAP_MCE
4437 :Parameters: struct kvm_x86_mce (in)
4438 :Returns: 0 on success,
4439 -EFAULT if struct kvm_x86_mce cannot be read,
4440 -EINVAL if the bank number is invalid,
4441 -EINVAL if VAL bit is not set in status field.
4443 Inject a machine check error (MCE) into the guest. The input
4446 struct kvm_x86_mce {
4456 If the MCE being reported is an uncorrected error, KVM will
4457 inject it as an MCE exception into the guest. If the guest
4458 MCG_STATUS register reports that an MCE is in progress, KVM
4459 causes an KVM_EXIT_SHUTDOWN vmexit.
4461 Otherwise, if the MCE is a corrected error, KVM will just
4462 store it in the corresponding bank (provided this bank is
4463 not holding a previously reported uncorrected error).
4465 4.107 KVM_S390_GET_CMMA_BITS
4466 ----------------------------
4468 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4469 :Architectures: s390
4471 :Parameters: struct kvm_s390_cmma_log (in, out)
4472 :Returns: 0 on success, a negative value on error
4474 This ioctl is used to get the values of the CMMA bits on the s390
4475 architecture. It is meant to be used in two scenarios:
4477 - During live migration to save the CMMA values. Live migration needs
4478 to be enabled via the KVM_REQ_START_MIGRATION VM property.
4479 - To non-destructively peek at the CMMA values, with the flag
4480 KVM_S390_CMMA_PEEK set.
4482 The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4483 values are written to a buffer whose location is indicated via the "values"
4484 member in the kvm_s390_cmma_log struct. The values in the input struct are
4485 also updated as needed.
4487 Each CMMA value takes up one byte.
4491 struct kvm_s390_cmma_log {
4502 start_gfn is the number of the first guest frame whose CMMA values are
4505 count is the length of the buffer in bytes,
4507 values points to the buffer where the result will be written to.
4509 If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4510 KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4513 The result is written in the buffer pointed to by the field values, and
4514 the values of the input parameter are updated as follows.
4516 Depending on the flags, different actions are performed. The only
4517 supported flag so far is KVM_S390_CMMA_PEEK.
4519 The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4520 start_gfn will indicate the first page frame whose CMMA bits were dirty.
4521 It is not necessarily the same as the one passed as input, as clean pages
4524 count will indicate the number of bytes actually written in the buffer.
4525 It can (and very often will) be smaller than the input value, since the
4526 buffer is only filled until 16 bytes of clean values are found (which
4527 are then not copied in the buffer). Since a CMMA migration block needs
4528 the base address and the length, for a total of 16 bytes, we will send
4529 back some clean data if there is some dirty data afterwards, as long as
4530 the size of the clean data does not exceed the size of the header. This
4531 allows to minimize the amount of data to be saved or transferred over
4532 the network at the expense of more roundtrips to userspace. The next
4533 invocation of the ioctl will skip over all the clean values, saving
4534 potentially more than just the 16 bytes we found.
4536 If KVM_S390_CMMA_PEEK is set:
4537 the existing storage attributes are read even when not in migration
4538 mode, and no other action is performed;
4540 the output start_gfn will be equal to the input start_gfn,
4542 the output count will be equal to the input count, except if the end of
4543 memory has been reached.
4546 the field "remaining" will indicate the total number of dirty CMMA values
4547 still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4552 values points to the userspace buffer where the result will be stored.
4554 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4555 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4556 KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4557 -EFAULT if the userspace address is invalid or if no page table is
4558 present for the addresses (e.g. when using hugepages).
4560 4.108 KVM_S390_SET_CMMA_BITS
4561 ----------------------------
4563 :Capability: KVM_CAP_S390_CMMA_MIGRATION
4564 :Architectures: s390
4566 :Parameters: struct kvm_s390_cmma_log (in)
4567 :Returns: 0 on success, a negative value on error
4569 This ioctl is used to set the values of the CMMA bits on the s390
4570 architecture. It is meant to be used during live migration to restore
4571 the CMMA values, but there are no restrictions on its use.
4572 The ioctl takes parameters via the kvm_s390_cmma_values struct.
4573 Each CMMA value takes up one byte.
4577 struct kvm_s390_cmma_log {
4588 start_gfn indicates the starting guest frame number,
4590 count indicates how many values are to be considered in the buffer,
4592 flags is not used and must be 0.
4594 mask indicates which PGSTE bits are to be considered.
4596 remaining is not used.
4598 values points to the buffer in userspace where to store the values.
4600 This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4601 complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4602 the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4603 if the flags field was not 0, with -EFAULT if the userspace address is
4604 invalid, if invalid pages are written to (e.g. after the end of memory)
4605 or if no page table is present for the addresses (e.g. when using
4608 4.109 KVM_PPC_GET_CPU_CHAR
4609 --------------------------
4611 :Capability: KVM_CAP_PPC_GET_CPU_CHAR
4612 :Architectures: powerpc
4614 :Parameters: struct kvm_ppc_cpu_char (out)
4615 :Returns: 0 on successful completion,
4616 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4618 This ioctl gives userspace information about certain characteristics
4619 of the CPU relating to speculative execution of instructions and
4620 possible information leakage resulting from speculative execution (see
4621 CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754). The information is
4622 returned in struct kvm_ppc_cpu_char, which looks like this::
4624 struct kvm_ppc_cpu_char {
4625 __u64 character; /* characteristics of the CPU */
4626 __u64 behaviour; /* recommended software behaviour */
4627 __u64 character_mask; /* valid bits in character */
4628 __u64 behaviour_mask; /* valid bits in behaviour */
4631 For extensibility, the character_mask and behaviour_mask fields
4632 indicate which bits of character and behaviour have been filled in by
4633 the kernel. If the set of defined bits is extended in future then
4634 userspace will be able to tell whether it is running on a kernel that
4635 knows about the new bits.
4637 The character field describes attributes of the CPU which can help
4638 with preventing inadvertent information disclosure - specifically,
4639 whether there is an instruction to flash-invalidate the L1 data cache
4640 (ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4641 to a mode where entries can only be used by the thread that created
4642 them, whether the bcctr[l] instruction prevents speculation, and
4643 whether a speculation barrier instruction (ori 31,31,0) is provided.
4645 The behaviour field describes actions that software should take to
4646 prevent inadvertent information disclosure, and thus describes which
4647 vulnerabilities the hardware is subject to; specifically whether the
4648 L1 data cache should be flushed when returning to user mode from the
4649 kernel, and whether a speculation barrier should be placed between an
4650 array bounds check and the array access.
4652 These fields use the same bit definitions as the new
4653 H_GET_CPU_CHARACTERISTICS hypercall.
4655 4.110 KVM_MEMORY_ENCRYPT_OP
4656 ---------------------------
4661 :Parameters: an opaque platform specific structure (in/out)
4662 :Returns: 0 on success; -1 on error
4664 If the platform supports creating encrypted VMs then this ioctl can be used
4665 for issuing platform-specific memory encryption commands to manage those
4668 Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4669 (SEV) commands on AMD Processors. The SEV commands are defined in
4670 Documentation/virt/kvm/amd-memory-encryption.rst.
4672 4.111 KVM_MEMORY_ENCRYPT_REG_REGION
4673 -----------------------------------
4678 :Parameters: struct kvm_enc_region (in)
4679 :Returns: 0 on success; -1 on error
4681 This ioctl can be used to register a guest memory region which may
4682 contain encrypted data (e.g. guest RAM, SMRAM etc).
4684 It is used in the SEV-enabled guest. When encryption is enabled, a guest
4685 memory region may contain encrypted data. The SEV memory encryption
4686 engine uses a tweak such that two identical plaintext pages, each at
4687 different locations will have differing ciphertexts. So swapping or
4688 moving ciphertext of those pages will not result in plaintext being
4689 swapped. So relocating (or migrating) physical backing pages for the SEV
4690 guest will require some additional steps.
4692 Note: The current SEV key management spec does not provide commands to
4693 swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4694 memory region registered with the ioctl.
4696 4.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4697 -------------------------------------
4702 :Parameters: struct kvm_enc_region (in)
4703 :Returns: 0 on success; -1 on error
4705 This ioctl can be used to unregister the guest memory region registered
4706 with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4708 4.113 KVM_HYPERV_EVENTFD
4709 ------------------------
4711 :Capability: KVM_CAP_HYPERV_EVENTFD
4714 :Parameters: struct kvm_hyperv_eventfd (in)
4716 This ioctl (un)registers an eventfd to receive notifications from the guest on
4717 the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4718 causing a user exit. SIGNAL_EVENT hypercall with non-zero event flag number
4719 (bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4723 struct kvm_hyperv_eventfd {
4730 The conn_id field should fit within 24 bits::
4732 #define KVM_HYPERV_CONN_ID_MASK 0x00ffffff
4734 The acceptable values for the flags field are::
4736 #define KVM_HYPERV_EVENTFD_DEASSIGN (1 << 0)
4738 :Returns: 0 on success,
4739 -EINVAL if conn_id or flags is outside the allowed range,
4740 -ENOENT on deassign if the conn_id isn't registered,
4741 -EEXIST on assign if the conn_id is already registered
4743 4.114 KVM_GET_NESTED_STATE
4744 --------------------------
4746 :Capability: KVM_CAP_NESTED_STATE
4749 :Parameters: struct kvm_nested_state (in/out)
4750 :Returns: 0 on success, -1 on error
4754 ===== =============================================================
4755 E2BIG the total state size exceeds the value of 'size' specified by
4756 the user; the size required will be written into size.
4757 ===== =============================================================
4761 struct kvm_nested_state {
4767 struct kvm_vmx_nested_state_hdr vmx;
4768 struct kvm_svm_nested_state_hdr svm;
4770 /* Pad the header to 128 bytes. */
4775 struct kvm_vmx_nested_state_data vmx[0];
4776 struct kvm_svm_nested_state_data svm[0];
4780 #define KVM_STATE_NESTED_GUEST_MODE 0x00000001
4781 #define KVM_STATE_NESTED_RUN_PENDING 0x00000002
4782 #define KVM_STATE_NESTED_EVMCS 0x00000004
4784 #define KVM_STATE_NESTED_FORMAT_VMX 0
4785 #define KVM_STATE_NESTED_FORMAT_SVM 1
4787 #define KVM_STATE_NESTED_VMX_VMCS_SIZE 0x1000
4789 #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE 0x00000001
4790 #define KVM_STATE_NESTED_VMX_SMM_VMXON 0x00000002
4792 #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4794 struct kvm_vmx_nested_state_hdr {
4803 __u64 preemption_timer_deadline;
4806 struct kvm_vmx_nested_state_data {
4807 __u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4808 __u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4811 This ioctl copies the vcpu's nested virtualization state from the kernel to
4814 The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4815 to the KVM_CHECK_EXTENSION ioctl().
4817 4.115 KVM_SET_NESTED_STATE
4818 --------------------------
4820 :Capability: KVM_CAP_NESTED_STATE
4823 :Parameters: struct kvm_nested_state (in)
4824 :Returns: 0 on success, -1 on error
4826 This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4827 For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4829 4.116 KVM_(UN)REGISTER_COALESCED_MMIO
4830 -------------------------------------
4832 :Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4833 KVM_CAP_COALESCED_PIO (for coalesced pio)
4836 :Parameters: struct kvm_coalesced_mmio_zone
4837 :Returns: 0 on success, < 0 on error
4839 Coalesced I/O is a performance optimization that defers hardware
4840 register write emulation so that userspace exits are avoided. It is
4841 typically used to reduce the overhead of emulating frequently accessed
4844 When a hardware register is configured for coalesced I/O, write accesses
4845 do not exit to userspace and their value is recorded in a ring buffer
4846 that is shared between kernel and userspace.
4848 Coalesced I/O is used if one or more write accesses to a hardware
4849 register can be deferred until a read or a write to another hardware
4850 register on the same device. This last access will cause a vmexit and
4851 userspace will process accesses from the ring buffer before emulating
4852 it. That will avoid exiting to userspace on repeated writes.
4854 Coalesced pio is based on coalesced mmio. There is little difference
4855 between coalesced mmio and pio except that coalesced pio records accesses
4858 4.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4859 ------------------------------------
4861 :Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4862 :Architectures: x86, arm64, mips
4864 :Parameters: struct kvm_clear_dirty_log (in)
4865 :Returns: 0 on success, -1 on error
4869 /* for KVM_CLEAR_DIRTY_LOG */
4870 struct kvm_clear_dirty_log {
4875 void __user *dirty_bitmap; /* one bit per page */
4880 The ioctl clears the dirty status of pages in a memory slot, according to
4881 the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4882 field. Bit 0 of the bitmap corresponds to page "first_page" in the
4883 memory slot, and num_pages is the size in bits of the input bitmap.
4884 first_page must be a multiple of 64; num_pages must also be a multiple of
4885 64 unless first_page + num_pages is the size of the memory slot. For each
4886 bit that is set in the input bitmap, the corresponding page is marked "clean"
4887 in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4888 (for example via write-protection, or by clearing the dirty bit in
4889 a page table entry).
4891 If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4892 the address space for which you want to clear the dirty status. See
4893 KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4895 This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4896 is enabled; for more information, see the description of the capability.
4897 However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4898 that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4900 4.118 KVM_GET_SUPPORTED_HV_CPUID
4901 --------------------------------
4903 :Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4905 :Type: system ioctl, vcpu ioctl
4906 :Parameters: struct kvm_cpuid2 (in/out)
4907 :Returns: 0 on success, -1 on error
4914 struct kvm_cpuid_entry2 entries[0];
4917 struct kvm_cpuid_entry2 {
4928 This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4929 KVM. Userspace can use the information returned by this ioctl to construct
4930 cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4931 Windows or Hyper-V guests).
4933 CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4934 Functional Specification (TLFS). These leaves can't be obtained with
4935 KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4936 leaves (0x40000000, 0x40000001).
4938 Currently, the following list of CPUID leaves are returned:
4940 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4941 - HYPERV_CPUID_INTERFACE
4942 - HYPERV_CPUID_VERSION
4943 - HYPERV_CPUID_FEATURES
4944 - HYPERV_CPUID_ENLIGHTMENT_INFO
4945 - HYPERV_CPUID_IMPLEMENT_LIMITS
4946 - HYPERV_CPUID_NESTED_FEATURES
4947 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4948 - HYPERV_CPUID_SYNDBG_INTERFACE
4949 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4951 Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4952 with the 'nent' field indicating the number of entries in the variable-size
4953 array 'entries'. If the number of entries is too low to describe all Hyper-V
4954 feature leaves, an error (E2BIG) is returned. If the number is more or equal
4955 to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4956 number of valid entries in the 'entries' array, which is then filled.
4958 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4959 userspace should not expect to get any particular value there.
4961 Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4962 system ioctl which exposes all supported feature bits unconditionally, vcpu
4963 version has the following quirks:
4965 - HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4966 feature bit are only exposed when Enlightened VMCS was previously enabled
4967 on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4968 - HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4969 (presumes KVM_CREATE_IRQCHIP has already been called).
4971 4.119 KVM_ARM_VCPU_FINALIZE
4972 ---------------------------
4974 :Architectures: arm64
4976 :Parameters: int feature (in)
4977 :Returns: 0 on success, -1 on error
4981 ====== ==============================================================
4982 EPERM feature not enabled, needs configuration, or already finalized
4983 EINVAL feature unknown or not present
4984 ====== ==============================================================
4986 Recognised values for feature:
4988 ===== ===========================================
4989 arm64 KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4990 ===== ===========================================
4992 Finalizes the configuration of the specified vcpu feature.
4994 The vcpu must already have been initialised, enabling the affected feature, by
4995 means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4998 For affected vcpu features, this is a mandatory step that must be performed
4999 before the vcpu is fully usable.
5001 Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
5002 configured by use of ioctls such as KVM_SET_ONE_REG. The exact configuration
5003 that should be performaned and how to do it are feature-dependent.
5005 Other calls that depend on a particular feature being finalized, such as
5006 KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
5007 -EPERM unless the feature has already been finalized by means of a
5008 KVM_ARM_VCPU_FINALIZE call.
5010 See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
5013 4.120 KVM_SET_PMU_EVENT_FILTER
5014 ------------------------------
5016 :Capability: KVM_CAP_PMU_EVENT_FILTER
5019 :Parameters: struct kvm_pmu_event_filter (in)
5020 :Returns: 0 on success, -1 on error
5024 struct kvm_pmu_event_filter {
5027 __u32 fixed_counter_bitmap;
5033 This ioctl restricts the set of PMU events that the guest can program.
5034 The argument holds a list of events which will be allowed or denied.
5035 The eventsel+umask of each event the guest attempts to program is compared
5036 against the events field to determine whether the guest should have access.
5037 The events field only controls general purpose counters; fixed purpose
5038 counters are controlled by the fixed_counter_bitmap.
5040 No flags are defined yet, the field must be zero.
5042 Valid values for 'action'::
5044 #define KVM_PMU_EVENT_ALLOW 0
5045 #define KVM_PMU_EVENT_DENY 1
5047 4.121 KVM_PPC_SVM_OFF
5048 ---------------------
5051 :Architectures: powerpc
5054 :Returns: 0 on successful completion,
5058 ====== ================================================================
5059 EINVAL if ultravisor failed to terminate the secure guest
5060 ENOMEM if hypervisor failed to allocate new radix page tables for guest
5061 ====== ================================================================
5063 This ioctl is used to turn off the secure mode of the guest or transition
5064 the guest from secure mode to normal mode. This is invoked when the guest
5065 is reset. This has no effect if called for a normal guest.
5067 This ioctl issues an ultravisor call to terminate the secure guest,
5068 unpins the VPA pages and releases all the device pages that are used to
5069 track the secure pages by hypervisor.
5071 4.122 KVM_S390_NORMAL_RESET
5072 ---------------------------
5074 :Capability: KVM_CAP_S390_VCPU_RESETS
5075 :Architectures: s390
5080 This ioctl resets VCPU registers and control structures according to
5081 the cpu reset definition in the POP (Principles Of Operation).
5083 4.123 KVM_S390_INITIAL_RESET
5084 ----------------------------
5087 :Architectures: s390
5092 This ioctl resets VCPU registers and control structures according to
5093 the initial cpu reset definition in the POP. However, the cpu is not
5094 put into ESA mode. This reset is a superset of the normal reset.
5096 4.124 KVM_S390_CLEAR_RESET
5097 --------------------------
5099 :Capability: KVM_CAP_S390_VCPU_RESETS
5100 :Architectures: s390
5105 This ioctl resets VCPU registers and control structures according to
5106 the clear cpu reset definition in the POP. However, the cpu is not put
5107 into ESA mode. This reset is a superset of the initial reset.
5110 4.125 KVM_S390_PV_COMMAND
5111 -------------------------
5113 :Capability: KVM_CAP_S390_PROTECTED
5114 :Architectures: s390
5116 :Parameters: struct kvm_pv_cmd
5117 :Returns: 0 on success, < 0 on error
5122 __u32 cmd; /* Command to be executed */
5123 __u16 rc; /* Ultravisor return code */
5124 __u16 rrc; /* Ultravisor return reason code */
5125 __u64 data; /* Data or address */
5126 __u32 flags; /* flags for future extensions. Must be 0 for now */
5133 Allocate memory and register the VM with the Ultravisor, thereby
5134 donating memory to the Ultravisor that will become inaccessible to
5135 KVM. All existing CPUs are converted to protected ones. After this
5136 command has succeeded, any CPU added via hotplug will become
5137 protected during its creation as well.
5141 ===== =============================
5142 EINTR an unmasked signal is pending
5143 ===== =============================
5147 Deregister the VM from the Ultravisor and reclaim the memory that
5148 had been donated to the Ultravisor, making it usable by the kernel
5149 again. All registered VCPUs are converted back to non-protected
5152 KVM_PV_VM_SET_SEC_PARMS
5153 Pass the image header from VM memory to the Ultravisor in
5154 preparation of image unpacking and verification.
5157 Unpack (protect and decrypt) a page of the encrypted boot image.
5160 Verify the integrity of the unpacked image. Only if this succeeds,
5161 KVM is allowed to start protected VCPUs.
5163 4.126 KVM_X86_SET_MSR_FILTER
5164 ----------------------------
5166 :Capability: KVM_CAP_X86_MSR_FILTER
5169 :Parameters: struct kvm_msr_filter
5170 :Returns: 0 on success, < 0 on error
5174 struct kvm_msr_filter_range {
5175 #define KVM_MSR_FILTER_READ (1 << 0)
5176 #define KVM_MSR_FILTER_WRITE (1 << 1)
5178 __u32 nmsrs; /* number of msrs in bitmap */
5179 __u32 base; /* MSR index the bitmap starts at */
5180 __u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
5183 #define KVM_MSR_FILTER_MAX_RANGES 16
5184 struct kvm_msr_filter {
5185 #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
5186 #define KVM_MSR_FILTER_DEFAULT_DENY (1 << 0)
5188 struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
5191 flags values for ``struct kvm_msr_filter_range``:
5193 ``KVM_MSR_FILTER_READ``
5195 Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
5196 indicates that a read should immediately fail, while a 1 indicates that
5197 a read for a particular MSR should be handled regardless of the default
5200 ``KVM_MSR_FILTER_WRITE``
5202 Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
5203 indicates that a write should immediately fail, while a 1 indicates that
5204 a write for a particular MSR should be handled regardless of the default
5207 ``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
5209 Filter both read and write accesses to MSRs using the given bitmap. A 0
5210 in the bitmap indicates that both reads and writes should immediately fail,
5211 while a 1 indicates that reads and writes for a particular MSR are not
5212 filtered by this range.
5214 flags values for ``struct kvm_msr_filter``:
5216 ``KVM_MSR_FILTER_DEFAULT_ALLOW``
5218 If no filter range matches an MSR index that is getting accessed, KVM will
5219 fall back to allowing access to the MSR.
5221 ``KVM_MSR_FILTER_DEFAULT_DENY``
5223 If no filter range matches an MSR index that is getting accessed, KVM will
5224 fall back to rejecting access to the MSR. In this mode, all MSRs that should
5225 be processed by KVM need to explicitly be marked as allowed in the bitmaps.
5227 This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
5228 specify whether a certain MSR access should be explicitly filtered for or not.
5230 If this ioctl has never been invoked, MSR accesses are not guarded and the
5231 default KVM in-kernel emulation behavior is fully preserved.
5233 Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
5234 filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
5237 As soon as the filtering is in place, every MSR access is processed through
5238 the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
5239 x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
5240 and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
5243 If a bit is within one of the defined ranges, read and write accesses are
5244 guarded by the bitmap's value for the MSR index if the kind of access
5245 is included in the ``struct kvm_msr_filter_range`` flags. If no range
5246 cover this particular access, the behavior is determined by the flags
5247 field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
5248 and ``KVM_MSR_FILTER_DEFAULT_DENY``.
5250 Each bitmap range specifies a range of MSRs to potentially allow access on.
5251 The range goes from MSR index [base .. base+nmsrs]. The flags field
5252 indicates whether reads, writes or both reads and writes are filtered
5253 by setting a 1 bit in the bitmap for the corresponding MSR index.
5255 If an MSR access is not permitted through the filtering, it generates a
5256 #GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
5257 allows user space to deflect and potentially handle various MSR accesses
5260 Note, invoking this ioctl with a vCPU is running is inherently racy. However,
5261 KVM does guarantee that vCPUs will see either the previous filter or the new
5262 filter, e.g. MSRs with identical settings in both the old and new filter will
5263 have deterministic behavior.
5265 4.127 KVM_XEN_HVM_SET_ATTR
5266 --------------------------
5268 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5271 :Parameters: struct kvm_xen_hvm_attr
5272 :Returns: 0 on success, < 0 on error
5276 struct kvm_xen_hvm_attr {
5287 __u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */
5296 __u32 port; /* Zero for eventfd */
5309 KVM_XEN_ATTR_TYPE_LONG_MODE
5310 Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
5311 determines the layout of the shared info pages exposed to the VM.
5313 KVM_XEN_ATTR_TYPE_SHARED_INFO
5314 Sets the guest physical frame number at which the Xen "shared info"
5315 page resides. Note that although Xen places vcpu_info for the first
5316 32 vCPUs in the shared_info page, KVM does not automatically do so
5317 and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
5318 explicitly even when the vcpu_info for a given vCPU resides at the
5319 "default" location in the shared_info page. This is because KVM is
5320 not aware of the Xen CPU id which is used as the index into the
5321 vcpu_info[] array, so cannot know the correct default location.
5323 Note that the shared info page may be constantly written to by KVM;
5324 it contains the event channel bitmap used to deliver interrupts to
5325 a Xen guest, amongst other things. It is exempt from dirty tracking
5326 mechanisms — KVM will not explicitly mark the page as dirty each
5327 time an event channel interrupt is delivered to the guest! Thus,
5328 userspace should always assume that the designated GFN is dirty if
5329 any vCPU has been running or any event channel interrupts can be
5330 routed to the guest.
5332 KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
5333 Sets the exception vector used to deliver Xen event channel upcalls.
5334 This is the HVM-wide vector injected directly by the hypervisor
5335 (not through the local APIC), typically configured by a guest via
5336 HVM_PARAM_CALLBACK_IRQ.
5338 KVM_XEN_ATTR_TYPE_EVTCHN
5339 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5340 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5341 an outbound port number for interception of EVTCHNOP_send requests
5342 from the guest. A given sending port number may be directed back
5343 to a specified vCPU (by APIC ID) / port / priority on the guest,
5344 or to trigger events on an eventfd. The vCPU and priority can be
5345 changed by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call,
5346 but other fields cannot change for a given sending port. A port
5347 mapping is removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags
5350 KVM_XEN_ATTR_TYPE_XEN_VERSION
5351 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5352 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5353 the 32-bit version code returned to the guest when it invokes the
5354 XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV
5355 Xen guests will often use this to as a dummy hypercall to trigger
5356 event channel delivery, so responding within the kernel without
5357 exiting to userspace is beneficial.
5359 4.127 KVM_XEN_HVM_GET_ATTR
5360 --------------------------
5362 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5365 :Parameters: struct kvm_xen_hvm_attr
5366 :Returns: 0 on success, < 0 on error
5368 Allows Xen VM attributes to be read. For the structure and types,
5369 see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN
5370 attribute cannot be read.
5372 4.128 KVM_XEN_VCPU_SET_ATTR
5373 ---------------------------
5375 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5378 :Parameters: struct kvm_xen_vcpu_attr
5379 :Returns: 0 on success, < 0 on error
5383 struct kvm_xen_vcpu_attr {
5391 __u64 state_entry_time;
5393 __u64 time_runnable;
5409 KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
5410 Sets the guest physical address of the vcpu_info for a given vCPU.
5411 As with the shared_info page for the VM, the corresponding page may be
5412 dirtied at any time if event channel interrupt delivery is enabled, so
5413 userspace should always assume that the page is dirty without relying
5416 KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
5417 Sets the guest physical address of an additional pvclock structure
5418 for a given vCPU. This is typically used for guest vsyscall support.
5420 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5421 Sets the guest physical address of the vcpu_runstate_info for a given
5422 vCPU. This is how a Xen guest tracks CPU state such as steal time.
5424 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5425 Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5426 the given vCPU from the .u.runstate.state member of the structure.
5427 KVM automatically accounts running and runnable time but blocked
5428 and offline states are only entered explicitly.
5430 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5431 Sets all fields of the vCPU runstate data from the .u.runstate member
5432 of the structure, including the current runstate. The state_entry_time
5433 must equal the sum of the other four times.
5435 KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5436 This *adds* the contents of the .u.runstate members of the structure
5437 to the corresponding members of the given vCPU's runstate data, thus
5438 permitting atomic adjustments to the runstate times. The adjustment
5439 to the state_entry_time must equal the sum of the adjustments to the
5440 other four times. The state field must be set to -1, or to a valid
5441 runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5442 or RUNSTATE_offline) to set the current accounted state as of the
5443 adjusted state_entry_time.
5445 KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID
5446 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5447 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen
5448 vCPU ID of the given vCPU, to allow timer-related VCPU operations to
5449 be intercepted by KVM.
5451 KVM_XEN_VCPU_ATTR_TYPE_TIMER
5452 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5453 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5454 event channel port/priority for the VIRQ_TIMER of the vCPU, as well
5455 as allowing a pending timer to be saved/restored.
5457 KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR
5458 This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5459 support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5460 per-vCPU local APIC upcall vector, configured by a Xen guest with
5461 the HVMOP_set_evtchn_upcall_vector hypercall. This is typically
5462 used by Windows guests, and is distinct from the HVM-wide upcall
5463 vector configured with HVM_PARAM_CALLBACK_IRQ.
5466 4.129 KVM_XEN_VCPU_GET_ATTR
5467 ---------------------------
5469 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5472 :Parameters: struct kvm_xen_vcpu_attr
5473 :Returns: 0 on success, < 0 on error
5475 Allows Xen vCPU attributes to be read. For the structure and types,
5476 see KVM_XEN_VCPU_SET_ATTR above.
5478 The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5479 with the KVM_XEN_VCPU_GET_ATTR ioctl.
5481 4.130 KVM_ARM_MTE_COPY_TAGS
5482 ---------------------------
5484 :Capability: KVM_CAP_ARM_MTE
5485 :Architectures: arm64
5487 :Parameters: struct kvm_arm_copy_mte_tags
5488 :Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5489 arguments, -EFAULT if memory cannot be accessed).
5493 struct kvm_arm_copy_mte_tags {
5501 Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5502 ``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr``
5503 field must point to a buffer which the tags will be copied to or from.
5505 ``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5506 ``KVM_ARM_TAGS_FROM_GUEST``.
5508 The size of the buffer to store the tags is ``(length / 16)`` bytes
5509 (granules in MTE are 16 bytes long). Each byte contains a single tag
5510 value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5511 ``PTRACE_POKEMTETAGS``.
5513 If an error occurs before any data is copied then a negative error code is
5514 returned. If some tags have been copied before an error occurs then the number
5515 of bytes successfully copied is returned. If the call completes successfully
5516 then ``length`` is returned.
5518 4.131 KVM_GET_SREGS2
5519 --------------------
5521 :Capability: KVM_CAP_SREGS2
5524 :Parameters: struct kvm_sregs2 (out)
5525 :Returns: 0 on success, -1 on error
5527 Reads special registers from the vcpu.
5528 This ioctl (when supported) replaces the KVM_GET_SREGS.
5533 /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5534 struct kvm_segment cs, ds, es, fs, gs, ss;
5535 struct kvm_segment tr, ldt;
5536 struct kvm_dtable gdt, idt;
5537 __u64 cr0, cr2, cr3, cr4, cr8;
5544 flags values for ``kvm_sregs2``:
5546 ``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5548 Indicates thats the struct contain valid PDPTR values.
5551 4.132 KVM_SET_SREGS2
5552 --------------------
5554 :Capability: KVM_CAP_SREGS2
5557 :Parameters: struct kvm_sregs2 (in)
5558 :Returns: 0 on success, -1 on error
5560 Writes special registers into the vcpu.
5561 See KVM_GET_SREGS2 for the data structures.
5562 This ioctl (when supported) replaces the KVM_SET_SREGS.
5564 4.133 KVM_GET_STATS_FD
5565 ----------------------
5567 :Capability: KVM_CAP_STATS_BINARY_FD
5569 :Type: vm ioctl, vcpu ioctl
5571 :Returns: statistics file descriptor on success, < 0 on error
5575 ====== ======================================================
5576 ENOMEM if the fd could not be created due to lack of memory
5577 EMFILE if the number of opened files exceeds the limit
5578 ====== ======================================================
5580 The returned file descriptor can be used to read VM/vCPU statistics data in
5581 binary format. The data in the file descriptor consists of four blocks
5582 organized as follows:
5594 Apart from the header starting at offset 0, please be aware that it is
5595 not guaranteed that the four blocks are adjacent or in the above order;
5596 the offsets of the id, descriptors and data blocks are found in the
5597 header. However, all four blocks are aligned to 64 bit offsets in the
5598 file and they do not overlap.
5600 All blocks except the data block are immutable. Userspace can read them
5601 only one time after retrieving the file descriptor, and then use ``pread`` or
5602 ``lseek`` to read the statistics repeatedly.
5604 All data is in system endianness.
5606 The format of the header is as follows::
5608 struct kvm_stats_header {
5617 The ``flags`` field is not used at the moment. It is always read as 0.
5619 The ``name_size`` field is the size (in byte) of the statistics name string
5620 (including trailing '\0') which is contained in the "id string" block and
5621 appended at the end of every descriptor.
5623 The ``num_desc`` field is the number of descriptors that are included in the
5624 descriptor block. (The actual number of values in the data block may be
5625 larger, since each descriptor may comprise more than one value).
5627 The ``id_offset`` field is the offset of the id string from the start of the
5628 file indicated by the file descriptor. It is a multiple of 8.
5630 The ``desc_offset`` field is the offset of the Descriptors block from the start
5631 of the file indicated by the file descriptor. It is a multiple of 8.
5633 The ``data_offset`` field is the offset of the Stats Data block from the start
5634 of the file indicated by the file descriptor. It is a multiple of 8.
5636 The id string block contains a string which identifies the file descriptor on
5637 which KVM_GET_STATS_FD was invoked. The size of the block, including the
5638 trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5640 The descriptors block is only needed to be read once for the lifetime of the
5641 file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5642 by a string of size ``name_size``.
5645 #define KVM_STATS_TYPE_SHIFT 0
5646 #define KVM_STATS_TYPE_MASK (0xF << KVM_STATS_TYPE_SHIFT)
5647 #define KVM_STATS_TYPE_CUMULATIVE (0x0 << KVM_STATS_TYPE_SHIFT)
5648 #define KVM_STATS_TYPE_INSTANT (0x1 << KVM_STATS_TYPE_SHIFT)
5649 #define KVM_STATS_TYPE_PEAK (0x2 << KVM_STATS_TYPE_SHIFT)
5650 #define KVM_STATS_TYPE_LINEAR_HIST (0x3 << KVM_STATS_TYPE_SHIFT)
5651 #define KVM_STATS_TYPE_LOG_HIST (0x4 << KVM_STATS_TYPE_SHIFT)
5652 #define KVM_STATS_TYPE_MAX KVM_STATS_TYPE_LOG_HIST
5654 #define KVM_STATS_UNIT_SHIFT 4
5655 #define KVM_STATS_UNIT_MASK (0xF << KVM_STATS_UNIT_SHIFT)
5656 #define KVM_STATS_UNIT_NONE (0x0 << KVM_STATS_UNIT_SHIFT)
5657 #define KVM_STATS_UNIT_BYTES (0x1 << KVM_STATS_UNIT_SHIFT)
5658 #define KVM_STATS_UNIT_SECONDS (0x2 << KVM_STATS_UNIT_SHIFT)
5659 #define KVM_STATS_UNIT_CYCLES (0x3 << KVM_STATS_UNIT_SHIFT)
5660 #define KVM_STATS_UNIT_BOOLEAN (0x4 << KVM_STATS_UNIT_SHIFT)
5661 #define KVM_STATS_UNIT_MAX KVM_STATS_UNIT_BOOLEAN
5663 #define KVM_STATS_BASE_SHIFT 8
5664 #define KVM_STATS_BASE_MASK (0xF << KVM_STATS_BASE_SHIFT)
5665 #define KVM_STATS_BASE_POW10 (0x0 << KVM_STATS_BASE_SHIFT)
5666 #define KVM_STATS_BASE_POW2 (0x1 << KVM_STATS_BASE_SHIFT)
5667 #define KVM_STATS_BASE_MAX KVM_STATS_BASE_POW2
5669 struct kvm_stats_desc {
5678 The ``flags`` field contains the type and unit of the statistics data described
5679 by this descriptor. Its endianness is CPU native.
5680 The following flags are supported:
5682 Bits 0-3 of ``flags`` encode the type:
5684 * ``KVM_STATS_TYPE_CUMULATIVE``
5685 The statistics reports a cumulative count. The value of data can only be increased.
5686 Most of the counters used in KVM are of this type.
5687 The corresponding ``size`` field for this type is always 1.
5688 All cumulative statistics data are read/write.
5689 * ``KVM_STATS_TYPE_INSTANT``
5690 The statistics reports an instantaneous value. Its value can be increased or
5691 decreased. This type is usually used as a measurement of some resources,
5692 like the number of dirty pages, the number of large pages, etc.
5693 All instant statistics are read only.
5694 The corresponding ``size`` field for this type is always 1.
5695 * ``KVM_STATS_TYPE_PEAK``
5696 The statistics data reports a peak value, for example the maximum number
5697 of items in a hash table bucket, the longest time waited and so on.
5698 The value of data can only be increased.
5699 The corresponding ``size`` field for this type is always 1.
5700 * ``KVM_STATS_TYPE_LINEAR_HIST``
5701 The statistic is reported as a linear histogram. The number of
5702 buckets is specified by the ``size`` field. The size of buckets is specified
5703 by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``)
5704 is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last
5705 bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity
5707 * ``KVM_STATS_TYPE_LOG_HIST``
5708 The statistic is reported as a logarithmic histogram. The number of
5709 buckets is specified by the ``size`` field. The range of the first bucket is
5710 [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF).
5711 Otherwise, The Nth bucket (1 < N < ``size``) covers
5712 [pow(2, N-2), pow(2, N-1)).
5714 Bits 4-7 of ``flags`` encode the unit:
5716 * ``KVM_STATS_UNIT_NONE``
5717 There is no unit for the value of statistics data. This usually means that
5718 the value is a simple counter of an event.
5719 * ``KVM_STATS_UNIT_BYTES``
5720 It indicates that the statistics data is used to measure memory size, in the
5721 unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
5722 determined by the ``exponent`` field in the descriptor.
5723 * ``KVM_STATS_UNIT_SECONDS``
5724 It indicates that the statistics data is used to measure time or latency.
5725 * ``KVM_STATS_UNIT_CYCLES``
5726 It indicates that the statistics data is used to measure CPU clock cycles.
5727 * ``KVM_STATS_UNIT_BOOLEAN``
5728 It indicates that the statistic will always be either 0 or 1. Boolean
5729 statistics of "peak" type will never go back from 1 to 0. Boolean
5730 statistics can be linear histograms (with two buckets) but not logarithmic
5733 Note that, in the case of histograms, the unit applies to the bucket
5734 ranges, while the bucket value indicates how many samples fell in the
5737 Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
5740 * ``KVM_STATS_BASE_POW10``
5741 The scale is based on power of 10. It is used for measurement of time and
5742 CPU clock cycles. For example, an exponent of -9 can be used with
5743 ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
5744 * ``KVM_STATS_BASE_POW2``
5745 The scale is based on power of 2. It is used for measurement of memory size.
5746 For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
5747 express that the unit is MiB.
5749 The ``size`` field is the number of values of this statistics data. Its
5750 value is usually 1 for most of simple statistics. 1 means it contains an
5751 unsigned 64bit data.
5753 The ``offset`` field is the offset from the start of Data Block to the start of
5754 the corresponding statistics data.
5756 The ``bucket_size`` field is used as a parameter for histogram statistics data.
5757 It is only used by linear histogram statistics data, specifying the size of a
5758 bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``.
5760 The ``name`` field is the name string of the statistics data. The name string
5761 starts at the end of ``struct kvm_stats_desc``. The maximum length including
5762 the trailing ``'\0'``, is indicated by ``name_size`` in the header.
5764 The Stats Data block contains an array of 64-bit values in the same order
5765 as the descriptors in Descriptors block.
5767 4.134 KVM_GET_XSAVE2
5768 --------------------
5770 :Capability: KVM_CAP_XSAVE2
5773 :Parameters: struct kvm_xsave (out)
5774 :Returns: 0 on success, -1 on error
5784 This ioctl would copy current vcpu's xsave struct to the userspace. It
5785 copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2)
5786 when invoked on the vm file descriptor. The size value returned by
5787 KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
5788 Currently, it is only greater than 4096 if a dynamic feature has been
5789 enabled with ``arch_prctl()``, but this may change in the future.
5791 The offsets of the state save areas in struct kvm_xsave follow the contents
5792 of CPUID leaf 0xD on the host.
5794 4.135 KVM_XEN_HVM_EVTCHN_SEND
5795 -----------------------------
5797 :Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND
5800 :Parameters: struct kvm_irq_routing_xen_evtchn
5801 :Returns: 0 on success, < 0 on error
5806 struct kvm_irq_routing_xen_evtchn {
5812 This ioctl injects an event channel interrupt directly to the guest vCPU.
5814 5. The kvm_run structure
5815 ========================
5817 Application code obtains a pointer to the kvm_run structure by
5818 mmap()ing a vcpu fd. From that point, application code can control
5819 execution by changing fields in kvm_run prior to calling the KVM_RUN
5820 ioctl, and obtain information about the reason KVM_RUN returned by
5821 looking up structure members.
5827 __u8 request_interrupt_window;
5829 Request that KVM_RUN return when it becomes possible to inject external
5830 interrupts into the guest. Useful in conjunction with KVM_INTERRUPT.
5834 __u8 immediate_exit;
5836 This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
5837 exits immediately, returning -EINTR. In the common scenario where a
5838 signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
5839 to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
5840 Rather than blocking the signal outside KVM_RUN, userspace can set up
5841 a signal handler that sets run->immediate_exit to a non-zero value.
5843 This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
5852 When KVM_RUN has returned successfully (return value 0), this informs
5853 application code why KVM_RUN has returned. Allowable values for this
5854 field are detailed below.
5858 __u8 ready_for_interrupt_injection;
5860 If request_interrupt_window has been specified, this field indicates
5861 an interrupt can be injected now with KVM_INTERRUPT.
5867 The value of the current interrupt flag. Only valid if in-kernel
5868 local APIC is not used.
5874 More architecture-specific flags detailing state of the VCPU that may
5875 affect the device's behavior. Current defined flags::
5877 /* x86, set if the VCPU is in system management mode */
5878 #define KVM_RUN_X86_SMM (1 << 0)
5879 /* x86, set if bus lock detected in VM */
5880 #define KVM_RUN_BUS_LOCK (1 << 1)
5881 /* arm64, set for KVM_EXIT_DEBUG */
5882 #define KVM_DEBUG_ARCH_HSR_HIGH_VALID (1 << 0)
5886 /* in (pre_kvm_run), out (post_kvm_run) */
5889 The value of the cr8 register. Only valid if in-kernel local APIC is
5890 not used. Both input and output.
5896 The value of the APIC BASE msr. Only valid if in-kernel local
5897 APIC is not used. Both input and output.
5902 /* KVM_EXIT_UNKNOWN */
5904 __u64 hardware_exit_reason;
5907 If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
5908 reasons. Further architecture-specific information is available in
5909 hardware_exit_reason.
5913 /* KVM_EXIT_FAIL_ENTRY */
5915 __u64 hardware_entry_failure_reason;
5916 __u32 cpu; /* if KVM_LAST_CPU */
5919 If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
5920 to unknown reasons. Further architecture-specific information is
5921 available in hardware_entry_failure_reason.
5925 /* KVM_EXIT_EXCEPTION */
5937 #define KVM_EXIT_IO_IN 0
5938 #define KVM_EXIT_IO_OUT 1
5940 __u8 size; /* bytes */
5943 __u64 data_offset; /* relative to kvm_run start */
5946 If exit_reason is KVM_EXIT_IO, then the vcpu has
5947 executed a port I/O instruction which could not be satisfied by kvm.
5948 data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
5949 where kvm expects application code to place the data for the next
5950 KVM_RUN invocation (KVM_EXIT_IO_IN). Data format is a packed array.
5954 /* KVM_EXIT_DEBUG */
5956 struct kvm_debug_exit_arch arch;
5959 If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
5960 for which architecture specific information is returned.
5972 If exit_reason is KVM_EXIT_MMIO, then the vcpu has
5973 executed a memory-mapped I/O instruction which could not be satisfied
5974 by kvm. The 'data' member contains the written data if 'is_write' is
5975 true, and should be filled by application code otherwise.
5977 The 'data' member contains, in its first 'len' bytes, the value as it would
5978 appear if the VCPU performed a load or store of the appropriate width directly
5983 For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
5984 KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5985 operations are complete (and guest state is consistent) only after userspace
5986 has re-entered the kernel with KVM_RUN. The kernel side will first finish
5987 incomplete operations and then check for pending signals.
5989 The pending state of the operation is not preserved in state which is
5990 visible to userspace, thus userspace should ensure that the operation is
5991 completed before performing a live migration. Userspace can re-enter the
5992 guest with an unmasked signal pending or with the immediate_exit field set
5993 to complete pending operations without allowing any further instructions
5998 /* KVM_EXIT_HYPERCALL */
6007 Unused. This was once used for 'hypercall to userspace'. To implement
6008 such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
6010 .. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
6014 /* KVM_EXIT_TPR_ACCESS */
6021 To be documented (KVM_TPR_ACCESS_REPORTING).
6025 /* KVM_EXIT_S390_SIEIC */
6028 __u64 mask; /* psw upper half */
6029 __u64 addr; /* psw lower half */
6038 /* KVM_EXIT_S390_RESET */
6039 #define KVM_S390_RESET_POR 1
6040 #define KVM_S390_RESET_CLEAR 2
6041 #define KVM_S390_RESET_SUBSYSTEM 4
6042 #define KVM_S390_RESET_CPU_INIT 8
6043 #define KVM_S390_RESET_IPL 16
6044 __u64 s390_reset_flags;
6050 /* KVM_EXIT_S390_UCONTROL */
6052 __u64 trans_exc_code;
6056 s390 specific. A page fault has occurred for a user controlled virtual
6057 machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
6058 resolved by the kernel.
6059 The program code and the translation exception code that were placed
6060 in the cpu's lowcore are presented here as defined by the z Architecture
6061 Principles of Operation Book in the Chapter for Dynamic Address Translation
6073 Deprecated - was used for 440 KVM.
6082 MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
6083 hypercalls and exit with this exit struct that contains all the guest gprs.
6085 If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
6086 Userspace can now handle the hypercall and when it's done modify the gprs as
6087 necessary. Upon guest entry all guest GPRs will then be replaced by the values
6092 /* KVM_EXIT_PAPR_HCALL */
6099 This is used on 64-bit PowerPC when emulating a pSeries partition,
6100 e.g. with the 'pseries' machine type in qemu. It occurs when the
6101 guest does a hypercall using the 'sc 1' instruction. The 'nr' field
6102 contains the hypercall number (from the guest R3), and 'args' contains
6103 the arguments (from the guest R4 - R12). Userspace should put the
6104 return code in 'ret' and any extra returned values in args[].
6105 The possible hypercalls are defined in the Power Architecture Platform
6106 Requirements (PAPR) document available from www.power.org (free
6107 developer registration required to access it).
6111 /* KVM_EXIT_S390_TSCH */
6113 __u16 subchannel_id;
6114 __u16 subchannel_nr;
6121 s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
6122 and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
6123 interrupt for the target subchannel has been dequeued and subchannel_id,
6124 subchannel_nr, io_int_parm and io_int_word contain the parameters for that
6125 interrupt. ipb is needed for instruction parameter decoding.
6134 On FSL BookE PowerPC chips, the interrupt controller has a fast patch
6135 interrupt acknowledge path to the core. When the core successfully
6136 delivers an interrupt, it automatically populates the EPR register with
6137 the interrupt vector number and acknowledges the interrupt inside
6138 the interrupt controller.
6140 In case the interrupt controller lives in user space, we need to do
6141 the interrupt acknowledge cycle through it to fetch the next to be
6142 delivered interrupt vector using this exit.
6144 It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
6145 external interrupt has just been delivered into the guest. User space
6146 should put the acknowledged interrupt vector into the 'epr' field.
6150 /* KVM_EXIT_SYSTEM_EVENT */
6152 #define KVM_SYSTEM_EVENT_SHUTDOWN 1
6153 #define KVM_SYSTEM_EVENT_RESET 2
6154 #define KVM_SYSTEM_EVENT_CRASH 3
6155 #define KVM_SYSTEM_EVENT_WAKEUP 4
6156 #define KVM_SYSTEM_EVENT_SUSPEND 5
6157 #define KVM_SYSTEM_EVENT_SEV_TERM 6
6163 If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
6164 a system-level event using some architecture specific mechanism (hypercall
6165 or some special instruction). In case of ARM64, this is triggered using
6166 HVC instruction based PSCI call from the vcpu.
6168 The 'type' field describes the system-level event type.
6169 Valid values for 'type' are:
6171 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
6172 VM. Userspace is not obliged to honour this, and if it does honour
6173 this does not need to destroy the VM synchronously (ie it may call
6174 KVM_RUN again before shutdown finally occurs).
6175 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
6176 As with SHUTDOWN, userspace can choose to ignore the request, or
6177 to schedule the reset to occur in the future and may call KVM_RUN again.
6178 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
6179 has requested a crash condition maintenance. Userspace can choose
6180 to ignore the request, or to gather VM memory core dump and/or
6181 reset/shutdown of the VM.
6182 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination.
6183 The guest physical address of the guest's GHCB is stored in `data[0]`.
6184 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and
6185 KVM has recognized a wakeup event. Userspace may honor this event by
6186 marking the exiting vCPU as runnable, or deny it and call KVM_RUN again.
6187 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of
6190 If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain
6191 architecture specific information for the system-level event. Only
6192 the first `ndata` items (possibly zero) of the data array are valid.
6194 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if
6195 the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI
6198 - for RISC-V, data[0] is set to the value of the second argument of the
6199 ``sbi_system_reset`` call.
6201 Previous versions of Linux defined a `flags` member in this struct. The
6202 field is now aliased to `data[0]`. Userspace can assume that it is only
6203 written if ndata is greater than 0.
6208 KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the
6209 KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI
6210 SYSTEM_SUSPEND function, KVM will exit to userspace with this event
6213 It is the sole responsibility of userspace to implement the PSCI
6214 SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND".
6215 KVM does not change the vCPU's state before exiting to userspace, so
6216 the call parameters are left in-place in the vCPU registers.
6218 Userspace is _required_ to take action for such an exit. It must
6221 - Honor the guest request to suspend the VM. Userspace can request
6222 in-kernel emulation of suspension by setting the calling vCPU's
6223 state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's
6224 state according to the parameters passed to the PSCI function when
6225 the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use"
6226 for details on the function parameters.
6228 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2
6229 "Caller responsibilities" for possible return values.
6233 /* KVM_EXIT_IOAPIC_EOI */
6238 Indicates that the VCPU's in-kernel local APIC received an EOI for a
6239 level-triggered IOAPIC interrupt. This exit only triggers when the
6240 IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
6241 the userspace IOAPIC should process the EOI and retrigger the interrupt if
6242 it is still asserted. Vector is the LAPIC interrupt vector for which the
6247 struct kvm_hyperv_exit {
6248 #define KVM_EXIT_HYPERV_SYNIC 1
6249 #define KVM_EXIT_HYPERV_HCALL 2
6250 #define KVM_EXIT_HYPERV_SYNDBG 3
6277 /* KVM_EXIT_HYPERV */
6278 struct kvm_hyperv_exit hyperv;
6280 Indicates that the VCPU exits into userspace to process some tasks
6281 related to Hyper-V emulation.
6283 Valid values for 'type' are:
6285 - KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
6287 Hyper-V SynIC state change. Notification is used to remap SynIC
6288 event/message pages and to enable/disable SynIC messages/events processing
6291 - KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
6293 Hyper-V Synthetic debugger state change. Notification is used to either update
6294 the pending_page location or to send a control command (send the buffer located
6295 in send_page or recv a buffer to recv_page).
6299 /* KVM_EXIT_ARM_NISV */
6305 Used on arm64 systems. If a guest accesses memory not in a memslot,
6306 KVM will typically return to userspace and ask it to do MMIO emulation on its
6307 behalf. However, for certain classes of instructions, no instruction decode
6308 (direction, length of memory access) is provided, and fetching and decoding
6309 the instruction from the VM is overly complicated to live in the kernel.
6311 Historically, when this situation occurred, KVM would print a warning and kill
6312 the VM. KVM assumed that if the guest accessed non-memslot memory, it was
6313 trying to do I/O, which just couldn't be emulated, and the warning message was
6314 phrased accordingly. However, what happened more often was that a guest bug
6315 caused access outside the guest memory areas which should lead to a more
6316 meaningful warning message and an external abort in the guest, if the access
6317 did not fall within an I/O window.
6319 Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
6320 this capability at VM creation. Once this is done, these types of errors will
6321 instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
6322 the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field.
6323 Userspace can either fix up the access if it's actually an I/O access by
6324 decoding the instruction from guest memory (if it's very brave) and continue
6325 executing the guest, or it can decide to suspend, dump, or restart the guest.
6327 Note that KVM does not skip the faulting instruction as it does for
6328 KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
6329 if it decides to decode and emulate the instruction.
6333 /* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
6335 __u8 error; /* user -> kernel */
6337 __u32 reason; /* kernel -> user */
6338 __u32 index; /* kernel -> user */
6339 __u64 data; /* kernel <-> user */
6342 Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
6343 enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
6344 will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
6347 The "reason" field specifies why the MSR trap occurred. User space will only
6348 receive MSR exit traps when a particular reason was requested during through
6349 ENABLE_CAP. Currently valid exit reasons are:
6351 KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
6352 KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
6353 KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
6355 For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
6356 wants to read. To respond to this request with a successful read, user space
6357 writes the respective data into the "data" field and must continue guest
6358 execution to ensure the read data is transferred into guest register state.
6360 If the RDMSR request was unsuccessful, user space indicates that with a "1" in
6361 the "error" field. This will inject a #GP into the guest when the VCPU is
6364 For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
6365 wants to write. Once finished processing the event, user space must continue
6366 vCPU execution. If the MSR write was unsuccessful, user space also sets the
6367 "error" field to "1".
6372 struct kvm_xen_exit {
6373 #define KVM_EXIT_XEN_HCALL 1
6386 struct kvm_hyperv_exit xen;
6388 Indicates that the VCPU exits into userspace to process some tasks
6389 related to Xen emulation.
6391 Valid values for 'type' are:
6393 - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
6394 Userspace is expected to place the hypercall result into the appropriate
6395 field before invoking KVM_RUN again.
6399 /* KVM_EXIT_RISCV_SBI */
6401 unsigned long extension_id;
6402 unsigned long function_id;
6403 unsigned long args[6];
6404 unsigned long ret[2];
6407 If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has
6408 done a SBI call which is not handled by KVM RISC-V kernel module. The details
6409 of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The
6410 'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the
6411 'function_id' field represents function ID of given SBI extension. The 'args'
6412 array field of 'riscv_sbi' represents parameters for the SBI call and 'ret'
6413 array field represents return values. The userspace should update the return
6414 values of SBI call before resuming the VCPU. For more details on RISC-V SBI
6415 spec refer, https://github.com/riscv/riscv-sbi-doc.
6419 /* Fix the size of the union. */
6424 * shared registers between kvm and userspace.
6425 * kvm_valid_regs specifies the register classes set by the host
6426 * kvm_dirty_regs specified the register classes dirtied by userspace
6427 * struct kvm_sync_regs is architecture specific, as well as the
6428 * bits for kvm_valid_regs and kvm_dirty_regs
6430 __u64 kvm_valid_regs;
6431 __u64 kvm_dirty_regs;
6433 struct kvm_sync_regs regs;
6434 char padding[SYNC_REGS_SIZE_BYTES];
6437 If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
6438 certain guest registers without having to call SET/GET_*REGS. Thus we can
6439 avoid some system call overhead if userspace has to handle the exit.
6440 Userspace can query the validity of the structure by checking
6441 kvm_valid_regs for specific bits. These bits are architecture specific
6442 and usually define the validity of a groups of registers. (e.g. one bit
6443 for general purpose registers)
6445 Please note that the kernel is allowed to use the kvm_run structure as the
6446 primary storage for certain register types. Therefore, the kernel may use the
6447 values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
6455 6. Capabilities that can be enabled on vCPUs
6456 ============================================
6458 There are certain capabilities that change the behavior of the virtual CPU or
6459 the virtual machine when enabled. To enable them, please see section 4.37.
6460 Below you can find a list of capabilities and what their effect on the vCPU or
6461 the virtual machine is when enabling them.
6463 The following information is provided along with the description:
6466 which instruction set architectures provide this ioctl.
6467 x86 includes both i386 and x86_64.
6470 whether this is a per-vcpu or per-vm capability.
6473 what parameters are accepted by the capability.
6476 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6477 are not detailed, but errors with specific meanings are.
6486 :Returns: 0 on success; -1 on error
6488 This capability enables interception of OSI hypercalls that otherwise would
6489 be treated as normal system calls to be injected into the guest. OSI hypercalls
6490 were invented by Mac-on-Linux to have a standardized communication mechanism
6491 between the guest and the host.
6493 When this capability is enabled, KVM_EXIT_OSI can occur.
6496 6.2 KVM_CAP_PPC_PAPR
6497 --------------------
6502 :Returns: 0 on success; -1 on error
6504 This capability enables interception of PAPR hypercalls. PAPR hypercalls are
6505 done using the hypercall instruction "sc 1".
6507 It also sets the guest privilege level to "supervisor" mode. Usually the guest
6508 runs in "hypervisor" privilege mode with a few missing features.
6510 In addition to the above, it changes the semantics of SDR1. In this mode, the
6511 HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
6512 HTAB invisible to the guest.
6514 When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
6522 :Parameters: args[0] is the address of a struct kvm_config_tlb
6523 :Returns: 0 on success; -1 on error
6527 struct kvm_config_tlb {
6534 Configures the virtual CPU's TLB array, establishing a shared memory area
6535 between userspace and KVM. The "params" and "array" fields are userspace
6536 addresses of mmu-type-specific data structures. The "array_len" field is an
6537 safety mechanism, and should be set to the size in bytes of the memory that
6538 userspace has reserved for the array. It must be at least the size dictated
6539 by "mmu_type" and "params".
6541 While KVM_RUN is active, the shared region is under control of KVM. Its
6542 contents are undefined, and any modification by userspace results in
6543 boundedly undefined behavior.
6545 On return from KVM_RUN, the shared region will reflect the current state of
6546 the guest's TLB. If userspace makes any changes, it must call KVM_DIRTY_TLB
6547 to tell KVM which entries have been changed, prior to calling KVM_RUN again
6550 For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
6552 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
6553 - The "array" field points to an array of type "struct
6554 kvm_book3e_206_tlb_entry".
6555 - The array consists of all entries in the first TLB, followed by all
6556 entries in the second TLB.
6557 - Within a TLB, entries are ordered first by increasing set number. Within a
6558 set, entries are ordered by way (increasing ESEL).
6559 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
6560 where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
6561 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
6562 hardware ignores this value for TLB0.
6564 6.4 KVM_CAP_S390_CSS_SUPPORT
6565 ----------------------------
6567 :Architectures: s390
6570 :Returns: 0 on success; -1 on error
6572 This capability enables support for handling of channel I/O instructions.
6574 TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
6575 handled in-kernel, while the other I/O instructions are passed to userspace.
6577 When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
6578 SUBCHANNEL intercepts.
6580 Note that even though this capability is enabled per-vcpu, the complete
6581 virtual machine is affected.
6588 :Parameters: args[0] defines whether the proxy facility is active
6589 :Returns: 0 on success; -1 on error
6591 This capability enables or disables the delivery of interrupts through the
6592 external proxy facility.
6594 When enabled (args[0] != 0), every time the guest gets an external interrupt
6595 delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
6596 to receive the topmost interrupt vector.
6598 When disabled (args[0] == 0), behavior is as if this facility is unsupported.
6600 When this capability is enabled, KVM_EXIT_EPR can occur.
6602 6.6 KVM_CAP_IRQ_MPIC
6603 --------------------
6606 :Parameters: args[0] is the MPIC device fd;
6607 args[1] is the MPIC CPU number for this vcpu
6609 This capability connects the vcpu to an in-kernel MPIC device.
6611 6.7 KVM_CAP_IRQ_XICS
6612 --------------------
6616 :Parameters: args[0] is the XICS device fd;
6617 args[1] is the XICS CPU number (server ID) for this vcpu
6619 This capability connects the vcpu to an in-kernel XICS device.
6621 6.8 KVM_CAP_S390_IRQCHIP
6622 ------------------------
6624 :Architectures: s390
6628 This capability enables the in-kernel irqchip for s390. Please refer to
6629 "4.24 KVM_CREATE_IRQCHIP" for details.
6631 6.9 KVM_CAP_MIPS_FPU
6632 --------------------
6634 :Architectures: mips
6636 :Parameters: args[0] is reserved for future use (should be 0).
6638 This capability allows the use of the host Floating Point Unit by the guest. It
6639 allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
6640 done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
6641 accessed (depending on the current guest FPU register mode), and the Status.FR,
6642 Config5.FRE bits are accessible via the KVM API and also from the guest,
6643 depending on them being supported by the FPU.
6645 6.10 KVM_CAP_MIPS_MSA
6646 ---------------------
6648 :Architectures: mips
6650 :Parameters: args[0] is reserved for future use (should be 0).
6652 This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
6653 It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
6654 Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
6655 registers can be accessed, and the Config5.MSAEn bit is accessible via the
6656 KVM API and also from the guest.
6658 6.74 KVM_CAP_SYNC_REGS
6659 ----------------------
6661 :Architectures: s390, x86
6662 :Target: s390: always enabled, x86: vcpu
6664 :Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
6666 (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
6668 As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
6669 KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
6670 without having to call SET/GET_*REGS". This reduces overhead by eliminating
6671 repeated ioctl calls for setting and/or getting register values. This is
6672 particularly important when userspace is making synchronous guest state
6673 modifications, e.g. when emulating and/or intercepting instructions in
6676 For s390 specifics, please refer to the source code.
6680 - the register sets to be copied out to kvm_run are selectable
6681 by userspace (rather that all sets being copied out for every exit).
6682 - vcpu_events are available in addition to regs and sregs.
6684 For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
6685 function as an input bit-array field set by userspace to indicate the
6686 specific register sets to be copied out on the next exit.
6688 To indicate when userspace has modified values that should be copied into
6689 the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
6690 This is done using the same bitflags as for the 'kvm_valid_regs' field.
6691 If the dirty bit is not set, then the register set values will not be copied
6692 into the vCPU even if they've been modified.
6694 Unused bitfields in the bitarrays must be set to zero.
6698 struct kvm_sync_regs {
6699 struct kvm_regs regs;
6700 struct kvm_sregs sregs;
6701 struct kvm_vcpu_events events;
6704 6.75 KVM_CAP_PPC_IRQ_XIVE
6705 -------------------------
6709 :Parameters: args[0] is the XIVE device fd;
6710 args[1] is the XIVE CPU number (server ID) for this vcpu
6712 This capability connects the vcpu to an in-kernel XIVE device.
6714 7. Capabilities that can be enabled on VMs
6715 ==========================================
6717 There are certain capabilities that change the behavior of the virtual
6718 machine when enabled. To enable them, please see section 4.37. Below
6719 you can find a list of capabilities and what their effect on the VM
6720 is when enabling them.
6722 The following information is provided along with the description:
6725 which instruction set architectures provide this ioctl.
6726 x86 includes both i386 and x86_64.
6729 what parameters are accepted by the capability.
6732 the return value. General error numbers (EBADF, ENOMEM, EINVAL)
6733 are not detailed, but errors with specific meanings are.
6736 7.1 KVM_CAP_PPC_ENABLE_HCALL
6737 ----------------------------
6740 :Parameters: args[0] is the sPAPR hcall number;
6741 args[1] is 0 to disable, 1 to enable in-kernel handling
6743 This capability controls whether individual sPAPR hypercalls (hcalls)
6744 get handled by the kernel or not. Enabling or disabling in-kernel
6745 handling of an hcall is effective across the VM. On creation, an
6746 initial set of hcalls are enabled for in-kernel handling, which
6747 consists of those hcalls for which in-kernel handlers were implemented
6748 before this capability was implemented. If disabled, the kernel will
6749 not to attempt to handle the hcall, but will always exit to userspace
6750 to handle it. Note that it may not make sense to enable some and
6751 disable others of a group of related hcalls, but KVM does not prevent
6752 userspace from doing that.
6754 If the hcall number specified is not one that has an in-kernel
6755 implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
6758 7.2 KVM_CAP_S390_USER_SIGP
6759 --------------------------
6761 :Architectures: s390
6764 This capability controls which SIGP orders will be handled completely in user
6765 space. With this capability enabled, all fast orders will be handled completely
6772 - CONDITIONAL EMERGENCY SIGNAL
6774 All other orders will be handled completely in user space.
6776 Only privileged operation exceptions will be checked for in the kernel (or even
6777 in the hardware prior to interception). If this capability is not enabled, the
6778 old way of handling SIGP orders is used (partially in kernel and user space).
6780 7.3 KVM_CAP_S390_VECTOR_REGISTERS
6781 ---------------------------------
6783 :Architectures: s390
6785 :Returns: 0 on success, negative value on error
6787 Allows use of the vector registers introduced with z13 processor, and
6788 provides for the synchronization between host and user space. Will
6789 return -EINVAL if the machine does not support vectors.
6791 7.4 KVM_CAP_S390_USER_STSI
6792 --------------------------
6794 :Architectures: s390
6797 This capability allows post-handlers for the STSI instruction. After
6798 initial handling in the kernel, KVM exits to user space with
6799 KVM_EXIT_S390_STSI to allow user space to insert further data.
6801 Before exiting to userspace, kvm handlers should fill in s390_stsi field of
6813 @addr - guest address of STSI SYSIB
6817 @ar - access register number
6819 KVM handlers should exit to userspace with rc = -EREMOTE.
6821 7.5 KVM_CAP_SPLIT_IRQCHIP
6822 -------------------------
6825 :Parameters: args[0] - number of routes reserved for userspace IOAPICs
6826 :Returns: 0 on success, -1 on error
6828 Create a local apic for each processor in the kernel. This can be used
6829 instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
6830 IOAPIC and PIC (and also the PIT, even though this has to be enabled
6833 This capability also enables in kernel routing of interrupt requests;
6834 when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
6835 used in the IRQ routing table. The first args[0] MSI routes are reserved
6836 for the IOAPIC pins. Whenever the LAPIC receives an EOI for these routes,
6837 a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
6839 Fails if VCPU has already been created, or if the irqchip is already in the
6840 kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
6845 :Architectures: s390
6848 Allows use of runtime-instrumentation introduced with zEC12 processor.
6849 Will return -EINVAL if the machine does not support runtime-instrumentation.
6850 Will return -EBUSY if a VCPU has already been created.
6852 7.7 KVM_CAP_X2APIC_API
6853 ----------------------
6856 :Parameters: args[0] - features that should be enabled
6857 :Returns: 0 on success, -EINVAL when args[0] contains invalid features
6859 Valid feature flags in args[0] are::
6861 #define KVM_X2APIC_API_USE_32BIT_IDS (1ULL << 0)
6862 #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK (1ULL << 1)
6864 Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
6865 KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
6866 allowing the use of 32-bit APIC IDs. See KVM_CAP_X2APIC_API in their
6867 respective sections.
6869 KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
6870 in logical mode or with more than 255 VCPUs. Otherwise, KVM treats 0xff
6871 as a broadcast even in x2APIC mode in order to support physical x2APIC
6872 without interrupt remapping. This is undesirable in logical mode,
6873 where 0xff represents CPUs 0-7 in cluster 0.
6875 7.8 KVM_CAP_S390_USER_INSTR0
6876 ----------------------------
6878 :Architectures: s390
6881 With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
6882 be intercepted and forwarded to user space. User space can use this
6883 mechanism e.g. to realize 2-byte software breakpoints. The kernel will
6884 not inject an operating exception for these instructions, user space has
6885 to take care of that.
6887 This capability can be enabled dynamically even if VCPUs were already
6888 created and are running.
6893 :Architectures: s390
6895 :Returns: 0 on success; -EINVAL if the machine does not support
6896 guarded storage; -EBUSY if a VCPU has already been created.
6898 Allows use of guarded storage for the KVM guest.
6900 7.10 KVM_CAP_S390_AIS
6901 ---------------------
6903 :Architectures: s390
6906 Allow use of adapter-interruption suppression.
6907 :Returns: 0 on success; -EBUSY if a VCPU has already been created.
6909 7.11 KVM_CAP_PPC_SMT
6910 --------------------
6913 :Parameters: vsmt_mode, flags
6915 Enabling this capability on a VM provides userspace with a way to set
6916 the desired virtual SMT mode (i.e. the number of virtual CPUs per
6917 virtual core). The virtual SMT mode, vsmt_mode, must be a power of 2
6918 between 1 and 8. On POWER8, vsmt_mode must also be no greater than
6919 the number of threads per subcore for the host. Currently flags must
6920 be 0. A successful call to enable this capability will result in
6921 vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
6922 subsequently queried for the VM. This capability is only supported by
6923 HV KVM, and can only be set before any VCPUs have been created.
6924 The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
6925 modes are available.
6927 7.12 KVM_CAP_PPC_FWNMI
6928 ----------------------
6933 With this capability a machine check exception in the guest address
6934 space will cause KVM to exit the guest with NMI exit reason. This
6935 enables QEMU to build error log and branch to guest kernel registered
6936 machine check handling routine. Without this capability KVM will
6937 branch to guests' 0x200 interrupt vector.
6939 7.13 KVM_CAP_X86_DISABLE_EXITS
6940 ------------------------------
6943 :Parameters: args[0] defines which exits are disabled
6944 :Returns: 0 on success, -EINVAL when args[0] contains invalid exits
6946 Valid bits in args[0] are::
6948 #define KVM_X86_DISABLE_EXITS_MWAIT (1 << 0)
6949 #define KVM_X86_DISABLE_EXITS_HLT (1 << 1)
6950 #define KVM_X86_DISABLE_EXITS_PAUSE (1 << 2)
6951 #define KVM_X86_DISABLE_EXITS_CSTATE (1 << 3)
6953 Enabling this capability on a VM provides userspace with a way to no
6954 longer intercept some instructions for improved latency in some
6955 workloads, and is suggested when vCPUs are associated to dedicated
6956 physical CPUs. More bits can be added in the future; userspace can
6957 just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
6960 Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
6962 7.14 KVM_CAP_S390_HPAGE_1M
6963 --------------------------
6965 :Architectures: s390
6967 :Returns: 0 on success, -EINVAL if hpage module parameter was not set
6968 or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
6971 With this capability the KVM support for memory backing with 1m pages
6972 through hugetlbfs can be enabled for a VM. After the capability is
6973 enabled, cmma can't be enabled anymore and pfmfi and the storage key
6974 interpretation are disabled. If cmma has already been enabled or the
6975 hpage module parameter is not set to 1, -EINVAL is returned.
6977 While it is generally possible to create a huge page backed VM without
6978 this capability, the VM will not be able to run.
6980 7.15 KVM_CAP_MSR_PLATFORM_INFO
6981 ------------------------------
6984 :Parameters: args[0] whether feature should be enabled or not
6986 With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
6987 a #GP would be raised when the guest tries to access. Currently, this
6988 capability does not enable write permissions of this MSR for the guest.
6990 7.16 KVM_CAP_PPC_NESTED_HV
6991 --------------------------
6995 :Returns: 0 on success, -EINVAL when the implementation doesn't support
6996 nested-HV virtualization.
6998 HV-KVM on POWER9 and later systems allows for "nested-HV"
6999 virtualization, which provides a way for a guest VM to run guests that
7000 can run using the CPU's supervisor mode (privileged non-hypervisor
7001 state). Enabling this capability on a VM depends on the CPU having
7002 the necessary functionality and on the facility being enabled with a
7003 kvm-hv module parameter.
7005 7.17 KVM_CAP_EXCEPTION_PAYLOAD
7006 ------------------------------
7009 :Parameters: args[0] whether feature should be enabled or not
7011 With this capability enabled, CR2 will not be modified prior to the
7012 emulated VM-exit when L1 intercepts a #PF exception that occurs in
7013 L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
7014 the emulated VM-exit when L1 intercepts a #DB exception that occurs in
7015 L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
7016 #DB) exception for L2, exception.has_payload will be set and the
7017 faulting address (or the new DR6 bits*) will be reported in the
7018 exception_payload field. Similarly, when userspace injects a #PF (or
7019 #DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
7020 exception.has_payload and to put the faulting address - or the new DR6
7021 bits\ [#]_ - in the exception_payload field.
7023 This capability also enables exception.pending in struct
7024 kvm_vcpu_events, which allows userspace to distinguish between pending
7025 and injected exceptions.
7028 .. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
7031 7.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
7033 :Architectures: x86, arm64, mips
7034 :Parameters: args[0] whether feature should be enabled or not
7038 #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (1 << 0)
7039 #define KVM_DIRTY_LOG_INITIALLY_SET (1 << 1)
7041 With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
7042 automatically clear and write-protect all pages that are returned as dirty.
7043 Rather, userspace will have to do this operation separately using
7044 KVM_CLEAR_DIRTY_LOG.
7046 At the cost of a slightly more complicated operation, this provides better
7047 scalability and responsiveness for two reasons. First,
7048 KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
7049 than requiring to sync a full memslot; this ensures that KVM does not
7050 take spinlocks for an extended period of time. Second, in some cases a
7051 large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
7052 userspace actually using the data in the page. Pages can be modified
7053 during this time, which is inefficient for both the guest and userspace:
7054 the guest will incur a higher penalty due to write protection faults,
7055 while userspace can see false reports of dirty pages. Manual reprotection
7056 helps reducing this time, improving guest performance and reducing the
7057 number of dirty log false positives.
7059 With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
7060 will be initialized to 1 when created. This also improves performance because
7061 dirty logging can be enabled gradually in small chunks on the first call
7062 to KVM_CLEAR_DIRTY_LOG. KVM_DIRTY_LOG_INITIALLY_SET depends on
7063 KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
7064 x86 and arm64 for now).
7066 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
7067 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
7068 it hard or impossible to use it correctly. The availability of
7069 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
7070 Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
7072 7.19 KVM_CAP_PPC_SECURE_GUEST
7073 ------------------------------
7077 This capability indicates that KVM is running on a host that has
7078 ultravisor firmware and thus can support a secure guest. On such a
7079 system, a guest can ask the ultravisor to make it a secure guest,
7080 one whose memory is inaccessible to the host except for pages which
7081 are explicitly requested to be shared with the host. The ultravisor
7082 notifies KVM when a guest requests to become a secure guest, and KVM
7083 has the opportunity to veto the transition.
7085 If present, this capability can be enabled for a VM, meaning that KVM
7086 will allow the transition to secure guest mode. Otherwise KVM will
7087 veto the transition.
7089 7.20 KVM_CAP_HALT_POLL
7090 ----------------------
7094 :Parameters: args[0] is the maximum poll time in nanoseconds
7095 :Returns: 0 on success; -1 on error
7097 This capability overrides the kvm module parameter halt_poll_ns for the
7100 VCPU polling allows a VCPU to poll for wakeup events instead of immediately
7101 scheduling during guest halts. The maximum time a VCPU can spend polling is
7102 controlled by the kvm module parameter halt_poll_ns. This capability allows
7103 the maximum halt time to specified on a per-VM basis, effectively overriding
7104 the module parameter for the target VM.
7106 7.21 KVM_CAP_X86_USER_SPACE_MSR
7107 -------------------------------
7111 :Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
7112 :Returns: 0 on success; -1 on error
7114 This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
7117 When a guest requests to read or write an MSR, KVM may not implement all MSRs
7118 that are relevant to a respective system. It also does not differentiate by
7121 To allow more fine grained control over MSR handling, user space may enable
7122 this capability. With it enabled, MSR accesses that match the mask specified in
7123 args[0] and trigger a #GP event inside the guest by KVM will instead trigger
7124 KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
7125 can then handle to implement model specific MSR handling and/or user notifications
7126 to inform a user that an MSR was not handled.
7128 7.22 KVM_CAP_X86_BUS_LOCK_EXIT
7129 -------------------------------
7133 :Parameters: args[0] defines the policy used when bus locks detected in guest
7134 :Returns: 0 on success, -EINVAL when args[0] contains invalid bits
7136 Valid bits in args[0] are::
7138 #define KVM_BUS_LOCK_DETECTION_OFF (1 << 0)
7139 #define KVM_BUS_LOCK_DETECTION_EXIT (1 << 1)
7141 Enabling this capability on a VM provides userspace with a way to select
7142 a policy to handle the bus locks detected in guest. Userspace can obtain
7143 the supported modes from the result of KVM_CHECK_EXTENSION and define it
7144 through the KVM_ENABLE_CAP.
7146 KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
7147 currently and mutually exclusive with each other. More bits can be added in
7150 With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
7151 so that no additional actions are needed. This is the default mode.
7153 With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
7154 in VM. KVM just exits to userspace when handling them. Userspace can enforce
7155 its own throttling or other policy based mitigations.
7157 This capability is aimed to address the thread that VM can exploit bus locks to
7158 degree the performance of the whole system. Once the userspace enable this
7159 capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
7160 KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
7161 the bus lock vm exit can be preempted by a higher priority VM exit, the exit
7162 notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
7163 KVM_RUN_BUS_LOCK flag is used to distinguish between them.
7165 7.23 KVM_CAP_PPC_DAWR1
7166 ----------------------
7170 :Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
7172 This capability can be used to check / enable 2nd DAWR feature provided
7173 by POWER10 processor.
7176 7.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
7177 -------------------------------------
7179 Architectures: x86 SEV enabled
7181 Parameters: args[0] is the fd of the source vm
7182 Returns: 0 on success; ENOTTY on error
7184 This capability enables userspace to copy encryption context from the vm
7185 indicated by the fd to the vm this is called on.
7187 This is intended to support in-guest workloads scheduled by the host. This
7188 allows the in-guest workload to maintain its own NPTs and keeps the two vms
7189 from accidentally clobbering each other with interrupts and the like (separate
7192 7.25 KVM_CAP_SGX_ATTRIBUTE
7193 --------------------------
7197 :Parameters: args[0] is a file handle of a SGX attribute file in securityfs
7198 :Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
7199 attribute is not supported by KVM.
7201 KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
7202 more priveleged enclave attributes. args[0] must hold a file handle to a valid
7203 SGX attribute file corresponding to an attribute that is supported/restricted
7204 by KVM (currently only PROVISIONKEY).
7206 The SGX subsystem restricts access to a subset of enclave attributes to provide
7207 additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
7208 is restricted to deter malware from using the PROVISIONKEY to obtain a stable
7209 system fingerprint. To prevent userspace from circumventing such restrictions
7210 by running an enclave in a VM, KVM prevents access to privileged attributes by
7213 See Documentation/x86/sgx.rst for more details.
7215 7.26 KVM_CAP_PPC_RPT_INVALIDATE
7216 -------------------------------
7218 :Capability: KVM_CAP_PPC_RPT_INVALIDATE
7222 This capability indicates that the kernel is capable of handling
7223 H_RPT_INVALIDATE hcall.
7225 In order to enable the use of H_RPT_INVALIDATE in the guest,
7226 user space might have to advertise it for the guest. For example,
7227 IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
7228 present in the "ibm,hypertas-functions" device-tree property.
7230 This capability is enabled for hypervisors on platforms like POWER9
7231 that support radix MMU.
7233 7.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
7234 --------------------------------------
7237 :Parameters: args[0] whether the feature should be enabled or not
7239 When this capability is enabled, an emulation failure will result in an exit
7240 to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
7241 to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
7242 to 15 instruction bytes for any exit to userspace resulting from an emulation
7243 failure. When these exits to userspace occur use the emulation_failure struct
7244 instead of the internal struct. They both have the same layout, but the
7245 emulation_failure struct matches the content better. It also explicitly
7246 defines the 'flags' field which is used to describe the fields in the struct
7247 that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
7248 set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
7251 7.28 KVM_CAP_ARM_MTE
7252 --------------------
7254 :Architectures: arm64
7257 This capability indicates that KVM (and the hardware) supports exposing the
7258 Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
7259 VMM before creating any VCPUs to allow the guest access. Note that MTE is only
7260 available to a guest running in AArch64 mode and enabling this capability will
7261 cause attempts to create AArch32 VCPUs to fail.
7263 When enabled the guest is able to access tags associated with any memory given
7264 to the guest. KVM will ensure that the tags are maintained during swap or
7265 hibernation of the host; however the VMM needs to manually save/restore the
7266 tags as appropriate if the VM is migrated.
7268 When this capability is enabled all memory in memslots must be mapped as
7269 not-shareable (no MAP_SHARED), attempts to create a memslot with a
7270 MAP_SHARED mmap will result in an -EINVAL return.
7272 When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
7273 perform a bulk copy of tags to/from the guest.
7275 7.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM
7276 -------------------------------------
7278 Architectures: x86 SEV enabled
7280 Parameters: args[0] is the fd of the source vm
7281 Returns: 0 on success
7283 This capability enables userspace to migrate the encryption context from the VM
7284 indicated by the fd to the VM this is called on.
7286 This is intended to support intra-host migration of VMs between userspace VMMs,
7287 upgrading the VMM process without interrupting the guest.
7289 7.30 KVM_CAP_PPC_AIL_MODE_3
7290 -------------------------------
7292 :Capability: KVM_CAP_PPC_AIL_MODE_3
7296 This capability indicates that the kernel supports the mode 3 setting for the
7297 "Address Translation Mode on Interrupt" aka "Alternate Interrupt Location"
7298 resource that is controlled with the H_SET_MODE hypercall.
7300 This capability allows a guest kernel to use a better-performance mode for
7301 handling interrupts and system calls.
7303 7.31 KVM_CAP_DISABLE_QUIRKS2
7304 ----------------------------
7306 :Capability: KVM_CAP_DISABLE_QUIRKS2
7307 :Parameters: args[0] - set of KVM quirks to disable
7311 This capability, if enabled, will cause KVM to disable some behavior
7314 Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
7315 quirks that can be disabled in KVM.
7317 The argument to KVM_ENABLE_CAP for this capability is a bitmask of
7318 quirks to disable, and must be a subset of the bitmask returned by
7319 KVM_CHECK_EXTENSION.
7321 The valid bits in cap.args[0] are:
7323 =================================== ============================================
7324 KVM_X86_QUIRK_LINT0_REENABLED By default, the reset value for the LVT
7325 LINT0 register is 0x700 (APIC_MODE_EXTINT).
7326 When this quirk is disabled, the reset value
7327 is 0x10000 (APIC_LVT_MASKED).
7329 KVM_X86_QUIRK_CD_NW_CLEARED By default, KVM clears CR0.CD and CR0.NW.
7330 When this quirk is disabled, KVM does not
7331 change the value of CR0.CD and CR0.NW.
7333 KVM_X86_QUIRK_LAPIC_MMIO_HOLE By default, the MMIO LAPIC interface is
7334 available even when configured for x2APIC
7335 mode. When this quirk is disabled, KVM
7336 disables the MMIO LAPIC interface if the
7337 LAPIC is in x2APIC mode.
7339 KVM_X86_QUIRK_OUT_7E_INC_RIP By default, KVM pre-increments %rip before
7340 exiting to userspace for an OUT instruction
7341 to port 0x7e. When this quirk is disabled,
7342 KVM does not pre-increment %rip before
7343 exiting to userspace.
7345 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets
7346 CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if
7347 IA32_MISC_ENABLE[bit 18] (MWAIT) is set.
7348 Additionally, when this quirk is disabled,
7349 KVM clears CPUID.01H:ECX[bit 3] if
7350 IA32_MISC_ENABLE[bit 18] is cleared.
7352 KVM_X86_QUIRK_FIX_HYPERCALL_INSN By default, KVM rewrites guest
7353 VMMCALL/VMCALL instructions to match the
7354 vendor's hypercall instruction for the
7355 system. When this quirk is disabled, KVM
7356 will no longer rewrite invalid guest
7357 hypercall instructions. Executing the
7358 incorrect hypercall instruction will
7359 generate a #UD within the guest.
7360 =================================== ============================================
7362 8. Other capabilities.
7363 ======================
7365 This section lists capabilities that give information about other
7366 features of the KVM implementation.
7368 8.1 KVM_CAP_PPC_HWRNG
7369 ---------------------
7373 This capability, if KVM_CHECK_EXTENSION indicates that it is
7374 available, means that the kernel has an implementation of the
7375 H_RANDOM hypercall backed by a hardware random-number generator.
7376 If present, the kernel H_RANDOM handler can be enabled for guest use
7377 with the KVM_CAP_PPC_ENABLE_HCALL capability.
7379 8.2 KVM_CAP_HYPERV_SYNIC
7380 ------------------------
7384 This capability, if KVM_CHECK_EXTENSION indicates that it is
7385 available, means that the kernel has an implementation of the
7386 Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
7387 used to support Windows Hyper-V based guest paravirt drivers(VMBus).
7389 In order to use SynIC, it has to be activated by setting this
7390 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
7391 will disable the use of APIC hardware virtualization even if supported
7392 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
7394 8.3 KVM_CAP_PPC_RADIX_MMU
7395 -------------------------
7399 This capability, if KVM_CHECK_EXTENSION indicates that it is
7400 available, means that the kernel can support guests using the
7401 radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
7404 8.4 KVM_CAP_PPC_HASH_MMU_V3
7405 ---------------------------
7409 This capability, if KVM_CHECK_EXTENSION indicates that it is
7410 available, means that the kernel can support guests using the
7411 hashed page table MMU defined in Power ISA V3.00 (as implemented in
7412 the POWER9 processor), including in-memory segment tables.
7417 :Architectures: mips
7419 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7420 it is available, means that full hardware assisted virtualization capabilities
7421 of the hardware are available for use through KVM. An appropriate
7422 KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
7425 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7426 available, it means that the VM is using full hardware assisted virtualization
7427 capabilities of the hardware. This is useful to check after creating a VM with
7428 KVM_VM_MIPS_DEFAULT.
7430 The value returned by KVM_CHECK_EXTENSION should be compared against known
7431 values (see below). All other values are reserved. This is to allow for the
7432 possibility of other hardware assisted virtualization implementations which
7433 may be incompatible with the MIPS VZ ASE.
7435 == ==========================================================================
7436 0 The trap & emulate implementation is in use to run guest code in user
7437 mode. Guest virtual memory segments are rearranged to fit the guest in the
7438 user mode address space.
7440 1 The MIPS VZ ASE is in use, providing full hardware assisted
7441 virtualization, including standard guest virtual memory segments.
7442 == ==========================================================================
7447 :Architectures: mips
7449 This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7450 it is available, means that the trap & emulate implementation is available to
7451 run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
7452 assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
7453 to KVM_CREATE_VM to create a VM which utilises it.
7455 If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7456 available, it means that the VM is using trap & emulate.
7458 8.7 KVM_CAP_MIPS_64BIT
7459 ----------------------
7461 :Architectures: mips
7463 This capability indicates the supported architecture type of the guest, i.e. the
7464 supported register and address width.
7466 The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
7467 kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
7468 be checked specifically against known values (see below). All other values are
7471 == ========================================================================
7472 0 MIPS32 or microMIPS32.
7473 Both registers and addresses are 32-bits wide.
7474 It will only be possible to run 32-bit guest code.
7476 1 MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
7477 Registers are 64-bits wide, but addresses are 32-bits wide.
7478 64-bit guest code may run but cannot access MIPS64 memory segments.
7479 It will also be possible to run 32-bit guest code.
7481 2 MIPS64 or microMIPS64 with access to all address segments.
7482 Both registers and addresses are 64-bits wide.
7483 It will be possible to run 64-bit or 32-bit guest code.
7484 == ========================================================================
7486 8.9 KVM_CAP_ARM_USER_IRQ
7487 ------------------------
7489 :Architectures: arm64
7491 This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
7492 that if userspace creates a VM without an in-kernel interrupt controller, it
7493 will be notified of changes to the output level of in-kernel emulated devices,
7494 which can generate virtual interrupts, presented to the VM.
7495 For such VMs, on every return to userspace, the kernel
7496 updates the vcpu's run->s.regs.device_irq_level field to represent the actual
7497 output level of the device.
7499 Whenever kvm detects a change in the device output level, kvm guarantees at
7500 least one return to userspace before running the VM. This exit could either
7501 be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
7502 userspace can always sample the device output level and re-compute the state of
7503 the userspace interrupt controller. Userspace should always check the state
7504 of run->s.regs.device_irq_level on every kvm exit.
7505 The value in run->s.regs.device_irq_level can represent both level and edge
7506 triggered interrupt signals, depending on the device. Edge triggered interrupt
7507 signals will exit to userspace with the bit in run->s.regs.device_irq_level
7508 set exactly once per edge signal.
7510 The field run->s.regs.device_irq_level is available independent of
7511 run->kvm_valid_regs or run->kvm_dirty_regs bits.
7513 If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
7514 number larger than 0 indicating the version of this capability is implemented
7515 and thereby which bits in run->s.regs.device_irq_level can signal values.
7517 Currently the following bits are defined for the device_irq_level bitmap::
7519 KVM_CAP_ARM_USER_IRQ >= 1:
7521 KVM_ARM_DEV_EL1_VTIMER - EL1 virtual timer
7522 KVM_ARM_DEV_EL1_PTIMER - EL1 physical timer
7523 KVM_ARM_DEV_PMU - ARM PMU overflow interrupt signal
7525 Future versions of kvm may implement additional events. These will get
7526 indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
7529 8.10 KVM_CAP_PPC_SMT_POSSIBLE
7530 -----------------------------
7534 Querying this capability returns a bitmap indicating the possible
7535 virtual SMT modes that can be set using KVM_CAP_PPC_SMT. If bit N
7536 (counting from the right) is set, then a virtual SMT mode of 2^N is
7539 8.11 KVM_CAP_HYPERV_SYNIC2
7540 --------------------------
7544 This capability enables a newer version of Hyper-V Synthetic interrupt
7545 controller (SynIC). The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
7546 doesn't clear SynIC message and event flags pages when they are enabled by
7547 writing to the respective MSRs.
7549 8.12 KVM_CAP_HYPERV_VP_INDEX
7550 ----------------------------
7554 This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr. Its
7555 value is used to denote the target vcpu for a SynIC interrupt. For
7556 compatibilty, KVM initializes this msr to KVM's internal vcpu index. When this
7557 capability is absent, userspace can still query this msr's value.
7559 8.13 KVM_CAP_S390_AIS_MIGRATION
7560 -------------------------------
7562 :Architectures: s390
7565 This capability indicates if the flic device will be able to get/set the
7566 AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
7567 to discover this without having to create a flic device.
7569 8.14 KVM_CAP_S390_PSW
7570 ---------------------
7572 :Architectures: s390
7574 This capability indicates that the PSW is exposed via the kvm_run structure.
7576 8.15 KVM_CAP_S390_GMAP
7577 ----------------------
7579 :Architectures: s390
7581 This capability indicates that the user space memory used as guest mapping can
7582 be anywhere in the user memory address space, as long as the memory slots are
7583 aligned and sized to a segment (1MB) boundary.
7585 8.16 KVM_CAP_S390_COW
7586 ---------------------
7588 :Architectures: s390
7590 This capability indicates that the user space memory used as guest mapping can
7591 use copy-on-write semantics as well as dirty pages tracking via read-only page
7594 8.17 KVM_CAP_S390_BPB
7595 ---------------------
7597 :Architectures: s390
7599 This capability indicates that kvm will implement the interfaces to handle
7600 reset, migration and nested KVM for branch prediction blocking. The stfle
7601 facility 82 should not be provided to the guest without this capability.
7603 8.18 KVM_CAP_HYPERV_TLBFLUSH
7604 ----------------------------
7608 This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
7610 HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
7611 HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
7613 8.19 KVM_CAP_ARM_INJECT_SERROR_ESR
7614 ----------------------------------
7616 :Architectures: arm64
7618 This capability indicates that userspace can specify (via the
7619 KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
7620 takes a virtual SError interrupt exception.
7621 If KVM advertises this capability, userspace can only specify the ISS field for
7622 the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
7623 CPU when the exception is taken. If this virtual SError is taken to EL1 using
7624 AArch64, this value will be reported in the ISS field of ESR_ELx.
7626 See KVM_CAP_VCPU_EVENTS for more details.
7628 8.20 KVM_CAP_HYPERV_SEND_IPI
7629 ----------------------------
7633 This capability indicates that KVM supports paravirtualized Hyper-V IPI send
7635 HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
7637 8.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
7638 -----------------------------------
7642 This capability indicates that KVM running on top of Hyper-V hypervisor
7643 enables Direct TLB flush for its guests meaning that TLB flush
7644 hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
7645 Due to the different ABI for hypercall parameters between Hyper-V and
7646 KVM, enabling this capability effectively disables all hypercall
7647 handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
7648 flush hypercalls by Hyper-V) so userspace should disable KVM identification
7649 in CPUID and only exposes Hyper-V identification. In this case, guest
7650 thinks it's running on Hyper-V and only use Hyper-V hypercalls.
7652 8.22 KVM_CAP_S390_VCPU_RESETS
7653 -----------------------------
7655 :Architectures: s390
7657 This capability indicates that the KVM_S390_NORMAL_RESET and
7658 KVM_S390_CLEAR_RESET ioctls are available.
7660 8.23 KVM_CAP_S390_PROTECTED
7661 ---------------------------
7663 :Architectures: s390
7665 This capability indicates that the Ultravisor has been initialized and
7666 KVM can therefore start protected VMs.
7667 This capability governs the KVM_S390_PV_COMMAND ioctl and the
7668 KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
7669 guests when the state change is invalid.
7671 8.24 KVM_CAP_STEAL_TIME
7672 -----------------------
7674 :Architectures: arm64, x86
7676 This capability indicates that KVM supports steal time accounting.
7677 When steal time accounting is supported it may be enabled with
7678 architecture-specific interfaces. This capability and the architecture-
7679 specific interfaces must be consistent, i.e. if one says the feature
7680 is supported, than the other should as well and vice versa. For arm64
7681 see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
7682 For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
7684 8.25 KVM_CAP_S390_DIAG318
7685 -------------------------
7687 :Architectures: s390
7689 This capability enables a guest to set information about its control program
7690 (i.e. guest kernel type and version). The information is helpful during
7691 system/firmware service events, providing additional data about the guest
7692 environments running on the machine.
7694 The information is associated with the DIAGNOSE 0x318 instruction, which sets
7695 an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
7696 a 7-byte Control Program Version Code (CPVC). The CPNC determines what
7697 environment the control program is running in (e.g. Linux, z/VM...), and the
7698 CPVC is used for information specific to OS (e.g. Linux version, Linux
7701 If this capability is available, then the CPNC and CPVC can be synchronized
7702 between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
7704 8.26 KVM_CAP_X86_USER_SPACE_MSR
7705 -------------------------------
7709 This capability indicates that KVM supports deflection of MSR reads and
7710 writes to user space. It can be enabled on a VM level. If enabled, MSR
7711 accesses that would usually trigger a #GP by KVM into the guest will
7712 instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
7713 KVM_EXIT_X86_WRMSR exit notifications.
7715 8.27 KVM_CAP_X86_MSR_FILTER
7716 ---------------------------
7720 This capability indicates that KVM supports that accesses to user defined MSRs
7721 may be rejected. With this capability exposed, KVM exports new VM ioctl
7722 KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
7723 ranges that KVM should reject access to.
7725 In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
7726 trap and emulate MSRs that are outside of the scope of KVM as well as
7727 limit the attack surface on KVM's MSR emulation code.
7729 8.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
7730 -------------------------------------
7734 When enabled, KVM will disable paravirtual features provided to the
7735 guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
7736 (0x40000001). Otherwise, a guest may use the paravirtual features
7737 regardless of what has actually been exposed through the CPUID leaf.
7739 8.29 KVM_CAP_DIRTY_LOG_RING
7740 ---------------------------
7743 :Parameters: args[0] - size of the dirty log ring
7745 KVM is capable of tracking dirty memory using ring buffers that are
7746 mmaped into userspace; there is one dirty ring per vcpu.
7748 The dirty ring is available to userspace as an array of
7749 ``struct kvm_dirty_gfn``. Each dirty entry it's defined as::
7751 struct kvm_dirty_gfn {
7753 __u32 slot; /* as_id | slot_id */
7757 The following values are defined for the flags field to define the
7758 current state of the entry::
7760 #define KVM_DIRTY_GFN_F_DIRTY BIT(0)
7761 #define KVM_DIRTY_GFN_F_RESET BIT(1)
7762 #define KVM_DIRTY_GFN_F_MASK 0x3
7764 Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
7765 ioctl to enable this capability for the new guest and set the size of
7766 the rings. Enabling the capability is only allowed before creating any
7767 vCPU, and the size of the ring must be a power of two. The larger the
7768 ring buffer, the less likely the ring is full and the VM is forced to
7769 exit to userspace. The optimal size depends on the workload, but it is
7770 recommended that it be at least 64 KiB (4096 entries).
7772 Just like for dirty page bitmaps, the buffer tracks writes to
7773 all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
7774 set in KVM_SET_USER_MEMORY_REGION. Once a memory region is registered
7775 with the flag set, userspace can start harvesting dirty pages from the
7778 An entry in the ring buffer can be unused (flag bits ``00``),
7779 dirty (flag bits ``01``) or harvested (flag bits ``1X``). The
7780 state machine for the entry is as follows::
7782 dirtied harvested reset
7783 00 -----------> 01 -------------> 1X -------+
7786 +------------------------------------------+
7788 To harvest the dirty pages, userspace accesses the mmaped ring buffer
7789 to read the dirty GFNs. If the flags has the DIRTY bit set (at this stage
7790 the RESET bit must be cleared), then it means this GFN is a dirty GFN.
7791 The userspace should harvest this GFN and mark the flags from state
7792 ``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
7793 to show that this GFN is harvested and waiting for a reset), and move
7794 on to the next GFN. The userspace should continue to do this until the
7795 flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
7796 all the dirty GFNs that were available.
7798 It's not necessary for userspace to harvest the all dirty GFNs at once.
7799 However it must collect the dirty GFNs in sequence, i.e., the userspace
7800 program cannot skip one dirty GFN to collect the one next to it.
7802 After processing one or more entries in the ring buffer, userspace
7803 calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
7804 it, so that the kernel will reprotect those collected GFNs.
7805 Therefore, the ioctl must be called *before* reading the content of
7808 The dirty ring can get full. When it happens, the KVM_RUN of the
7809 vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
7811 The dirty ring interface has a major difference comparing to the
7812 KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
7813 userspace, it's still possible that the kernel has not yet flushed the
7814 processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
7815 flushing is done by the KVM_GET_DIRTY_LOG ioctl). To achieve that, one
7816 needs to kick the vcpu out of KVM_RUN using a signal. The resulting
7817 vmexit ensures that all dirty GFNs are flushed to the dirty rings.
7819 NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
7820 ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
7821 KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG. After enabling
7822 KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
7823 machine will switch to ring-buffer dirty page tracking and further
7824 KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
7826 8.30 KVM_CAP_XEN_HVM
7827 --------------------
7831 This capability indicates the features that Xen supports for hosting Xen
7832 PVHVM guests. Valid flags are::
7834 #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0)
7835 #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1)
7836 #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2)
7837 #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3)
7838 #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL (1 << 4)
7839 #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND (1 << 5)
7841 The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
7842 ioctl is available, for the guest to set its hypercall page.
7844 If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
7845 provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
7846 contents, to request that KVM generate hypercall page content automatically
7847 and also enable interception of guest hypercalls with KVM_EXIT_XEN.
7849 The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
7850 KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
7851 KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
7852 for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
7855 The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
7856 features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
7857 supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
7859 The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries
7860 of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority
7861 field set to indicate 2 level event channel delivery.
7863 The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports
7864 injecting event channel events directly into the guest with the
7865 KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the
7866 KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the
7867 KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes.
7868 related to event channel delivery, timers, and the XENVER_version
7871 8.31 KVM_CAP_PPC_MULTITCE
7872 -------------------------
7874 :Capability: KVM_CAP_PPC_MULTITCE
7878 This capability means the kernel is capable of handling hypercalls
7879 H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
7880 space. This significantly accelerates DMA operations for PPC KVM guests.
7881 User space should expect that its handlers for these hypercalls
7882 are not going to be called if user space previously registered LIOBN
7883 in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
7885 In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
7886 user space might have to advertise it for the guest. For example,
7887 IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
7888 present in the "ibm,hypertas-functions" device-tree property.
7890 The hypercalls mentioned above may or may not be processed successfully
7891 in the kernel based fast path. If they can not be handled by the kernel,
7892 they will get passed on to user space. So user space still has to have
7893 an implementation for these despite the in kernel acceleration.
7895 This capability is always enabled.
7897 8.32 KVM_CAP_PTP_KVM
7898 --------------------
7900 :Architectures: arm64
7902 This capability indicates that the KVM virtual PTP service is
7903 supported in the host. A VMM can check whether the service is
7904 available to the guest on migration.
7906 8.33 KVM_CAP_HYPERV_ENFORCE_CPUID
7907 ---------------------------------
7911 When enabled, KVM will disable emulated Hyper-V features provided to the
7912 guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
7913 currently implmented Hyper-V features are provided unconditionally when
7914 Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
7917 8.34 KVM_CAP_EXIT_HYPERCALL
7918 ---------------------------
7920 :Capability: KVM_CAP_EXIT_HYPERCALL
7924 This capability, if enabled, will cause KVM to exit to userspace
7925 with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
7927 Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
7928 of hypercalls that can be configured to exit to userspace.
7929 Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
7931 The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
7932 of the result of KVM_CHECK_EXTENSION. KVM will forward to userspace
7933 the hypercalls whose corresponding bit is in the argument, and return
7934 ENOSYS for the others.
7936 8.35 KVM_CAP_PMU_CAPABILITY
7937 ---------------------------
7939 :Capability KVM_CAP_PMU_CAPABILITY
7942 :Parameters: arg[0] is bitmask of PMU virtualization capabilities.
7943 :Returns 0 on success, -EINVAL when arg[0] contains invalid bits
7945 This capability alters PMU virtualization in KVM.
7947 Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
7948 PMU virtualization capabilities that can be adjusted on a VM.
7950 The argument to KVM_ENABLE_CAP is also a bitmask and selects specific
7951 PMU virtualization capabilities to be applied to the VM. This can
7952 only be invoked on a VM prior to the creation of VCPUs.
7954 At this time, KVM_PMU_CAP_DISABLE is the only capability. Setting
7955 this capability will disable PMU virtualization for that VM. Usermode
7956 should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
7958 8.36 KVM_CAP_ARM_SYSTEM_SUSPEND
7959 -------------------------------
7961 :Capability: KVM_CAP_ARM_SYSTEM_SUSPEND
7962 :Architectures: arm64
7965 When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of
7966 type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request.
7968 9. Known KVM API problems
7969 =========================
7971 In some cases, KVM's API has some inconsistencies or common pitfalls
7972 that userspace need to be aware of. This section details some of
7975 Most of them are architecture specific, so the section is split by
7981 ``KVM_GET_SUPPORTED_CPUID`` issues
7982 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
7984 In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible
7985 to take its result and pass it directly to ``KVM_SET_CPUID2``. This section
7986 documents some cases in which that requires some care.
7991 CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``,
7992 but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or
7993 ``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of
7996 The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature.
7998 CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``.
7999 It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel
8000 has enabled in-kernel emulation of the local APIC.
8002 Obsolete ioctls and capabilities
8003 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8005 KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually
8006 available. Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if
8009 Ordering of KVM_GET_*/KVM_SET_* ioctls
8010 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^