xref: /linux/Documentation/virt/kvm/api.rst (revision 1fd1dc41724319406b0aff221a352a400b0ddfc5)
1.. SPDX-License-Identifier: GPL-2.0
2
3===================================================================
4The Definitive KVM (Kernel-based Virtual Machine) API Documentation
5===================================================================
6
71. General description
8======================
9
10The kvm API is centered around different kinds of file descriptors
11and ioctls that can be issued to these file descriptors.  An initial
12open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
13can be used to issue system ioctls.  A KVM_CREATE_VM ioctl on this
14handle will create a VM file descriptor which can be used to issue VM
15ioctls.  A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
16create a virtual cpu or device and return a file descriptor pointing to
17the new resource.
18
19In other words, the kvm API is a set of ioctls that are issued to
20different kinds of file descriptor in order to control various aspects of
21a virtual machine.  Depending on the file descriptor that accepts them,
22ioctls belong to the following classes:
23
24 - System ioctls: These query and set global attributes which affect the
25   whole kvm subsystem.  In addition a system ioctl is used to create
26   virtual machines.
27
28 - VM ioctls: These query and set attributes that affect an entire virtual
29   machine, for example memory layout.  In addition a VM ioctl is used to
30   create virtual cpus (vcpus) and devices.
31
32   VM ioctls must be issued from the same process (address space) that was
33   used to create the VM.
34
35 - vcpu ioctls: These query and set attributes that control the operation
36   of a single virtual cpu.
37
38   vcpu ioctls should be issued from the same thread that was used to create
39   the vcpu, except for asynchronous vcpu ioctl that are marked as such in
40   the documentation.  Otherwise, the first ioctl after switching threads
41   could see a performance impact.
42
43 - device ioctls: These query and set attributes that control the operation
44   of a single device.
45
46   device ioctls must be issued from the same process (address space) that
47   was used to create the VM.
48
49While most ioctls are specific to one kind of file descriptor, in some
50cases the same ioctl can belong to more than one class.
51
52The KVM API grew over time.  For this reason, KVM defines many constants
53of the form ``KVM_CAP_*``, each corresponding to a set of functionality
54provided by one or more ioctls.  Availability of these "capabilities" can
55be checked with :ref:`KVM_CHECK_EXTENSION <KVM_CHECK_EXTENSION>`.  Some
56capabilities also need to be enabled for VMs or VCPUs where their
57functionality is desired (see :ref:`cap_enable` and :ref:`cap_enable_vm`).
58
59
602. Restrictions
61===============
62
63In general file descriptors can be migrated among processes by means
64of fork() and the SCM_RIGHTS facility of unix domain socket.  These
65kinds of tricks are explicitly not supported by kvm.  While they will
66not cause harm to the host, their actual behavior is not guaranteed by
67the API.  See "General description" for details on the ioctl usage
68model that is supported by KVM.
69
70It is important to note that although VM ioctls may only be issued from
71the process that created the VM, a VM's lifecycle is associated with its
72file descriptor, not its creator (process).  In other words, the VM and
73its resources, *including the associated address space*, are not freed
74until the last reference to the VM's file descriptor has been released.
75For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will
76not be freed until both the parent (original) process and its child have
77put their references to the VM's file descriptor.
78
79Because a VM's resources are not freed until the last reference to its
80file descriptor is released, creating additional references to a VM
81via fork(), dup(), etc... without careful consideration is strongly
82discouraged and may have unwanted side effects, e.g. memory allocated
83by and on behalf of the VM's process may not be freed/unaccounted when
84the VM is shut down.
85
86
873. Extensions
88=============
89
90As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
91incompatible change are allowed.  However, there is an extension
92facility that allows backward-compatible extensions to the API to be
93queried and used.
94
95The extension mechanism is not based on the Linux version number.
96Instead, kvm defines extension identifiers and a facility to query
97whether a particular extension identifier is available.  If it is, a
98set of ioctls is available for application use.
99
100
1014. API description
102==================
103
104This section describes ioctls that can be used to control kvm guests.
105For each ioctl, the following information is provided along with a
106description:
107
108  Capability:
109      which KVM extension provides this ioctl.  Can be 'basic',
110      which means that is will be provided by any kernel that supports
111      API version 12 (see :ref:`KVM_GET_API_VERSION <KVM_GET_API_VERSION>`),
112      or a KVM_CAP_xyz constant that can be checked with
113      :ref:`KVM_CHECK_EXTENSION <KVM_CHECK_EXTENSION>`.
114
115  Architectures:
116      which instruction set architectures provide this ioctl.
117      x86 includes both i386 and x86_64.
118
119  Type:
120      system, vm, or vcpu.
121
122  Parameters:
123      what parameters are accepted by the ioctl.
124
125  Returns:
126      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
127      are not detailed, but errors with specific meanings are.
128
129
130.. _KVM_GET_API_VERSION:
131
1324.1 KVM_GET_API_VERSION
133-----------------------
134
135:Capability: basic
136:Architectures: all
137:Type: system ioctl
138:Parameters: none
139:Returns: the constant KVM_API_VERSION (=12)
140
141This identifies the API version as the stable kvm API. It is not
142expected that this number will change.  However, Linux 2.6.20 and
1432.6.21 report earlier versions; these are not documented and not
144supported.  Applications should refuse to run if KVM_GET_API_VERSION
145returns a value other than 12.  If this check passes, all ioctls
146described as 'basic' will be available.
147
148
1494.2 KVM_CREATE_VM
150-----------------
151
152:Capability: basic
153:Architectures: all
154:Type: system ioctl
155:Parameters: machine type identifier (KVM_VM_*)
156:Returns: a VM fd that can be used to control the new virtual machine.
157
158The new VM has no virtual cpus and no memory.
159You probably want to use 0 as machine type.
160
161X86:
162^^^^
163
164Supported X86 VM types can be queried via KVM_CAP_VM_TYPES.
165
166S390:
167^^^^^
168
169In order to create user controlled virtual machines on S390, check
170KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
171privileged user (CAP_SYS_ADMIN).
172
173MIPS:
174^^^^^
175
176To use hardware assisted virtualization on MIPS (VZ ASE) rather than
177the default trap & emulate implementation (which changes the virtual
178memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
179flag KVM_VM_MIPS_VZ.
180
181ARM64:
182^^^^^^
183
184On arm64, the physical address size for a VM (IPA Size limit) is limited
185to 40bits by default. The limit can be configured if the host supports the
186extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
187KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
188identifier, where IPA_Bits is the maximum width of any physical
189address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
190machine type identifier.
191
192e.g, to configure a guest to use 48bit physical address size::
193
194    vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
195
196The requested size (IPA_Bits) must be:
197
198 ==   =========================================================
199  0   Implies default size, 40bits (for backward compatibility)
200  N   Implies N bits, where N is a positive integer such that,
201      32 <= N <= Host_IPA_Limit
202 ==   =========================================================
203
204Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
205is dependent on the CPU capability and the kernel configuration. The limit can
206be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
207ioctl() at run-time.
208
209Creation of the VM will fail if the requested IPA size (whether it is
210implicit or explicit) is unsupported on the host.
211
212Please note that configuring the IPA size does not affect the capability
213exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
214size of the address translated by the stage2 level (guest physical to
215host physical address translations).
216
217
2184.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
219----------------------------------------------------------
220
221:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
222:Architectures: x86
223:Type: system ioctl
224:Parameters: struct kvm_msr_list (in/out)
225:Returns: 0 on success; -1 on error
226
227Errors:
228
229  ======     ============================================================
230  EFAULT     the msr index list cannot be read from or written to
231  E2BIG      the msr index list is too big to fit in the array specified by
232             the user.
233  ======     ============================================================
234
235::
236
237  struct kvm_msr_list {
238	__u32 nmsrs; /* number of msrs in entries */
239	__u32 indices[0];
240  };
241
242The user fills in the size of the indices array in nmsrs, and in return
243kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
244indices array with their numbers.
245
246KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported.  The list
247varies by kvm version and host processor, but does not change otherwise.
248
249Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
250not returned in the MSR list, as different vcpus can have a different number
251of banks, as set via the KVM_X86_SETUP_MCE ioctl.
252
253KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
254to the KVM_GET_MSRS system ioctl.  This lets userspace probe host capabilities
255and processor features that are exposed via MSRs (e.g., VMX capabilities).
256This list also varies by kvm version and host processor, but does not change
257otherwise.
258
259
260.. _KVM_CHECK_EXTENSION:
261
2624.4 KVM_CHECK_EXTENSION
263-----------------------
264
265:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
266:Architectures: all
267:Type: system ioctl, vm ioctl
268:Parameters: extension identifier (KVM_CAP_*)
269:Returns: 0 if unsupported; 1 (or some other positive integer) if supported
270
271The API allows the application to query about extensions to the core
272kvm API.  Userspace passes an extension identifier (an integer) and
273receives an integer that describes the extension availability.
274Generally 0 means no and 1 means yes, but some extensions may report
275additional information in the integer return value.
276
277Based on their initialization different VMs may have different capabilities.
278It is thus encouraged to use the vm ioctl to query for capabilities (available
279with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
280
2814.5 KVM_GET_VCPU_MMAP_SIZE
282--------------------------
283
284:Capability: basic
285:Architectures: all
286:Type: system ioctl
287:Parameters: none
288:Returns: size of vcpu mmap area, in bytes
289
290The KVM_RUN ioctl (cf.) communicates with userspace via a shared
291memory region.  This ioctl returns the size of that region.  See the
292KVM_RUN documentation for details.
293
294Besides the size of the KVM_RUN communication region, other areas of
295the VCPU file descriptor can be mmap-ed, including:
296
297- if KVM_CAP_COALESCED_MMIO is available, a page at
298  KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
299  this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
300  KVM_CAP_COALESCED_MMIO is not documented yet.
301
302- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
303  KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE.  For more information on
304  KVM_CAP_DIRTY_LOG_RING, see :ref:`KVM_CAP_DIRTY_LOG_RING`.
305
306
3074.7 KVM_CREATE_VCPU
308-------------------
309
310:Capability: basic
311:Architectures: all
312:Type: vm ioctl
313:Parameters: vcpu id (apic id on x86)
314:Returns: vcpu fd on success, -1 on error
315
316This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
317The vcpu id is an integer in the range [0, max_vcpu_id).
318
319The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
320the KVM_CHECK_EXTENSION ioctl() at run-time.
321The maximum possible value for max_vcpus can be retrieved using the
322KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
323
324If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
325cpus max.
326If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
327same as the value returned from KVM_CAP_NR_VCPUS.
328
329The maximum possible value for max_vcpu_id can be retrieved using the
330KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
331
332If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
333is the same as the value returned from KVM_CAP_MAX_VCPUS.
334
335On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
336threads in one or more virtual CPU cores.  (This is because the
337hardware requires all the hardware threads in a CPU core to be in the
338same partition.)  The KVM_CAP_PPC_SMT capability indicates the number
339of vcpus per virtual core (vcore).  The vcore id is obtained by
340dividing the vcpu id by the number of vcpus per vcore.  The vcpus in a
341given vcore will always be in the same physical core as each other
342(though that might be a different physical core from time to time).
343Userspace can control the threading (SMT) mode of the guest by its
344allocation of vcpu ids.  For example, if userspace wants
345single-threaded guest vcpus, it should make all vcpu ids be a multiple
346of the number of vcpus per vcore.
347
348For virtual cpus that have been created with S390 user controlled virtual
349machines, the resulting vcpu fd can be memory mapped at page offset
350KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
351cpu's hardware control block.
352
353
3544.8 KVM_GET_DIRTY_LOG
355---------------------
356
357:Capability: basic
358:Architectures: all
359:Type: vm ioctl
360:Parameters: struct kvm_dirty_log (in/out)
361:Returns: 0 on success, -1 on error
362
363::
364
365  /* for KVM_GET_DIRTY_LOG */
366  struct kvm_dirty_log {
367	__u32 slot;
368	__u32 padding;
369	union {
370		void __user *dirty_bitmap; /* one bit per page */
371		__u64 padding;
372	};
373  };
374
375Given a memory slot, return a bitmap containing any pages dirtied
376since the last call to this ioctl.  Bit 0 is the first page in the
377memory slot.  Ensure the entire structure is cleared to avoid padding
378issues.
379
380If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
381the address space for which you want to return the dirty bitmap.  See
382KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
383
384The bits in the dirty bitmap are cleared before the ioctl returns, unless
385KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled.  For more information,
386see the description of the capability.
387
388Note that the Xen shared_info page, if configured, shall always be assumed
389to be dirty. KVM will not explicitly mark it such.
390
391
3924.10 KVM_RUN
393------------
394
395:Capability: basic
396:Architectures: all
397:Type: vcpu ioctl
398:Parameters: none
399:Returns: 0 on success, -1 on error
400
401Errors:
402
403  =======    ==============================================================
404  EINTR      an unmasked signal is pending
405  ENOEXEC    the vcpu hasn't been initialized or the guest tried to execute
406             instructions from device memory (arm64)
407  ENOSYS     data abort outside memslots with no syndrome info and
408             KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
409  EPERM      SVE feature set but not finalized (arm64)
410  =======    ==============================================================
411
412This ioctl is used to run a guest virtual cpu.  While there are no
413explicit parameters, there is an implicit parameter block that can be
414obtained by mmap()ing the vcpu fd at offset 0, with the size given by
415KVM_GET_VCPU_MMAP_SIZE.  The parameter block is formatted as a 'struct
416kvm_run' (see below).
417
418
4194.11 KVM_GET_REGS
420-----------------
421
422:Capability: basic
423:Architectures: all except arm64
424:Type: vcpu ioctl
425:Parameters: struct kvm_regs (out)
426:Returns: 0 on success, -1 on error
427
428Reads the general purpose registers from the vcpu.
429
430::
431
432  /* x86 */
433  struct kvm_regs {
434	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
435	__u64 rax, rbx, rcx, rdx;
436	__u64 rsi, rdi, rsp, rbp;
437	__u64 r8,  r9,  r10, r11;
438	__u64 r12, r13, r14, r15;
439	__u64 rip, rflags;
440  };
441
442  /* mips */
443  struct kvm_regs {
444	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
445	__u64 gpr[32];
446	__u64 hi;
447	__u64 lo;
448	__u64 pc;
449  };
450
451  /* LoongArch */
452  struct kvm_regs {
453	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
454	unsigned long gpr[32];
455	unsigned long pc;
456  };
457
458
4594.12 KVM_SET_REGS
460-----------------
461
462:Capability: basic
463:Architectures: all except arm64
464:Type: vcpu ioctl
465:Parameters: struct kvm_regs (in)
466:Returns: 0 on success, -1 on error
467
468Writes the general purpose registers into the vcpu.
469
470See KVM_GET_REGS for the data structure.
471
472
4734.13 KVM_GET_SREGS
474------------------
475
476:Capability: basic
477:Architectures: x86, ppc
478:Type: vcpu ioctl
479:Parameters: struct kvm_sregs (out)
480:Returns: 0 on success, -1 on error
481
482Reads special registers from the vcpu.
483
484::
485
486  /* x86 */
487  struct kvm_sregs {
488	struct kvm_segment cs, ds, es, fs, gs, ss;
489	struct kvm_segment tr, ldt;
490	struct kvm_dtable gdt, idt;
491	__u64 cr0, cr2, cr3, cr4, cr8;
492	__u64 efer;
493	__u64 apic_base;
494	__u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
495  };
496
497  /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
498
499interrupt_bitmap is a bitmap of pending external interrupts.  At most
500one bit may be set.  This interrupt has been acknowledged by the APIC
501but not yet injected into the cpu core.
502
503
5044.14 KVM_SET_SREGS
505------------------
506
507:Capability: basic
508:Architectures: x86, ppc
509:Type: vcpu ioctl
510:Parameters: struct kvm_sregs (in)
511:Returns: 0 on success, -1 on error
512
513Writes special registers into the vcpu.  See KVM_GET_SREGS for the
514data structures.
515
516
5174.15 KVM_TRANSLATE
518------------------
519
520:Capability: basic
521:Architectures: x86
522:Type: vcpu ioctl
523:Parameters: struct kvm_translation (in/out)
524:Returns: 0 on success, -1 on error
525
526Translates a virtual address according to the vcpu's current address
527translation mode.
528
529::
530
531  struct kvm_translation {
532	/* in */
533	__u64 linear_address;
534
535	/* out */
536	__u64 physical_address;
537	__u8  valid;
538	__u8  writeable;
539	__u8  usermode;
540	__u8  pad[5];
541  };
542
543
5444.16 KVM_INTERRUPT
545------------------
546
547:Capability: basic
548:Architectures: x86, ppc, mips, riscv, loongarch
549:Type: vcpu ioctl
550:Parameters: struct kvm_interrupt (in)
551:Returns: 0 on success, negative on failure.
552
553Queues a hardware interrupt vector to be injected.
554
555::
556
557  /* for KVM_INTERRUPT */
558  struct kvm_interrupt {
559	/* in */
560	__u32 irq;
561  };
562
563X86:
564^^^^
565
566:Returns:
567
568	========= ===================================
569	  0       on success,
570	 -EEXIST  if an interrupt is already enqueued
571	 -EINVAL  the irq number is invalid
572	 -ENXIO   if the PIC is in the kernel
573	 -EFAULT  if the pointer is invalid
574	========= ===================================
575
576Note 'irq' is an interrupt vector, not an interrupt pin or line. This
577ioctl is useful if the in-kernel PIC is not used.
578
579PPC:
580^^^^
581
582Queues an external interrupt to be injected. This ioctl is overloaded
583with 3 different irq values:
584
585a) KVM_INTERRUPT_SET
586
587   This injects an edge type external interrupt into the guest once it's ready
588   to receive interrupts. When injected, the interrupt is done.
589
590b) KVM_INTERRUPT_UNSET
591
592   This unsets any pending interrupt.
593
594   Only available with KVM_CAP_PPC_UNSET_IRQ.
595
596c) KVM_INTERRUPT_SET_LEVEL
597
598   This injects a level type external interrupt into the guest context. The
599   interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
600   is triggered.
601
602   Only available with KVM_CAP_PPC_IRQ_LEVEL.
603
604Note that any value for 'irq' other than the ones stated above is invalid
605and incurs unexpected behavior.
606
607This is an asynchronous vcpu ioctl and can be invoked from any thread.
608
609MIPS:
610^^^^^
611
612Queues an external interrupt to be injected into the virtual CPU. A negative
613interrupt number dequeues the interrupt.
614
615This is an asynchronous vcpu ioctl and can be invoked from any thread.
616
617RISC-V:
618^^^^^^^
619
620Queues an external interrupt to be injected into the virtual CPU. This ioctl
621is overloaded with 2 different irq values:
622
623a) KVM_INTERRUPT_SET
624
625   This sets external interrupt for a virtual CPU and it will receive
626   once it is ready.
627
628b) KVM_INTERRUPT_UNSET
629
630   This clears pending external interrupt for a virtual CPU.
631
632This is an asynchronous vcpu ioctl and can be invoked from any thread.
633
634LOONGARCH:
635^^^^^^^^^^
636
637Queues an external interrupt to be injected into the virtual CPU. A negative
638interrupt number dequeues the interrupt.
639
640This is an asynchronous vcpu ioctl and can be invoked from any thread.
641
642
6434.18 KVM_GET_MSRS
644-----------------
645
646:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
647:Architectures: x86
648:Type: system ioctl, vcpu ioctl
649:Parameters: struct kvm_msrs (in/out)
650:Returns: number of msrs successfully returned;
651          -1 on error
652
653When used as a system ioctl:
654Reads the values of MSR-based features that are available for the VM.  This
655is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
656The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
657in a system ioctl.
658
659When used as a vcpu ioctl:
660Reads model-specific registers from the vcpu.  Supported msr indices can
661be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
662
663::
664
665  struct kvm_msrs {
666	__u32 nmsrs; /* number of msrs in entries */
667	__u32 pad;
668
669	struct kvm_msr_entry entries[0];
670  };
671
672  struct kvm_msr_entry {
673	__u32 index;
674	__u32 reserved;
675	__u64 data;
676  };
677
678Application code should set the 'nmsrs' member (which indicates the
679size of the entries array) and the 'index' member of each array entry.
680kvm will fill in the 'data' member.
681
682
6834.19 KVM_SET_MSRS
684-----------------
685
686:Capability: basic
687:Architectures: x86
688:Type: vcpu ioctl
689:Parameters: struct kvm_msrs (in)
690:Returns: number of msrs successfully set (see below), -1 on error
691
692Writes model-specific registers to the vcpu.  See KVM_GET_MSRS for the
693data structures.
694
695Application code should set the 'nmsrs' member (which indicates the
696size of the entries array), and the 'index' and 'data' members of each
697array entry.
698
699It tries to set the MSRs in array entries[] one by one. If setting an MSR
700fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
701by KVM, etc..., it stops processing the MSR list and returns the number of
702MSRs that have been set successfully.
703
704
7054.20 KVM_SET_CPUID
706------------------
707
708:Capability: basic
709:Architectures: x86
710:Type: vcpu ioctl
711:Parameters: struct kvm_cpuid (in)
712:Returns: 0 on success, -1 on error
713
714Defines the vcpu responses to the cpuid instruction.  Applications
715should use the KVM_SET_CPUID2 ioctl if available.
716
717Caveat emptor:
718  - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID
719    configuration (if there is) is not corrupted. Userspace can get a copy
720    of the resulting CPUID configuration through KVM_GET_CPUID2 in case.
721  - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model
722    after running the guest, may cause guest instability.
723  - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc...
724    may cause guest instability.
725
726::
727
728  struct kvm_cpuid_entry {
729	__u32 function;
730	__u32 eax;
731	__u32 ebx;
732	__u32 ecx;
733	__u32 edx;
734	__u32 padding;
735  };
736
737  /* for KVM_SET_CPUID */
738  struct kvm_cpuid {
739	__u32 nent;
740	__u32 padding;
741	struct kvm_cpuid_entry entries[0];
742  };
743
744
7454.21 KVM_SET_SIGNAL_MASK
746------------------------
747
748:Capability: basic
749:Architectures: all
750:Type: vcpu ioctl
751:Parameters: struct kvm_signal_mask (in)
752:Returns: 0 on success, -1 on error
753
754Defines which signals are blocked during execution of KVM_RUN.  This
755signal mask temporarily overrides the threads signal mask.  Any
756unblocked signal received (except SIGKILL and SIGSTOP, which retain
757their traditional behaviour) will cause KVM_RUN to return with -EINTR.
758
759Note the signal will only be delivered if not blocked by the original
760signal mask.
761
762::
763
764  /* for KVM_SET_SIGNAL_MASK */
765  struct kvm_signal_mask {
766	__u32 len;
767	__u8  sigset[0];
768  };
769
770
7714.22 KVM_GET_FPU
772----------------
773
774:Capability: basic
775:Architectures: x86, loongarch
776:Type: vcpu ioctl
777:Parameters: struct kvm_fpu (out)
778:Returns: 0 on success, -1 on error
779
780Reads the floating point state from the vcpu.
781
782::
783
784  /* x86: for KVM_GET_FPU and KVM_SET_FPU */
785  struct kvm_fpu {
786	__u8  fpr[8][16];
787	__u16 fcw;
788	__u16 fsw;
789	__u8  ftwx;  /* in fxsave format */
790	__u8  pad1;
791	__u16 last_opcode;
792	__u64 last_ip;
793	__u64 last_dp;
794	__u8  xmm[16][16];
795	__u32 mxcsr;
796	__u32 pad2;
797  };
798
799  /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */
800  struct kvm_fpu {
801	__u32 fcsr;
802	__u64 fcc;
803	struct kvm_fpureg {
804		__u64 val64[4];
805	}fpr[32];
806  };
807
808
8094.23 KVM_SET_FPU
810----------------
811
812:Capability: basic
813:Architectures: x86, loongarch
814:Type: vcpu ioctl
815:Parameters: struct kvm_fpu (in)
816:Returns: 0 on success, -1 on error
817
818Writes the floating point state to the vcpu.
819
820::
821
822  /* x86: for KVM_GET_FPU and KVM_SET_FPU */
823  struct kvm_fpu {
824	__u8  fpr[8][16];
825	__u16 fcw;
826	__u16 fsw;
827	__u8  ftwx;  /* in fxsave format */
828	__u8  pad1;
829	__u16 last_opcode;
830	__u64 last_ip;
831	__u64 last_dp;
832	__u8  xmm[16][16];
833	__u32 mxcsr;
834	__u32 pad2;
835  };
836
837  /* LoongArch: for KVM_GET_FPU and KVM_SET_FPU */
838  struct kvm_fpu {
839	__u32 fcsr;
840	__u64 fcc;
841	struct kvm_fpureg {
842		__u64 val64[4];
843	}fpr[32];
844  };
845
846
8474.24 KVM_CREATE_IRQCHIP
848-----------------------
849
850:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
851:Architectures: x86, arm64, s390
852:Type: vm ioctl
853:Parameters: none
854:Returns: 0 on success, -1 on error
855
856Creates an interrupt controller model in the kernel.
857On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
858future vcpus to have a local APIC.  IRQ routing for GSIs 0-15 is set to both
859PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
860On arm64, a GICv2 is created. Any other GIC versions require the usage of
861KVM_CREATE_DEVICE, which also supports creating a GICv2.  Using
862KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
863On s390, a dummy irq routing table is created.
864
865Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
866before KVM_CREATE_IRQCHIP can be used.
867
868
8694.25 KVM_IRQ_LINE
870-----------------
871
872:Capability: KVM_CAP_IRQCHIP
873:Architectures: x86, arm64
874:Type: vm ioctl
875:Parameters: struct kvm_irq_level
876:Returns: 0 on success, -1 on error
877
878Sets the level of a GSI input to the interrupt controller model in the kernel.
879On some architectures it is required that an interrupt controller model has
880been previously created with KVM_CREATE_IRQCHIP.  Note that edge-triggered
881interrupts require the level to be set to 1 and then back to 0.
882
883On real hardware, interrupt pins can be active-low or active-high.  This
884does not matter for the level field of struct kvm_irq_level: 1 always
885means active (asserted), 0 means inactive (deasserted).
886
887x86 allows the operating system to program the interrupt polarity
888(active-low/active-high) for level-triggered interrupts, and KVM used
889to consider the polarity.  However, due to bitrot in the handling of
890active-low interrupts, the above convention is now valid on x86 too.
891This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED.  Userspace
892should not present interrupts to the guest as active-low unless this
893capability is present (or unless it is not using the in-kernel irqchip,
894of course).
895
896
897arm64 can signal an interrupt either at the CPU level, or at the
898in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
899use PPIs designated for specific cpus.  The irq field is interpreted
900like this::
901
902  bits:  |  31 ... 28  | 27 ... 24 | 23  ... 16 | 15 ... 0 |
903  field: | vcpu2_index | irq_type  | vcpu_index |  irq_id  |
904
905The irq_type field has the following values:
906
907- KVM_ARM_IRQ_TYPE_CPU:
908	       out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
909- KVM_ARM_IRQ_TYPE_SPI:
910	       in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
911               (the vcpu_index field is ignored)
912- KVM_ARM_IRQ_TYPE_PPI:
913	       in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
914
915(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
916
917In both cases, level is used to assert/deassert the line.
918
919When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
920identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
921must be zero.
922
923Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions
924injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
925be used for a userspace interrupt controller.
926
927::
928
929  struct kvm_irq_level {
930	union {
931		__u32 irq;     /* GSI */
932		__s32 status;  /* not used for KVM_IRQ_LEVEL */
933	};
934	__u32 level;           /* 0 or 1 */
935  };
936
937
9384.26 KVM_GET_IRQCHIP
939--------------------
940
941:Capability: KVM_CAP_IRQCHIP
942:Architectures: x86
943:Type: vm ioctl
944:Parameters: struct kvm_irqchip (in/out)
945:Returns: 0 on success, -1 on error
946
947Reads the state of a kernel interrupt controller created with
948KVM_CREATE_IRQCHIP into a buffer provided by the caller.
949
950::
951
952  struct kvm_irqchip {
953	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
954	__u32 pad;
955        union {
956		char dummy[512];  /* reserving space */
957		struct kvm_pic_state pic;
958		struct kvm_ioapic_state ioapic;
959	} chip;
960  };
961
962
9634.27 KVM_SET_IRQCHIP
964--------------------
965
966:Capability: KVM_CAP_IRQCHIP
967:Architectures: x86
968:Type: vm ioctl
969:Parameters: struct kvm_irqchip (in)
970:Returns: 0 on success, -1 on error
971
972Sets the state of a kernel interrupt controller created with
973KVM_CREATE_IRQCHIP from a buffer provided by the caller.
974
975::
976
977  struct kvm_irqchip {
978	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
979	__u32 pad;
980        union {
981		char dummy[512];  /* reserving space */
982		struct kvm_pic_state pic;
983		struct kvm_ioapic_state ioapic;
984	} chip;
985  };
986
987
9884.28 KVM_XEN_HVM_CONFIG
989-----------------------
990
991:Capability: KVM_CAP_XEN_HVM
992:Architectures: x86
993:Type: vm ioctl
994:Parameters: struct kvm_xen_hvm_config (in)
995:Returns: 0 on success, -1 on error
996
997Sets the MSR that the Xen HVM guest uses to initialize its hypercall
998page, and provides the starting address and size of the hypercall
999blobs in userspace.  When the guest writes the MSR, kvm copies one
1000page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
1001memory.
1002
1003The MSR index must be in the range [0x40000000, 0x4fffffff], i.e. must reside
1004in the range that is unofficially reserved for use by hypervisors.  The min/max
1005values are enumerated via KVM_XEN_MSR_MIN_INDEX and KVM_XEN_MSR_MAX_INDEX.
1006
1007::
1008
1009  struct kvm_xen_hvm_config {
1010	__u32 flags;
1011	__u32 msr;
1012	__u64 blob_addr_32;
1013	__u64 blob_addr_64;
1014	__u8 blob_size_32;
1015	__u8 blob_size_64;
1016	__u8 pad2[30];
1017  };
1018
1019If certain flags are returned from the KVM_CAP_XEN_HVM check, they may
1020be set in the flags field of this ioctl:
1021
1022The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate
1023the contents of the hypercall page automatically; hypercalls will be
1024intercepted and passed to userspace through KVM_EXIT_XEN.  In this
1025case, all of the blob size and address fields must be zero.
1026
1027The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace
1028will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event
1029channel interrupts rather than manipulating the guest's shared_info
1030structures directly. This, in turn, may allow KVM to enable features
1031such as intercepting the SCHEDOP_poll hypercall to accelerate PV
1032spinlock operation for the guest. Userspace may still use the ioctl
1033to deliver events if it was advertised, even if userspace does not
1034send this indication that it will always do so
1035
1036No other flags are currently valid in the struct kvm_xen_hvm_config.
1037
10384.29 KVM_GET_CLOCK
1039------------------
1040
1041:Capability: KVM_CAP_ADJUST_CLOCK
1042:Architectures: x86
1043:Type: vm ioctl
1044:Parameters: struct kvm_clock_data (out)
1045:Returns: 0 on success, -1 on error
1046
1047Gets the current timestamp of kvmclock as seen by the current guest. In
1048conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
1049such as migration.
1050
1051When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
1052set of bits that KVM can return in struct kvm_clock_data's flag member.
1053
1054The following flags are defined:
1055
1056KVM_CLOCK_TSC_STABLE
1057  If set, the returned value is the exact kvmclock
1058  value seen by all VCPUs at the instant when KVM_GET_CLOCK was called.
1059  If clear, the returned value is simply CLOCK_MONOTONIC plus a constant
1060  offset; the offset can be modified with KVM_SET_CLOCK.  KVM will try
1061  to make all VCPUs follow this clock, but the exact value read by each
1062  VCPU could differ, because the host TSC is not stable.
1063
1064KVM_CLOCK_REALTIME
1065  If set, the `realtime` field in the kvm_clock_data
1066  structure is populated with the value of the host's real time
1067  clocksource at the instant when KVM_GET_CLOCK was called. If clear,
1068  the `realtime` field does not contain a value.
1069
1070KVM_CLOCK_HOST_TSC
1071  If set, the `host_tsc` field in the kvm_clock_data
1072  structure is populated with the value of the host's timestamp counter (TSC)
1073  at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field
1074  does not contain a value.
1075
1076::
1077
1078  struct kvm_clock_data {
1079	__u64 clock;  /* kvmclock current value */
1080	__u32 flags;
1081	__u32 pad0;
1082	__u64 realtime;
1083	__u64 host_tsc;
1084	__u32 pad[4];
1085  };
1086
1087
10884.30 KVM_SET_CLOCK
1089------------------
1090
1091:Capability: KVM_CAP_ADJUST_CLOCK
1092:Architectures: x86
1093:Type: vm ioctl
1094:Parameters: struct kvm_clock_data (in)
1095:Returns: 0 on success, -1 on error
1096
1097Sets the current timestamp of kvmclock to the value specified in its parameter.
1098In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1099such as migration.
1100
1101The following flags can be passed:
1102
1103KVM_CLOCK_REALTIME
1104  If set, KVM will compare the value of the `realtime` field
1105  with the value of the host's real time clocksource at the instant when
1106  KVM_SET_CLOCK was called. The difference in elapsed time is added to the final
1107  kvmclock value that will be provided to guests.
1108
1109Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored.
1110
1111::
1112
1113  struct kvm_clock_data {
1114	__u64 clock;  /* kvmclock current value */
1115	__u32 flags;
1116	__u32 pad0;
1117	__u64 realtime;
1118	__u64 host_tsc;
1119	__u32 pad[4];
1120  };
1121
1122
11234.31 KVM_GET_VCPU_EVENTS
1124------------------------
1125
1126:Capability: KVM_CAP_VCPU_EVENTS
1127:Extended by: KVM_CAP_INTR_SHADOW
1128:Architectures: x86, arm64
1129:Type: vcpu ioctl
1130:Parameters: struct kvm_vcpu_events (out)
1131:Returns: 0 on success, -1 on error
1132
1133X86:
1134^^^^
1135
1136Gets currently pending exceptions, interrupts, and NMIs as well as related
1137states of the vcpu.
1138
1139::
1140
1141  struct kvm_vcpu_events {
1142	struct {
1143		__u8 injected;
1144		__u8 nr;
1145		__u8 has_error_code;
1146		__u8 pending;
1147		__u32 error_code;
1148	} exception;
1149	struct {
1150		__u8 injected;
1151		__u8 nr;
1152		__u8 soft;
1153		__u8 shadow;
1154	} interrupt;
1155	struct {
1156		__u8 injected;
1157		__u8 pending;
1158		__u8 masked;
1159		__u8 pad;
1160	} nmi;
1161	__u32 sipi_vector;
1162	__u32 flags;
1163	struct {
1164		__u8 smm;
1165		__u8 pending;
1166		__u8 smm_inside_nmi;
1167		__u8 latched_init;
1168	} smi;
1169	__u8 reserved[27];
1170	__u8 exception_has_payload;
1171	__u64 exception_payload;
1172  };
1173
1174The following bits are defined in the flags field:
1175
1176- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1177  interrupt.shadow contains a valid state.
1178
1179- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1180  valid state.
1181
1182- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1183  exception_has_payload, exception_payload, and exception.pending
1184  fields contain a valid state. This bit will be set whenever
1185  KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1186
1187- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the
1188  triple_fault_pending field contains a valid state. This bit will
1189  be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled.
1190
1191ARM64:
1192^^^^^^
1193
1194If the guest accesses a device that is being emulated by the host kernel in
1195such a way that a real device would generate a physical SError, KVM may make
1196a virtual SError pending for that VCPU. This system error interrupt remains
1197pending until the guest takes the exception by unmasking PSTATE.A.
1198
1199Running the VCPU may cause it to take a pending SError, or make an access that
1200causes an SError to become pending. The event's description is only valid while
1201the VPCU is not running.
1202
1203This API provides a way to read and write the pending 'event' state that is not
1204visible to the guest. To save, restore or migrate a VCPU the struct representing
1205the state can be read then written using this GET/SET API, along with the other
1206guest-visible registers. It is not possible to 'cancel' an SError that has been
1207made pending.
1208
1209A device being emulated in user-space may also wish to generate an SError. To do
1210this the events structure can be populated by user-space. The current state
1211should be read first, to ensure no existing SError is pending. If an existing
1212SError is pending, the architecture's 'Multiple SError interrupts' rules should
1213be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1214Serviceability (RAS) Specification").
1215
1216SError exceptions always have an ESR value. Some CPUs have the ability to
1217specify what the virtual SError's ESR value should be. These systems will
1218advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1219always have a non-zero value when read, and the agent making an SError pending
1220should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1221the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1222with exception.has_esr as zero, KVM will choose an ESR.
1223
1224Specifying exception.has_esr on a system that does not support it will return
1225-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1226will return -EINVAL.
1227
1228It is not possible to read back a pending external abort (injected via
1229KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1230directly to the virtual CPU).
1231
1232Calling this ioctl on a vCPU that hasn't been initialized will return
1233-ENOEXEC.
1234
1235::
1236
1237  struct kvm_vcpu_events {
1238	struct {
1239		__u8 serror_pending;
1240		__u8 serror_has_esr;
1241		__u8 ext_dabt_pending;
1242		/* Align it to 8 bytes */
1243		__u8 pad[5];
1244		__u64 serror_esr;
1245	} exception;
1246	__u32 reserved[12];
1247  };
1248
12494.32 KVM_SET_VCPU_EVENTS
1250------------------------
1251
1252:Capability: KVM_CAP_VCPU_EVENTS
1253:Extended by: KVM_CAP_INTR_SHADOW
1254:Architectures: x86, arm64
1255:Type: vcpu ioctl
1256:Parameters: struct kvm_vcpu_events (in)
1257:Returns: 0 on success, -1 on error
1258
1259X86:
1260^^^^
1261
1262Set pending exceptions, interrupts, and NMIs as well as related states of the
1263vcpu.
1264
1265See KVM_GET_VCPU_EVENTS for the data structure.
1266
1267Fields that may be modified asynchronously by running VCPUs can be excluded
1268from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1269smi.pending. Keep the corresponding bits in the flags field cleared to
1270suppress overwriting the current in-kernel state. The bits are:
1271
1272===============================  ==================================
1273KVM_VCPUEVENT_VALID_NMI_PENDING  transfer nmi.pending to the kernel
1274KVM_VCPUEVENT_VALID_SIPI_VECTOR  transfer sipi_vector
1275KVM_VCPUEVENT_VALID_SMM          transfer the smi sub-struct.
1276===============================  ==================================
1277
1278If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1279the flags field to signal that interrupt.shadow contains a valid state and
1280shall be written into the VCPU.
1281
1282KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1283
1284If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1285can be set in the flags field to signal that the
1286exception_has_payload, exception_payload, and exception.pending fields
1287contain a valid state and shall be written into the VCPU.
1288
1289If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT
1290can be set in flags field to signal that the triple_fault field contains
1291a valid state and shall be written into the VCPU.
1292
1293ARM64:
1294^^^^^^
1295
1296User space may need to inject several types of events to the guest.
1297
1298Set the pending SError exception state for this VCPU. It is not possible to
1299'cancel' an Serror that has been made pending.
1300
1301If the guest performed an access to I/O memory which could not be handled by
1302userspace, for example because of missing instruction syndrome decode
1303information or because there is no device mapped at the accessed IPA, then
1304userspace can ask the kernel to inject an external abort using the address
1305from the exiting fault on the VCPU. It is a programming error to set
1306ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO,
1307KVM_EXIT_ARM_NISV, or KVM_EXIT_ARM_LDST64B. This feature is only available if
1308the system supports KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which
1309provides commonality in how userspace reports accesses for the above cases to
1310guests, across different userspace implementations. Nevertheless, userspace
1311can still emulate all Arm exceptions by manipulating individual registers
1312using the KVM_SET_ONE_REG API.
1313
1314See KVM_GET_VCPU_EVENTS for the data structure.
1315
1316Calling this ioctl on a vCPU that hasn't been initialized will return
1317-ENOEXEC.
1318
13194.33 KVM_GET_DEBUGREGS
1320----------------------
1321
1322:Capability: KVM_CAP_DEBUGREGS
1323:Architectures: x86
1324:Type: vcpu ioctl
1325:Parameters: struct kvm_debugregs (out)
1326:Returns: 0 on success, -1 on error
1327
1328Reads debug registers from the vcpu.
1329
1330::
1331
1332  struct kvm_debugregs {
1333	__u64 db[4];
1334	__u64 dr6;
1335	__u64 dr7;
1336	__u64 flags;
1337	__u64 reserved[9];
1338  };
1339
1340
13414.34 KVM_SET_DEBUGREGS
1342----------------------
1343
1344:Capability: KVM_CAP_DEBUGREGS
1345:Architectures: x86
1346:Type: vcpu ioctl
1347:Parameters: struct kvm_debugregs (in)
1348:Returns: 0 on success, -1 on error
1349
1350Writes debug registers into the vcpu.
1351
1352See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1353yet and must be cleared on entry.
1354
1355
13564.35 KVM_SET_USER_MEMORY_REGION
1357-------------------------------
1358
1359:Capability: KVM_CAP_USER_MEMORY
1360:Architectures: all
1361:Type: vm ioctl
1362:Parameters: struct kvm_userspace_memory_region (in)
1363:Returns: 0 on success, -1 on error
1364
1365::
1366
1367  struct kvm_userspace_memory_region {
1368	__u32 slot;
1369	__u32 flags;
1370	__u64 guest_phys_addr;
1371	__u64 memory_size; /* bytes */
1372	__u64 userspace_addr; /* start of the userspace allocated memory */
1373  };
1374
1375  /* for kvm_userspace_memory_region::flags */
1376  #define KVM_MEM_LOG_DIRTY_PAGES	(1UL << 0)
1377  #define KVM_MEM_READONLY	(1UL << 1)
1378
1379This ioctl allows the user to create, modify or delete a guest physical
1380memory slot.  Bits 0-15 of "slot" specify the slot id and this value
1381should be less than the maximum number of user memory slots supported per
1382VM.  The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1383Slots may not overlap in guest physical address space.
1384
1385If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1386specifies the address space which is being modified.  They must be
1387less than the value that KVM_CHECK_EXTENSION returns for the
1388KVM_CAP_MULTI_ADDRESS_SPACE capability.  Slots in separate address spaces
1389are unrelated; the restriction on overlapping slots only applies within
1390each address space.
1391
1392Deleting a slot is done by passing zero for memory_size.  When changing
1393an existing slot, it may be moved in the guest physical memory space,
1394or its flags may be modified, but it may not be resized.
1395
1396Memory for the region is taken starting at the address denoted by the
1397field userspace_addr, which must point at user addressable memory for
1398the entire memory slot size.  Any object may back this memory, including
1399anonymous memory, ordinary files, and hugetlbfs.  Changes in the backing
1400of the memory region are automatically reflected into the guest.
1401For example, an mmap() that affects the region will be made visible
1402immediately.  Another example is madvise(MADV_DROP).
1403
1404On architectures that support a form of address tagging, userspace_addr must
1405be an untagged address.
1406
1407It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1408be identical.  This allows large pages in the guest to be backed by large
1409pages in the host.
1410
1411The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1412KVM_MEM_READONLY.  The former can be set to instruct KVM to keep track of
1413writes to memory within the slot.  See KVM_GET_DIRTY_LOG ioctl to know how to
1414use it.  The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1415to make a new slot read-only.  In this case, writes to this memory will be
1416posted to userspace as KVM_EXIT_MMIO exits.
1417
1418For TDX guest, deleting/moving memory region loses guest memory contents.
1419Read only region isn't supported.  Only as-id 0 is supported.
1420
1421Note: On arm64, a write generated by the page-table walker (to update
1422the Access and Dirty flags, for example) never results in a
1423KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This
1424is because KVM cannot provide the data that would be written by the
1425page-table walker, making it impossible to emulate the access.
1426Instead, an abort (data abort if the cause of the page-table update
1427was a load or a store, instruction abort if it was an instruction
1428fetch) is injected in the guest.
1429
1430S390:
1431^^^^^
1432
1433Returns -EINVAL or -EEXIST if the VM has the KVM_VM_S390_UCONTROL flag set.
1434Returns -EINVAL if called on a protected VM.
1435
14364.36 KVM_SET_TSS_ADDR
1437---------------------
1438
1439:Capability: KVM_CAP_SET_TSS_ADDR
1440:Architectures: x86
1441:Type: vm ioctl
1442:Parameters: unsigned long tss_address (in)
1443:Returns: 0 on success, -1 on error
1444
1445This ioctl defines the physical address of a three-page region in the guest
1446physical address space.  The region must be within the first 4GB of the
1447guest physical address space and must not conflict with any memory slot
1448or any mmio address.  The guest may malfunction if it accesses this memory
1449region.
1450
1451This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1452because of a quirk in the virtualization implementation (see the internals
1453documentation when it pops into existence).
1454
1455
1456.. _KVM_ENABLE_CAP:
1457
14584.37 KVM_ENABLE_CAP
1459-------------------
1460
1461:Capability: KVM_CAP_ENABLE_CAP
1462:Architectures: mips, ppc, s390, x86, loongarch
1463:Type: vcpu ioctl
1464:Parameters: struct kvm_enable_cap (in)
1465:Returns: 0 on success; -1 on error
1466
1467:Capability: KVM_CAP_ENABLE_CAP_VM
1468:Architectures: all
1469:Type: vm ioctl
1470:Parameters: struct kvm_enable_cap (in)
1471:Returns: 0 on success; -1 on error
1472
1473.. note::
1474
1475   Not all extensions are enabled by default. Using this ioctl the application
1476   can enable an extension, making it available to the guest.
1477
1478On systems that do not support this ioctl, it always fails. On systems that
1479do support it, it only works for extensions that are supported for enablement.
1480
1481To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1482be used.
1483
1484::
1485
1486  struct kvm_enable_cap {
1487       /* in */
1488       __u32 cap;
1489
1490The capability that is supposed to get enabled.
1491
1492::
1493
1494       __u32 flags;
1495
1496A bitfield indicating future enhancements. Has to be 0 for now.
1497
1498::
1499
1500       __u64 args[4];
1501
1502Arguments for enabling a feature. If a feature needs initial values to
1503function properly, this is the place to put them.
1504
1505::
1506
1507       __u8  pad[64];
1508  };
1509
1510The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1511for vm-wide capabilities.
1512
15134.38 KVM_GET_MP_STATE
1514---------------------
1515
1516:Capability: KVM_CAP_MP_STATE
1517:Architectures: x86, s390, arm64, riscv, loongarch
1518:Type: vcpu ioctl
1519:Parameters: struct kvm_mp_state (out)
1520:Returns: 0 on success; -1 on error
1521
1522::
1523
1524  struct kvm_mp_state {
1525	__u32 mp_state;
1526  };
1527
1528Returns the vcpu's current "multiprocessing state" (though also valid on
1529uniprocessor guests).
1530
1531Possible values are:
1532
1533   ==========================    ===============================================
1534   KVM_MP_STATE_RUNNABLE         the vcpu is currently running
1535                                 [x86,arm64,riscv,loongarch]
1536   KVM_MP_STATE_UNINITIALIZED    the vcpu is an application processor (AP)
1537                                 which has not yet received an INIT signal [x86]
1538   KVM_MP_STATE_INIT_RECEIVED    the vcpu has received an INIT signal, and is
1539                                 now ready for a SIPI [x86]
1540   KVM_MP_STATE_HALTED           the vcpu has executed a HLT instruction and
1541                                 is waiting for an interrupt [x86]
1542   KVM_MP_STATE_SIPI_RECEIVED    the vcpu has just received a SIPI (vector
1543                                 accessible via KVM_GET_VCPU_EVENTS) [x86]
1544   KVM_MP_STATE_STOPPED          the vcpu is stopped [s390,arm64,riscv]
1545   KVM_MP_STATE_CHECK_STOP       the vcpu is in a special error state [s390]
1546   KVM_MP_STATE_OPERATING        the vcpu is operating (running or halted)
1547                                 [s390]
1548   KVM_MP_STATE_LOAD             the vcpu is in a special load/startup state
1549                                 [s390]
1550   KVM_MP_STATE_SUSPENDED        the vcpu is in a suspend state and is waiting
1551                                 for a wakeup event [arm64]
1552   ==========================    ===============================================
1553
1554On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1555in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1556these architectures.
1557
1558For arm64:
1559^^^^^^^^^^
1560
1561If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the
1562architectural execution of a WFI instruction.
1563
1564If a wakeup event is recognized, KVM will exit to userspace with a
1565KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If
1566userspace wants to honor the wakeup, it must set the vCPU's MP state to
1567KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup
1568event in subsequent calls to KVM_RUN.
1569
1570.. warning::
1571
1572     If userspace intends to keep the vCPU in a SUSPENDED state, it is
1573     strongly recommended that userspace take action to suppress the
1574     wakeup event (such as masking an interrupt). Otherwise, subsequent
1575     calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP
1576     event and inadvertently waste CPU cycles.
1577
1578     Additionally, if userspace takes action to suppress a wakeup event,
1579     it is strongly recommended that it also restores the vCPU to its
1580     original state when the vCPU is made RUNNABLE again. For example,
1581     if userspace masked a pending interrupt to suppress the wakeup,
1582     the interrupt should be unmasked before returning control to the
1583     guest.
1584
1585For riscv:
1586^^^^^^^^^^
1587
1588The only states that are valid are KVM_MP_STATE_STOPPED and
1589KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1590
1591On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect
1592whether the vcpu is runnable.
1593
15944.39 KVM_SET_MP_STATE
1595---------------------
1596
1597:Capability: KVM_CAP_MP_STATE
1598:Architectures: x86, s390, arm64, riscv, loongarch
1599:Type: vcpu ioctl
1600:Parameters: struct kvm_mp_state (in)
1601:Returns: 0 on success; -1 on error
1602
1603Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1604arguments.
1605
1606On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1607in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1608these architectures.
1609
1610For arm64/riscv:
1611^^^^^^^^^^^^^^^^
1612
1613The only states that are valid are KVM_MP_STATE_STOPPED and
1614KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1615
1616On LoongArch, only the KVM_MP_STATE_RUNNABLE state is used to reflect
1617whether the vcpu is runnable.
1618
16194.40 KVM_SET_IDENTITY_MAP_ADDR
1620------------------------------
1621
1622:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1623:Architectures: x86
1624:Type: vm ioctl
1625:Parameters: unsigned long identity (in)
1626:Returns: 0 on success, -1 on error
1627
1628This ioctl defines the physical address of a one-page region in the guest
1629physical address space.  The region must be within the first 4GB of the
1630guest physical address space and must not conflict with any memory slot
1631or any mmio address.  The guest may malfunction if it accesses this memory
1632region.
1633
1634Setting the address to 0 will result in resetting the address to its default
1635(0xfffbc000).
1636
1637This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1638because of a quirk in the virtualization implementation (see the internals
1639documentation when it pops into existence).
1640
1641Fails if any VCPU has already been created.
1642
16434.41 KVM_SET_BOOT_CPU_ID
1644------------------------
1645
1646:Capability: KVM_CAP_SET_BOOT_CPU_ID
1647:Architectures: x86
1648:Type: vm ioctl
1649:Parameters: unsigned long vcpu_id
1650:Returns: 0 on success, -1 on error
1651
1652Define which vcpu is the Bootstrap Processor (BSP).  Values are the same
1653as the vcpu id in KVM_CREATE_VCPU.  If this ioctl is not called, the default
1654is vcpu 0. This ioctl has to be called before vcpu creation,
1655otherwise it will return EBUSY error.
1656
1657
16584.42 KVM_GET_XSAVE
1659------------------
1660
1661:Capability: KVM_CAP_XSAVE
1662:Architectures: x86
1663:Type: vcpu ioctl
1664:Parameters: struct kvm_xsave (out)
1665:Returns: 0 on success, -1 on error
1666
1667
1668::
1669
1670  struct kvm_xsave {
1671	__u32 region[1024];
1672	__u32 extra[0];
1673  };
1674
1675This ioctl would copy current vcpu's xsave struct to the userspace.
1676
1677
16784.43 KVM_SET_XSAVE
1679------------------
1680
1681:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2
1682:Architectures: x86
1683:Type: vcpu ioctl
1684:Parameters: struct kvm_xsave (in)
1685:Returns: 0 on success, -1 on error
1686
1687::
1688
1689
1690  struct kvm_xsave {
1691	__u32 region[1024];
1692	__u32 extra[0];
1693  };
1694
1695This ioctl would copy userspace's xsave struct to the kernel. It copies
1696as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2),
1697when invoked on the vm file descriptor. The size value returned by
1698KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
1699Currently, it is only greater than 4096 if a dynamic feature has been
1700enabled with ``arch_prctl()``, but this may change in the future.
1701
1702The offsets of the state save areas in struct kvm_xsave follow the
1703contents of CPUID leaf 0xD on the host.
1704
1705
17064.44 KVM_GET_XCRS
1707-----------------
1708
1709:Capability: KVM_CAP_XCRS
1710:Architectures: x86
1711:Type: vcpu ioctl
1712:Parameters: struct kvm_xcrs (out)
1713:Returns: 0 on success, -1 on error
1714
1715::
1716
1717  struct kvm_xcr {
1718	__u32 xcr;
1719	__u32 reserved;
1720	__u64 value;
1721  };
1722
1723  struct kvm_xcrs {
1724	__u32 nr_xcrs;
1725	__u32 flags;
1726	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1727	__u64 padding[16];
1728  };
1729
1730This ioctl would copy current vcpu's xcrs to the userspace.
1731
1732
17334.45 KVM_SET_XCRS
1734-----------------
1735
1736:Capability: KVM_CAP_XCRS
1737:Architectures: x86
1738:Type: vcpu ioctl
1739:Parameters: struct kvm_xcrs (in)
1740:Returns: 0 on success, -1 on error
1741
1742::
1743
1744  struct kvm_xcr {
1745	__u32 xcr;
1746	__u32 reserved;
1747	__u64 value;
1748  };
1749
1750  struct kvm_xcrs {
1751	__u32 nr_xcrs;
1752	__u32 flags;
1753	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1754	__u64 padding[16];
1755  };
1756
1757This ioctl would set vcpu's xcr to the value userspace specified.
1758
1759
17604.46 KVM_GET_SUPPORTED_CPUID
1761----------------------------
1762
1763:Capability: KVM_CAP_EXT_CPUID
1764:Architectures: x86
1765:Type: system ioctl
1766:Parameters: struct kvm_cpuid2 (in/out)
1767:Returns: 0 on success, -1 on error
1768
1769::
1770
1771  struct kvm_cpuid2 {
1772	__u32 nent;
1773	__u32 padding;
1774	struct kvm_cpuid_entry2 entries[0];
1775  };
1776
1777  #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
1778  #define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1) /* deprecated */
1779  #define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2) /* deprecated */
1780
1781  struct kvm_cpuid_entry2 {
1782	__u32 function;
1783	__u32 index;
1784	__u32 flags;
1785	__u32 eax;
1786	__u32 ebx;
1787	__u32 ecx;
1788	__u32 edx;
1789	__u32 padding[3];
1790  };
1791
1792This ioctl returns x86 cpuid features which are supported by both the
1793hardware and kvm in its default configuration.  Userspace can use the
1794information returned by this ioctl to construct cpuid information (for
1795KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1796userspace capabilities, and with user requirements (for example, the
1797user may wish to constrain cpuid to emulate older hardware, or for
1798feature consistency across a cluster).
1799
1800Dynamically-enabled feature bits need to be requested with
1801``arch_prctl()`` before calling this ioctl. Feature bits that have not
1802been requested are excluded from the result.
1803
1804Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1805expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1806its default configuration. If userspace enables such capabilities, it
1807is responsible for modifying the results of this ioctl appropriately.
1808
1809Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1810with the 'nent' field indicating the number of entries in the variable-size
1811array 'entries'.  If the number of entries is too low to describe the cpu
1812capabilities, an error (E2BIG) is returned.  If the number is too high,
1813the 'nent' field is adjusted and an error (ENOMEM) is returned.  If the
1814number is just right, the 'nent' field is adjusted to the number of valid
1815entries in the 'entries' array, which is then filled.
1816
1817The entries returned are the host cpuid as returned by the cpuid instruction,
1818with unknown or unsupported features masked out.  Some features (for example,
1819x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1820emulate them efficiently. The fields in each entry are defined as follows:
1821
1822  function:
1823         the eax value used to obtain the entry
1824
1825  index:
1826         the ecx value used to obtain the entry (for entries that are
1827         affected by ecx)
1828
1829  flags:
1830     an OR of zero or more of the following:
1831
1832        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1833           if the index field is valid
1834
1835   eax, ebx, ecx, edx:
1836         the values returned by the cpuid instruction for
1837         this function/index combination
1838
1839x2APIC (CPUID leaf 1, ecx[21) and TSC deadline timer (CPUID leaf 1, ecx[24])
1840may be returned as true, but they depend on KVM_CREATE_IRQCHIP for in-kernel
1841emulation of the local APIC.  TSC deadline timer support is also reported via::
1842
1843  ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1844
1845if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1846feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1847
1848Enabling x2APIC in KVM_SET_CPUID2 requires KVM_CREATE_IRQCHIP as KVM doesn't
1849support forwarding x2APIC MSR accesses to userspace, i.e. KVM does not support
1850emulating x2APIC in userspace.
1851
18524.47 KVM_PPC_GET_PVINFO
1853-----------------------
1854
1855:Capability: KVM_CAP_PPC_GET_PVINFO
1856:Architectures: ppc
1857:Type: vm ioctl
1858:Parameters: struct kvm_ppc_pvinfo (out)
1859:Returns: 0 on success, !0 on error
1860
1861::
1862
1863  struct kvm_ppc_pvinfo {
1864	__u32 flags;
1865	__u32 hcall[4];
1866	__u8  pad[108];
1867  };
1868
1869This ioctl fetches PV specific information that need to be passed to the guest
1870using the device tree or other means from vm context.
1871
1872The hcall array defines 4 instructions that make up a hypercall.
1873
1874If any additional field gets added to this structure later on, a bit for that
1875additional piece of information will be set in the flags bitmap.
1876
1877The flags bitmap is defined as::
1878
1879   /* the host supports the ePAPR idle hcall
1880   #define KVM_PPC_PVINFO_FLAGS_EV_IDLE   (1<<0)
1881
18824.52 KVM_SET_GSI_ROUTING
1883------------------------
1884
1885:Capability: KVM_CAP_IRQ_ROUTING
1886:Architectures: x86 s390 arm64
1887:Type: vm ioctl
1888:Parameters: struct kvm_irq_routing (in)
1889:Returns: 0 on success, -1 on error
1890
1891Sets the GSI routing table entries, overwriting any previously set entries.
1892
1893On arm64, GSI routing has the following limitation:
1894
1895- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1896
1897::
1898
1899  struct kvm_irq_routing {
1900	__u32 nr;
1901	__u32 flags;
1902	struct kvm_irq_routing_entry entries[0];
1903  };
1904
1905No flags are specified so far, the corresponding field must be set to zero.
1906
1907::
1908
1909  struct kvm_irq_routing_entry {
1910	__u32 gsi;
1911	__u32 type;
1912	__u32 flags;
1913	__u32 pad;
1914	union {
1915		struct kvm_irq_routing_irqchip irqchip;
1916		struct kvm_irq_routing_msi msi;
1917		struct kvm_irq_routing_s390_adapter adapter;
1918		struct kvm_irq_routing_hv_sint hv_sint;
1919		struct kvm_irq_routing_xen_evtchn xen_evtchn;
1920		__u32 pad[8];
1921	} u;
1922  };
1923
1924  /* gsi routing entry types */
1925  #define KVM_IRQ_ROUTING_IRQCHIP 1
1926  #define KVM_IRQ_ROUTING_MSI 2
1927  #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1928  #define KVM_IRQ_ROUTING_HV_SINT 4
1929  #define KVM_IRQ_ROUTING_XEN_EVTCHN 5
1930
1931On s390, adding a KVM_IRQ_ROUTING_S390_ADAPTER is rejected on ucontrol VMs with
1932error -EINVAL.
1933
1934flags:
1935
1936- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1937  type, specifies that the devid field contains a valid value.  The per-VM
1938  KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1939  the device ID.  If this capability is not available, userspace should
1940  never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1941- zero otherwise
1942
1943::
1944
1945  struct kvm_irq_routing_irqchip {
1946	__u32 irqchip;
1947	__u32 pin;
1948  };
1949
1950  struct kvm_irq_routing_msi {
1951	__u32 address_lo;
1952	__u32 address_hi;
1953	__u32 data;
1954	union {
1955		__u32 pad;
1956		__u32 devid;
1957	};
1958  };
1959
1960If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1961for the device that wrote the MSI message.  For PCI, this is usually a
1962BDF identifier in the lower 16 bits.
1963
1964On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1965feature of KVM_CAP_X2APIC_API capability is enabled.  If it is enabled,
1966address_hi bits 31-8 provide bits 31-8 of the destination id.  Bits 7-0 of
1967address_hi must be zero.
1968
1969::
1970
1971  struct kvm_irq_routing_s390_adapter {
1972	__u64 ind_addr;
1973	__u64 summary_addr;
1974	__u64 ind_offset;
1975	__u32 summary_offset;
1976	__u32 adapter_id;
1977  };
1978
1979  struct kvm_irq_routing_hv_sint {
1980	__u32 vcpu;
1981	__u32 sint;
1982  };
1983
1984  struct kvm_irq_routing_xen_evtchn {
1985	__u32 port;
1986	__u32 vcpu;
1987	__u32 priority;
1988  };
1989
1990
1991When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit
1992in its indication of supported features, routing to Xen event channels
1993is supported. Although the priority field is present, only the value
1994KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by
19952 level event channels. FIFO event channel support may be added in
1996the future.
1997
1998
19994.55 KVM_SET_TSC_KHZ
2000--------------------
2001
2002:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL
2003:Architectures: x86
2004:Type: vcpu ioctl / vm ioctl
2005:Parameters: virtual tsc_khz
2006:Returns: 0 on success, -1 on error
2007
2008Specifies the tsc frequency for the virtual machine. The unit of the
2009frequency is KHz.
2010
2011If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also
2012be used as a vm ioctl to set the initial tsc frequency of subsequently
2013created vCPUs.  Note, the vm ioctl is only allowed prior to creating vCPUs.
2014
2015For TSC protected Confidential Computing (CoCo) VMs where TSC frequency
2016is configured once at VM scope and remains unchanged during VM's
2017lifetime, the vm ioctl should be used to configure the TSC frequency
2018and the vcpu ioctl is not supported.
2019
2020Example of such CoCo VMs: TDX guests.
2021
20224.56 KVM_GET_TSC_KHZ
2023--------------------
2024
2025:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL
2026:Architectures: x86
2027:Type: vcpu ioctl / vm ioctl
2028:Parameters: none
2029:Returns: virtual tsc-khz on success, negative value on error
2030
2031Returns the tsc frequency of the guest. The unit of the return value is
2032KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
2033error.
2034
2035
20364.57 KVM_GET_LAPIC
2037------------------
2038
2039:Capability: KVM_CAP_IRQCHIP
2040:Architectures: x86
2041:Type: vcpu ioctl
2042:Parameters: struct kvm_lapic_state (out)
2043:Returns: 0 on success, -1 on error
2044
2045::
2046
2047  #define KVM_APIC_REG_SIZE 0x400
2048  struct kvm_lapic_state {
2049	char regs[KVM_APIC_REG_SIZE];
2050  };
2051
2052Reads the Local APIC registers and copies them into the input argument.  The
2053data format and layout are the same as documented in the architecture manual.
2054
2055If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
2056enabled, then the format of APIC_ID register depends on the APIC mode
2057(reported by MSR_IA32_APICBASE) of its VCPU.  x2APIC stores APIC ID in
2058the APIC_ID register (bytes 32-35).  xAPIC only allows an 8-bit APIC ID
2059which is stored in bits 31-24 of the APIC register, or equivalently in
2060byte 35 of struct kvm_lapic_state's regs field.  KVM_GET_LAPIC must then
2061be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
2062
2063If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
2064always uses xAPIC format.
2065
2066
20674.58 KVM_SET_LAPIC
2068------------------
2069
2070:Capability: KVM_CAP_IRQCHIP
2071:Architectures: x86
2072:Type: vcpu ioctl
2073:Parameters: struct kvm_lapic_state (in)
2074:Returns: 0 on success, -1 on error
2075
2076::
2077
2078  #define KVM_APIC_REG_SIZE 0x400
2079  struct kvm_lapic_state {
2080	char regs[KVM_APIC_REG_SIZE];
2081  };
2082
2083Copies the input argument into the Local APIC registers.  The data format
2084and layout are the same as documented in the architecture manual.
2085
2086The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
2087regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
2088See the note in KVM_GET_LAPIC.
2089
2090
20914.59 KVM_IOEVENTFD
2092------------------
2093
2094:Capability: KVM_CAP_IOEVENTFD
2095:Architectures: all
2096:Type: vm ioctl
2097:Parameters: struct kvm_ioeventfd (in)
2098:Returns: 0 on success, !0 on error
2099
2100This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
2101within the guest.  A guest write in the registered address will signal the
2102provided event instead of triggering an exit.
2103
2104::
2105
2106  struct kvm_ioeventfd {
2107	__u64 datamatch;
2108	__u64 addr;        /* legal pio/mmio address */
2109	__u32 len;         /* 0, 1, 2, 4, or 8 bytes    */
2110	__s32 fd;
2111	__u32 flags;
2112	__u8  pad[36];
2113  };
2114
2115For the special case of virtio-ccw devices on s390, the ioevent is matched
2116to a subchannel/virtqueue tuple instead.
2117
2118The following flags are defined::
2119
2120  #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
2121  #define KVM_IOEVENTFD_FLAG_PIO       (1 << kvm_ioeventfd_flag_nr_pio)
2122  #define KVM_IOEVENTFD_FLAG_DEASSIGN  (1 << kvm_ioeventfd_flag_nr_deassign)
2123  #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2124	(1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2125
2126If datamatch flag is set, the event will be signaled only if the written value
2127to the registered address is equal to datamatch in struct kvm_ioeventfd.
2128
2129For virtio-ccw devices, addr contains the subchannel id and datamatch the
2130virtqueue index.
2131
2132With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2133the kernel will ignore the length of guest write and may get a faster vmexit.
2134The speedup may only apply to specific architectures, but the ioeventfd will
2135work anyway.
2136
21374.60 KVM_DIRTY_TLB
2138------------------
2139
2140:Capability: KVM_CAP_SW_TLB
2141:Architectures: ppc
2142:Type: vcpu ioctl
2143:Parameters: struct kvm_dirty_tlb (in)
2144:Returns: 0 on success, -1 on error
2145
2146::
2147
2148  struct kvm_dirty_tlb {
2149	__u64 bitmap;
2150	__u32 num_dirty;
2151  };
2152
2153This must be called whenever userspace has changed an entry in the shared
2154TLB, prior to calling KVM_RUN on the associated vcpu.
2155
2156The "bitmap" field is the userspace address of an array.  This array
2157consists of a number of bits, equal to the total number of TLB entries as
2158determined by the last successful call to ``KVM_ENABLE_CAP(KVM_CAP_SW_TLB)``,
2159rounded up to the nearest multiple of 64.
2160
2161Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2162array.
2163
2164The array is little-endian: the bit 0 is the least significant bit of the
2165first byte, bit 8 is the least significant bit of the second byte, etc.
2166This avoids any complications with differing word sizes.
2167
2168The "num_dirty" field is a performance hint for KVM to determine whether it
2169should skip processing the bitmap and just invalidate everything.  It must
2170be set to the number of set bits in the bitmap.
2171
2172
21734.62 KVM_CREATE_SPAPR_TCE
2174-------------------------
2175
2176:Capability: KVM_CAP_SPAPR_TCE
2177:Architectures: powerpc
2178:Type: vm ioctl
2179:Parameters: struct kvm_create_spapr_tce (in)
2180:Returns: file descriptor for manipulating the created TCE table
2181
2182This creates a virtual TCE (translation control entry) table, which
2183is an IOMMU for PAPR-style virtual I/O.  It is used to translate
2184logical addresses used in virtual I/O into guest physical addresses,
2185and provides a scatter/gather capability for PAPR virtual I/O.
2186
2187::
2188
2189  /* for KVM_CAP_SPAPR_TCE */
2190  struct kvm_create_spapr_tce {
2191	__u64 liobn;
2192	__u32 window_size;
2193  };
2194
2195The liobn field gives the logical IO bus number for which to create a
2196TCE table.  The window_size field specifies the size of the DMA window
2197which this TCE table will translate - the table will contain one 64
2198bit TCE entry for every 4kiB of the DMA window.
2199
2200When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2201table has been created using this ioctl(), the kernel will handle it
2202in real mode, updating the TCE table.  H_PUT_TCE calls for other
2203liobns will cause a vm exit and must be handled by userspace.
2204
2205The return value is a file descriptor which can be passed to mmap(2)
2206to map the created TCE table into userspace.  This lets userspace read
2207the entries written by kernel-handled H_PUT_TCE calls, and also lets
2208userspace update the TCE table directly which is useful in some
2209circumstances.
2210
2211
22124.64 KVM_NMI
2213------------
2214
2215:Capability: KVM_CAP_USER_NMI
2216:Architectures: x86
2217:Type: vcpu ioctl
2218:Parameters: none
2219:Returns: 0 on success, -1 on error
2220
2221Queues an NMI on the thread's vcpu.  Note this is well defined only
2222when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2223between the virtual cpu core and virtual local APIC.  After KVM_CREATE_IRQCHIP
2224has been called, this interface is completely emulated within the kernel.
2225
2226To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2227following algorithm:
2228
2229  - pause the vcpu
2230  - read the local APIC's state (KVM_GET_LAPIC)
2231  - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2232  - if so, issue KVM_NMI
2233  - resume the vcpu
2234
2235Some guests configure the LINT1 NMI input to cause a panic, aiding in
2236debugging.
2237
2238
22394.65 KVM_S390_UCAS_MAP
2240----------------------
2241
2242:Capability: KVM_CAP_S390_UCONTROL
2243:Architectures: s390
2244:Type: vcpu ioctl
2245:Parameters: struct kvm_s390_ucas_mapping (in)
2246:Returns: 0 in case of success
2247
2248The parameter is defined like this::
2249
2250	struct kvm_s390_ucas_mapping {
2251		__u64 user_addr;
2252		__u64 vcpu_addr;
2253		__u64 length;
2254	};
2255
2256This ioctl maps the memory at "user_addr" with the length "length" to
2257the vcpu's address space starting at "vcpu_addr". All parameters need to
2258be aligned by 1 megabyte.
2259
2260
22614.66 KVM_S390_UCAS_UNMAP
2262------------------------
2263
2264:Capability: KVM_CAP_S390_UCONTROL
2265:Architectures: s390
2266:Type: vcpu ioctl
2267:Parameters: struct kvm_s390_ucas_mapping (in)
2268:Returns: 0 in case of success
2269
2270The parameter is defined like this::
2271
2272	struct kvm_s390_ucas_mapping {
2273		__u64 user_addr;
2274		__u64 vcpu_addr;
2275		__u64 length;
2276	};
2277
2278This ioctl unmaps the memory in the vcpu's address space starting at
2279"vcpu_addr" with the length "length". The field "user_addr" is ignored.
2280All parameters need to be aligned by 1 megabyte.
2281
2282
22834.67 KVM_S390_VCPU_FAULT
2284------------------------
2285
2286:Capability: KVM_CAP_S390_UCONTROL
2287:Architectures: s390
2288:Type: vcpu ioctl
2289:Parameters: vcpu absolute address (in)
2290:Returns: 0 in case of success
2291
2292This call creates a page table entry on the virtual cpu's address space
2293(for user controlled virtual machines) or the virtual machine's address
2294space (for regular virtual machines). This only works for minor faults,
2295thus it's recommended to access subject memory page via the user page
2296table upfront. This is useful to handle validity intercepts for user
2297controlled virtual machines to fault in the virtual cpu's lowcore pages
2298prior to calling the KVM_RUN ioctl.
2299
2300
23014.68 KVM_SET_ONE_REG
2302--------------------
2303
2304:Capability: KVM_CAP_ONE_REG
2305:Architectures: all
2306:Type: vcpu ioctl
2307:Parameters: struct kvm_one_reg (in)
2308:Returns: 0 on success, negative value on failure
2309
2310Errors:
2311
2312  ======   ============================================================
2313  ENOENT   no such register
2314  EINVAL   invalid register ID, or no such register or used with VMs in
2315           protected virtualization mode on s390
2316  EPERM    (arm64) register access not allowed before vcpu finalization
2317  EBUSY    (riscv) changing register value not allowed after the vcpu
2318           has run at least once
2319  ======   ============================================================
2320
2321(These error codes are indicative only: do not rely on a specific error
2322code being returned in a specific situation.)
2323
2324::
2325
2326  struct kvm_one_reg {
2327       __u64 id;
2328       __u64 addr;
2329 };
2330
2331Using this ioctl, a single vcpu register can be set to a specific value
2332defined by user space with the passed in struct kvm_one_reg, where id
2333refers to the register identifier as described below and addr is a pointer
2334to a variable with the respective size. There can be architecture agnostic
2335and architecture specific registers. Each have their own range of operation
2336and their own constants and width. To keep track of the implemented
2337registers, find a list below:
2338
2339  ======= =============================== ============
2340  Arch              Register              Width (bits)
2341  ======= =============================== ============
2342  PPC     KVM_REG_PPC_HIOR                64
2343  PPC     KVM_REG_PPC_IAC1                64
2344  PPC     KVM_REG_PPC_IAC2                64
2345  PPC     KVM_REG_PPC_IAC3                64
2346  PPC     KVM_REG_PPC_IAC4                64
2347  PPC     KVM_REG_PPC_DAC1                64
2348  PPC     KVM_REG_PPC_DAC2                64
2349  PPC     KVM_REG_PPC_DABR                64
2350  PPC     KVM_REG_PPC_DSCR                64
2351  PPC     KVM_REG_PPC_PURR                64
2352  PPC     KVM_REG_PPC_SPURR               64
2353  PPC     KVM_REG_PPC_DAR                 64
2354  PPC     KVM_REG_PPC_DSISR               32
2355  PPC     KVM_REG_PPC_AMR                 64
2356  PPC     KVM_REG_PPC_UAMOR               64
2357  PPC     KVM_REG_PPC_MMCR0               64
2358  PPC     KVM_REG_PPC_MMCR1               64
2359  PPC     KVM_REG_PPC_MMCRA               64
2360  PPC     KVM_REG_PPC_MMCR2               64
2361  PPC     KVM_REG_PPC_MMCRS               64
2362  PPC     KVM_REG_PPC_MMCR3               64
2363  PPC     KVM_REG_PPC_SIAR                64
2364  PPC     KVM_REG_PPC_SDAR                64
2365  PPC     KVM_REG_PPC_SIER                64
2366  PPC     KVM_REG_PPC_SIER2               64
2367  PPC     KVM_REG_PPC_SIER3               64
2368  PPC     KVM_REG_PPC_PMC1                32
2369  PPC     KVM_REG_PPC_PMC2                32
2370  PPC     KVM_REG_PPC_PMC3                32
2371  PPC     KVM_REG_PPC_PMC4                32
2372  PPC     KVM_REG_PPC_PMC5                32
2373  PPC     KVM_REG_PPC_PMC6                32
2374  PPC     KVM_REG_PPC_PMC7                32
2375  PPC     KVM_REG_PPC_PMC8                32
2376  PPC     KVM_REG_PPC_FPR0                64
2377  ...
2378  PPC     KVM_REG_PPC_FPR31               64
2379  PPC     KVM_REG_PPC_VR0                 128
2380  ...
2381  PPC     KVM_REG_PPC_VR31                128
2382  PPC     KVM_REG_PPC_VSR0                128
2383  ...
2384  PPC     KVM_REG_PPC_VSR31               128
2385  PPC     KVM_REG_PPC_FPSCR               64
2386  PPC     KVM_REG_PPC_VSCR                32
2387  PPC     KVM_REG_PPC_VPA_ADDR            64
2388  PPC     KVM_REG_PPC_VPA_SLB             128
2389  PPC     KVM_REG_PPC_VPA_DTL             128
2390  PPC     KVM_REG_PPC_EPCR                32
2391  PPC     KVM_REG_PPC_EPR                 32
2392  PPC     KVM_REG_PPC_TCR                 32
2393  PPC     KVM_REG_PPC_TSR                 32
2394  PPC     KVM_REG_PPC_OR_TSR              32
2395  PPC     KVM_REG_PPC_CLEAR_TSR           32
2396  PPC     KVM_REG_PPC_MAS0                32
2397  PPC     KVM_REG_PPC_MAS1                32
2398  PPC     KVM_REG_PPC_MAS2                64
2399  PPC     KVM_REG_PPC_MAS7_3              64
2400  PPC     KVM_REG_PPC_MAS4                32
2401  PPC     KVM_REG_PPC_MAS6                32
2402  PPC     KVM_REG_PPC_MMUCFG              32
2403  PPC     KVM_REG_PPC_TLB0CFG             32
2404  PPC     KVM_REG_PPC_TLB1CFG             32
2405  PPC     KVM_REG_PPC_TLB2CFG             32
2406  PPC     KVM_REG_PPC_TLB3CFG             32
2407  PPC     KVM_REG_PPC_TLB0PS              32
2408  PPC     KVM_REG_PPC_TLB1PS              32
2409  PPC     KVM_REG_PPC_TLB2PS              32
2410  PPC     KVM_REG_PPC_TLB3PS              32
2411  PPC     KVM_REG_PPC_EPTCFG              32
2412  PPC     KVM_REG_PPC_ICP_STATE           64
2413  PPC     KVM_REG_PPC_VP_STATE            128
2414  PPC     KVM_REG_PPC_TB_OFFSET           64
2415  PPC     KVM_REG_PPC_SPMC1               32
2416  PPC     KVM_REG_PPC_SPMC2               32
2417  PPC     KVM_REG_PPC_IAMR                64
2418  PPC     KVM_REG_PPC_TFHAR               64
2419  PPC     KVM_REG_PPC_TFIAR               64
2420  PPC     KVM_REG_PPC_TEXASR              64
2421  PPC     KVM_REG_PPC_FSCR                64
2422  PPC     KVM_REG_PPC_PSPB                32
2423  PPC     KVM_REG_PPC_EBBHR               64
2424  PPC     KVM_REG_PPC_EBBRR               64
2425  PPC     KVM_REG_PPC_BESCR               64
2426  PPC     KVM_REG_PPC_TAR                 64
2427  PPC     KVM_REG_PPC_DPDES               64
2428  PPC     KVM_REG_PPC_DAWR                64
2429  PPC     KVM_REG_PPC_DAWRX               64
2430  PPC     KVM_REG_PPC_CIABR               64
2431  PPC     KVM_REG_PPC_IC                  64
2432  PPC     KVM_REG_PPC_VTB                 64
2433  PPC     KVM_REG_PPC_CSIGR               64
2434  PPC     KVM_REG_PPC_TACR                64
2435  PPC     KVM_REG_PPC_TCSCR               64
2436  PPC     KVM_REG_PPC_PID                 64
2437  PPC     KVM_REG_PPC_ACOP                64
2438  PPC     KVM_REG_PPC_VRSAVE              32
2439  PPC     KVM_REG_PPC_LPCR                32
2440  PPC     KVM_REG_PPC_LPCR_64             64
2441  PPC     KVM_REG_PPC_PPR                 64
2442  PPC     KVM_REG_PPC_ARCH_COMPAT         32
2443  PPC     KVM_REG_PPC_DABRX               32
2444  PPC     KVM_REG_PPC_WORT                64
2445  PPC	  KVM_REG_PPC_SPRG9               64
2446  PPC	  KVM_REG_PPC_DBSR                32
2447  PPC     KVM_REG_PPC_TIDR                64
2448  PPC     KVM_REG_PPC_PSSCR               64
2449  PPC     KVM_REG_PPC_DEC_EXPIRY          64
2450  PPC     KVM_REG_PPC_PTCR                64
2451  PPC     KVM_REG_PPC_HASHKEYR            64
2452  PPC     KVM_REG_PPC_HASHPKEYR           64
2453  PPC     KVM_REG_PPC_DAWR1               64
2454  PPC     KVM_REG_PPC_DAWRX1              64
2455  PPC     KVM_REG_PPC_DEXCR               64
2456  PPC     KVM_REG_PPC_TM_GPR0             64
2457  ...
2458  PPC     KVM_REG_PPC_TM_GPR31            64
2459  PPC     KVM_REG_PPC_TM_VSR0             128
2460  ...
2461  PPC     KVM_REG_PPC_TM_VSR63            128
2462  PPC     KVM_REG_PPC_TM_CR               64
2463  PPC     KVM_REG_PPC_TM_LR               64
2464  PPC     KVM_REG_PPC_TM_CTR              64
2465  PPC     KVM_REG_PPC_TM_FPSCR            64
2466  PPC     KVM_REG_PPC_TM_AMR              64
2467  PPC     KVM_REG_PPC_TM_PPR              64
2468  PPC     KVM_REG_PPC_TM_VRSAVE           64
2469  PPC     KVM_REG_PPC_TM_VSCR             32
2470  PPC     KVM_REG_PPC_TM_DSCR             64
2471  PPC     KVM_REG_PPC_TM_TAR              64
2472  PPC     KVM_REG_PPC_TM_XER              64
2473
2474  MIPS    KVM_REG_MIPS_R0                 64
2475  ...
2476  MIPS    KVM_REG_MIPS_R31                64
2477  MIPS    KVM_REG_MIPS_HI                 64
2478  MIPS    KVM_REG_MIPS_LO                 64
2479  MIPS    KVM_REG_MIPS_PC                 64
2480  MIPS    KVM_REG_MIPS_CP0_INDEX          32
2481  MIPS    KVM_REG_MIPS_CP0_ENTRYLO0       64
2482  MIPS    KVM_REG_MIPS_CP0_ENTRYLO1       64
2483  MIPS    KVM_REG_MIPS_CP0_CONTEXT        64
2484  MIPS    KVM_REG_MIPS_CP0_CONTEXTCONFIG  32
2485  MIPS    KVM_REG_MIPS_CP0_USERLOCAL      64
2486  MIPS    KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2487  MIPS    KVM_REG_MIPS_CP0_PAGEMASK       32
2488  MIPS    KVM_REG_MIPS_CP0_PAGEGRAIN      32
2489  MIPS    KVM_REG_MIPS_CP0_SEGCTL0        64
2490  MIPS    KVM_REG_MIPS_CP0_SEGCTL1        64
2491  MIPS    KVM_REG_MIPS_CP0_SEGCTL2        64
2492  MIPS    KVM_REG_MIPS_CP0_PWBASE         64
2493  MIPS    KVM_REG_MIPS_CP0_PWFIELD        64
2494  MIPS    KVM_REG_MIPS_CP0_PWSIZE         64
2495  MIPS    KVM_REG_MIPS_CP0_WIRED          32
2496  MIPS    KVM_REG_MIPS_CP0_PWCTL          32
2497  MIPS    KVM_REG_MIPS_CP0_HWRENA         32
2498  MIPS    KVM_REG_MIPS_CP0_BADVADDR       64
2499  MIPS    KVM_REG_MIPS_CP0_BADINSTR       32
2500  MIPS    KVM_REG_MIPS_CP0_BADINSTRP      32
2501  MIPS    KVM_REG_MIPS_CP0_COUNT          32
2502  MIPS    KVM_REG_MIPS_CP0_ENTRYHI        64
2503  MIPS    KVM_REG_MIPS_CP0_COMPARE        32
2504  MIPS    KVM_REG_MIPS_CP0_STATUS         32
2505  MIPS    KVM_REG_MIPS_CP0_INTCTL         32
2506  MIPS    KVM_REG_MIPS_CP0_CAUSE          32
2507  MIPS    KVM_REG_MIPS_CP0_EPC            64
2508  MIPS    KVM_REG_MIPS_CP0_PRID           32
2509  MIPS    KVM_REG_MIPS_CP0_EBASE          64
2510  MIPS    KVM_REG_MIPS_CP0_CONFIG         32
2511  MIPS    KVM_REG_MIPS_CP0_CONFIG1        32
2512  MIPS    KVM_REG_MIPS_CP0_CONFIG2        32
2513  MIPS    KVM_REG_MIPS_CP0_CONFIG3        32
2514  MIPS    KVM_REG_MIPS_CP0_CONFIG4        32
2515  MIPS    KVM_REG_MIPS_CP0_CONFIG5        32
2516  MIPS    KVM_REG_MIPS_CP0_CONFIG7        32
2517  MIPS    KVM_REG_MIPS_CP0_XCONTEXT       64
2518  MIPS    KVM_REG_MIPS_CP0_ERROREPC       64
2519  MIPS    KVM_REG_MIPS_CP0_KSCRATCH1      64
2520  MIPS    KVM_REG_MIPS_CP0_KSCRATCH2      64
2521  MIPS    KVM_REG_MIPS_CP0_KSCRATCH3      64
2522  MIPS    KVM_REG_MIPS_CP0_KSCRATCH4      64
2523  MIPS    KVM_REG_MIPS_CP0_KSCRATCH5      64
2524  MIPS    KVM_REG_MIPS_CP0_KSCRATCH6      64
2525  MIPS    KVM_REG_MIPS_CP0_MAAR(0..63)    64
2526  MIPS    KVM_REG_MIPS_COUNT_CTL          64
2527  MIPS    KVM_REG_MIPS_COUNT_RESUME       64
2528  MIPS    KVM_REG_MIPS_COUNT_HZ           64
2529  MIPS    KVM_REG_MIPS_FPR_32(0..31)      32
2530  MIPS    KVM_REG_MIPS_FPR_64(0..31)      64
2531  MIPS    KVM_REG_MIPS_VEC_128(0..31)     128
2532  MIPS    KVM_REG_MIPS_FCR_IR             32
2533  MIPS    KVM_REG_MIPS_FCR_CSR            32
2534  MIPS    KVM_REG_MIPS_MSA_IR             32
2535  MIPS    KVM_REG_MIPS_MSA_CSR            32
2536  ======= =============================== ============
2537
2538ARM registers are mapped using the lower 32 bits.  The upper 16 of that
2539is the register group type, or coprocessor number:
2540
2541ARM core registers have the following id bit patterns::
2542
2543  0x4020 0000 0010 <index into the kvm_regs struct:16>
2544
2545ARM 32-bit CP15 registers have the following id bit patterns::
2546
2547  0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2548
2549ARM 64-bit CP15 registers have the following id bit patterns::
2550
2551  0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2552
2553ARM CCSIDR registers are demultiplexed by CSSELR value::
2554
2555  0x4020 0000 0011 00 <csselr:8>
2556
2557ARM 32-bit VFP control registers have the following id bit patterns::
2558
2559  0x4020 0000 0012 1 <regno:12>
2560
2561ARM 64-bit FP registers have the following id bit patterns::
2562
2563  0x4030 0000 0012 0 <regno:12>
2564
2565ARM firmware pseudo-registers have the following bit pattern::
2566
2567  0x4030 0000 0014 <regno:16>
2568
2569
2570arm64 registers are mapped using the lower 32 bits. The upper 16 of
2571that is the register group type, or coprocessor number:
2572
2573arm64 core/FP-SIMD registers have the following id bit patterns. Note
2574that the size of the access is variable, as the kvm_regs structure
2575contains elements ranging from 32 to 128 bits. The index is a 32bit
2576value in the kvm_regs structure seen as a 32bit array::
2577
2578  0x60x0 0000 0010 <index into the kvm_regs struct:16>
2579
2580Specifically:
2581
2582======================= ========= ===== =======================================
2583    Encoding            Register  Bits  kvm_regs member
2584======================= ========= ===== =======================================
2585  0x6030 0000 0010 0000 X0          64  regs.regs[0]
2586  0x6030 0000 0010 0002 X1          64  regs.regs[1]
2587  ...
2588  0x6030 0000 0010 003c X30         64  regs.regs[30]
2589  0x6030 0000 0010 003e SP          64  regs.sp
2590  0x6030 0000 0010 0040 PC          64  regs.pc
2591  0x6030 0000 0010 0042 PSTATE      64  regs.pstate
2592  0x6030 0000 0010 0044 SP_EL1      64  sp_el1
2593  0x6030 0000 0010 0046 ELR_EL1     64  elr_el1
2594  0x6030 0000 0010 0048 SPSR_EL1    64  spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2595  0x6030 0000 0010 004a SPSR_ABT    64  spsr[KVM_SPSR_ABT]
2596  0x6030 0000 0010 004c SPSR_UND    64  spsr[KVM_SPSR_UND]
2597  0x6030 0000 0010 004e SPSR_IRQ    64  spsr[KVM_SPSR_IRQ]
2598  0x6030 0000 0010 0050 SPSR_FIQ    64  spsr[KVM_SPSR_FIQ]
2599  0x6040 0000 0010 0054 V0         128  fp_regs.vregs[0]    [1]_
2600  0x6040 0000 0010 0058 V1         128  fp_regs.vregs[1]    [1]_
2601  ...
2602  0x6040 0000 0010 00d0 V31        128  fp_regs.vregs[31]   [1]_
2603  0x6020 0000 0010 00d4 FPSR        32  fp_regs.fpsr
2604  0x6020 0000 0010 00d5 FPCR        32  fp_regs.fpcr
2605======================= ========= ===== =======================================
2606
2607.. [1] These encodings are not accepted for SVE-enabled vcpus.  See
2608       :ref:`KVM_ARM_VCPU_INIT`.
2609
2610       The equivalent register content can be accessed via bits [127:0] of
2611       the corresponding SVE Zn registers instead for vcpus that have SVE
2612       enabled (see below).
2613
2614arm64 CCSIDR registers are demultiplexed by CSSELR value::
2615
2616  0x6020 0000 0011 00 <csselr:8>
2617
2618arm64 system registers have the following id bit patterns::
2619
2620  0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2621
2622.. warning::
2623
2624     Two system register IDs do not follow the specified pattern.  These
2625     are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2626     system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively.  These
2627     two had their values accidentally swapped, which means TIMER_CVAL is
2628     derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2629     derived from the register encoding for CNTV_CVAL_EL0.  As this is
2630     API, it must remain this way.
2631
2632arm64 firmware pseudo-registers have the following bit pattern::
2633
2634  0x6030 0000 0014 <regno:16>
2635
2636arm64 SVE registers have the following bit patterns::
2637
2638  0x6080 0000 0015 00 <n:5> <slice:5>   Zn bits[2048*slice + 2047 : 2048*slice]
2639  0x6050 0000 0015 04 <n:4> <slice:5>   Pn bits[256*slice + 255 : 256*slice]
2640  0x6050 0000 0015 060 <slice:5>        FFR bits[256*slice + 255 : 256*slice]
2641  0x6060 0000 0015 ffff                 KVM_REG_ARM64_SVE_VLS pseudo-register
2642
2643Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2644ENOENT.  max_vq is the vcpu's maximum supported vector length in 128-bit
2645quadwords: see [2]_ below.
2646
2647These registers are only accessible on vcpus for which SVE is enabled.
2648See KVM_ARM_VCPU_INIT for details.
2649
2650In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2651accessible until the vcpu's SVE configuration has been finalized
2652using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).  See KVM_ARM_VCPU_INIT
2653and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2654
2655KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2656lengths supported by the vcpu to be discovered and configured by
2657userspace.  When transferred to or from user memory via KVM_GET_ONE_REG
2658or KVM_SET_ONE_REG, the value of this register is of type
2659__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2660follows::
2661
2662  __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2663
2664  if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2665      ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2666		((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2667	/* Vector length vq * 16 bytes supported */
2668  else
2669	/* Vector length vq * 16 bytes not supported */
2670
2671.. [2] The maximum value vq for which the above condition is true is
2672       max_vq.  This is the maximum vector length available to the guest on
2673       this vcpu, and determines which register slices are visible through
2674       this ioctl interface.
2675
2676(See Documentation/arch/arm64/sve.rst for an explanation of the "vq"
2677nomenclature.)
2678
2679KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2680KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2681the host supports.
2682
2683Userspace may subsequently modify it if desired until the vcpu's SVE
2684configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2685
2686Apart from simply removing all vector lengths from the host set that
2687exceed some value, support for arbitrarily chosen sets of vector lengths
2688is hardware-dependent and may not be available.  Attempting to configure
2689an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2690EINVAL.
2691
2692After the vcpu's SVE configuration is finalized, further attempts to
2693write this register will fail with EPERM.
2694
2695arm64 bitmap feature firmware pseudo-registers have the following bit pattern::
2696
2697  0x6030 0000 0016 <regno:16>
2698
2699The bitmap feature firmware registers exposes the hypercall services that
2700are available for userspace to configure. The set bits corresponds to the
2701services that are available for the guests to access. By default, KVM
2702sets all the supported bits during VM initialization. The userspace can
2703discover the available services via KVM_GET_ONE_REG, and write back the
2704bitmap corresponding to the features that it wishes guests to see via
2705KVM_SET_ONE_REG.
2706
2707Note: These registers are immutable once any of the vCPUs of the VM has
2708run at least once. A KVM_SET_ONE_REG in such a scenario will return
2709a -EBUSY to userspace.
2710
2711(See Documentation/virt/kvm/arm/hypercalls.rst for more details.)
2712
2713
2714MIPS registers are mapped using the lower 32 bits.  The upper 16 of that is
2715the register group type:
2716
2717MIPS core registers (see above) have the following id bit patterns::
2718
2719  0x7030 0000 0000 <reg:16>
2720
2721MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2722patterns depending on whether they're 32-bit or 64-bit registers::
2723
2724  0x7020 0000 0001 00 <reg:5> <sel:3>   (32-bit)
2725  0x7030 0000 0001 00 <reg:5> <sel:3>   (64-bit)
2726
2727Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2728versions of the EntryLo registers regardless of the word size of the host
2729hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2730with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2731the PFNX field starting at bit 30.
2732
2733MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2734patterns::
2735
2736  0x7030 0000 0001 01 <reg:8>
2737
2738MIPS KVM control registers (see above) have the following id bit patterns::
2739
2740  0x7030 0000 0002 <reg:16>
2741
2742MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2743id bit patterns depending on the size of the register being accessed. They are
2744always accessed according to the current guest FPU mode (Status.FR and
2745Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2746if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2747registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2748overlap the FPU registers::
2749
2750  0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2751  0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2752  0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2753
2754MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2755following id bit patterns::
2756
2757  0x7020 0000 0003 01 <0:3> <reg:5>
2758
2759MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2760following id bit patterns::
2761
2762  0x7020 0000 0003 02 <0:3> <reg:5>
2763
2764RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of
2765that is the register group type.
2766
2767RISC-V config registers are meant for configuring a Guest VCPU and it has
2768the following id bit patterns::
2769
2770  0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host)
2771  0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host)
2772
2773Following are the RISC-V config registers:
2774
2775======================= ========= =============================================
2776    Encoding            Register  Description
2777======================= ========= =============================================
2778  0x80x0 0000 0100 0000 isa       ISA feature bitmap of Guest VCPU
2779======================= ========= =============================================
2780
2781The isa config register can be read anytime but can only be written before
2782a Guest VCPU runs. It will have ISA feature bits matching underlying host
2783set by default.
2784
2785RISC-V core registers represent the general execution state of a Guest VCPU
2786and it has the following id bit patterns::
2787
2788  0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host)
2789  0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host)
2790
2791Following are the RISC-V core registers:
2792
2793======================= ========= =============================================
2794    Encoding            Register  Description
2795======================= ========= =============================================
2796  0x80x0 0000 0200 0000 regs.pc   Program counter
2797  0x80x0 0000 0200 0001 regs.ra   Return address
2798  0x80x0 0000 0200 0002 regs.sp   Stack pointer
2799  0x80x0 0000 0200 0003 regs.gp   Global pointer
2800  0x80x0 0000 0200 0004 regs.tp   Task pointer
2801  0x80x0 0000 0200 0005 regs.t0   Caller saved register 0
2802  0x80x0 0000 0200 0006 regs.t1   Caller saved register 1
2803  0x80x0 0000 0200 0007 regs.t2   Caller saved register 2
2804  0x80x0 0000 0200 0008 regs.s0   Callee saved register 0
2805  0x80x0 0000 0200 0009 regs.s1   Callee saved register 1
2806  0x80x0 0000 0200 000a regs.a0   Function argument (or return value) 0
2807  0x80x0 0000 0200 000b regs.a1   Function argument (or return value) 1
2808  0x80x0 0000 0200 000c regs.a2   Function argument 2
2809  0x80x0 0000 0200 000d regs.a3   Function argument 3
2810  0x80x0 0000 0200 000e regs.a4   Function argument 4
2811  0x80x0 0000 0200 000f regs.a5   Function argument 5
2812  0x80x0 0000 0200 0010 regs.a6   Function argument 6
2813  0x80x0 0000 0200 0011 regs.a7   Function argument 7
2814  0x80x0 0000 0200 0012 regs.s2   Callee saved register 2
2815  0x80x0 0000 0200 0013 regs.s3   Callee saved register 3
2816  0x80x0 0000 0200 0014 regs.s4   Callee saved register 4
2817  0x80x0 0000 0200 0015 regs.s5   Callee saved register 5
2818  0x80x0 0000 0200 0016 regs.s6   Callee saved register 6
2819  0x80x0 0000 0200 0017 regs.s7   Callee saved register 7
2820  0x80x0 0000 0200 0018 regs.s8   Callee saved register 8
2821  0x80x0 0000 0200 0019 regs.s9   Callee saved register 9
2822  0x80x0 0000 0200 001a regs.s10  Callee saved register 10
2823  0x80x0 0000 0200 001b regs.s11  Callee saved register 11
2824  0x80x0 0000 0200 001c regs.t3   Caller saved register 3
2825  0x80x0 0000 0200 001d regs.t4   Caller saved register 4
2826  0x80x0 0000 0200 001e regs.t5   Caller saved register 5
2827  0x80x0 0000 0200 001f regs.t6   Caller saved register 6
2828  0x80x0 0000 0200 0020 mode      Privilege mode (1 = S-mode or 0 = U-mode)
2829======================= ========= =============================================
2830
2831RISC-V csr registers represent the supervisor mode control/status registers
2832of a Guest VCPU and it has the following id bit patterns::
2833
2834  0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host)
2835  0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host)
2836
2837Following are the RISC-V csr registers:
2838
2839======================= ========= =============================================
2840    Encoding            Register  Description
2841======================= ========= =============================================
2842  0x80x0 0000 0300 0000 sstatus   Supervisor status
2843  0x80x0 0000 0300 0001 sie       Supervisor interrupt enable
2844  0x80x0 0000 0300 0002 stvec     Supervisor trap vector base
2845  0x80x0 0000 0300 0003 sscratch  Supervisor scratch register
2846  0x80x0 0000 0300 0004 sepc      Supervisor exception program counter
2847  0x80x0 0000 0300 0005 scause    Supervisor trap cause
2848  0x80x0 0000 0300 0006 stval     Supervisor bad address or instruction
2849  0x80x0 0000 0300 0007 sip       Supervisor interrupt pending
2850  0x80x0 0000 0300 0008 satp      Supervisor address translation and protection
2851======================= ========= =============================================
2852
2853RISC-V timer registers represent the timer state of a Guest VCPU and it has
2854the following id bit patterns::
2855
2856  0x8030 0000 04 <index into the kvm_riscv_timer struct:24>
2857
2858Following are the RISC-V timer registers:
2859
2860======================= ========= =============================================
2861    Encoding            Register  Description
2862======================= ========= =============================================
2863  0x8030 0000 0400 0000 frequency Time base frequency (read-only)
2864  0x8030 0000 0400 0001 time      Time value visible to Guest
2865  0x8030 0000 0400 0002 compare   Time compare programmed by Guest
2866  0x8030 0000 0400 0003 state     Time compare state (1 = ON or 0 = OFF)
2867======================= ========= =============================================
2868
2869RISC-V F-extension registers represent the single precision floating point
2870state of a Guest VCPU and it has the following id bit patterns::
2871
2872  0x8020 0000 05 <index into the __riscv_f_ext_state struct:24>
2873
2874Following are the RISC-V F-extension registers:
2875
2876======================= ========= =============================================
2877    Encoding            Register  Description
2878======================= ========= =============================================
2879  0x8020 0000 0500 0000 f[0]      Floating point register 0
2880  ...
2881  0x8020 0000 0500 001f f[31]     Floating point register 31
2882  0x8020 0000 0500 0020 fcsr      Floating point control and status register
2883======================= ========= =============================================
2884
2885RISC-V D-extension registers represent the double precision floating point
2886state of a Guest VCPU and it has the following id bit patterns::
2887
2888  0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr)
2889  0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr)
2890
2891Following are the RISC-V D-extension registers:
2892
2893======================= ========= =============================================
2894    Encoding            Register  Description
2895======================= ========= =============================================
2896  0x8030 0000 0600 0000 f[0]      Floating point register 0
2897  ...
2898  0x8030 0000 0600 001f f[31]     Floating point register 31
2899  0x8020 0000 0600 0020 fcsr      Floating point control and status register
2900======================= ========= =============================================
2901
2902LoongArch registers are mapped using the lower 32 bits. The upper 16 bits of
2903that is the register group type.
2904
2905LoongArch csr registers are used to control guest cpu or get status of guest
2906cpu, and they have the following id bit patterns::
2907
2908  0x9030 0000 0001 00 <reg:5> <sel:3>   (64-bit)
2909
2910LoongArch KVM control registers are used to implement some new defined functions
2911such as set vcpu counter or reset vcpu, and they have the following id bit patterns::
2912
2913  0x9030 0000 0002 <reg:16>
2914
2915x86 MSR registers have the following id bit patterns::
2916  0x2030 0002 <msr number:32>
2917
2918Following are the KVM-defined registers for x86:
2919
2920======================= ========= =============================================
2921    Encoding            Register  Description
2922======================= ========= =============================================
2923  0x2030 0003 0000 0000 SSP       Shadow Stack Pointer
2924======================= ========= =============================================
2925
29264.69 KVM_GET_ONE_REG
2927--------------------
2928
2929:Capability: KVM_CAP_ONE_REG
2930:Architectures: all
2931:Type: vcpu ioctl
2932:Parameters: struct kvm_one_reg (in and out)
2933:Returns: 0 on success, negative value on failure
2934
2935Errors include:
2936
2937  ======== ============================================================
2938  ENOENT   no such register
2939  EINVAL   invalid register ID, or no such register or used with VMs in
2940           protected virtualization mode on s390
2941  EPERM    (arm64) register access not allowed before vcpu finalization
2942  ======== ============================================================
2943
2944(These error codes are indicative only: do not rely on a specific error
2945code being returned in a specific situation.)
2946
2947This ioctl allows to receive the value of a single register implemented
2948in a vcpu. The register to read is indicated by the "id" field of the
2949kvm_one_reg struct passed in. On success, the register value can be found
2950at the memory location pointed to by "addr".
2951
2952The list of registers accessible using this interface is identical to the
2953list in 4.68.
2954
2955
29564.70 KVM_KVMCLOCK_CTRL
2957----------------------
2958
2959:Capability: KVM_CAP_KVMCLOCK_CTRL
2960:Architectures: Any that implement pvclocks (currently x86 only)
2961:Type: vcpu ioctl
2962:Parameters: None
2963:Returns: 0 on success, -1 on error
2964
2965This ioctl sets a flag accessible to the guest indicating that the specified
2966vCPU has been paused by the host userspace.
2967
2968The host will set a flag in the pvclock structure that is checked from the
2969soft lockup watchdog.  The flag is part of the pvclock structure that is
2970shared between guest and host, specifically the second bit of the flags
2971field of the pvclock_vcpu_time_info structure.  It will be set exclusively by
2972the host and read/cleared exclusively by the guest.  The guest operation of
2973checking and clearing the flag must be an atomic operation so
2974load-link/store-conditional, or equivalent must be used.  There are two cases
2975where the guest will clear the flag: when the soft lockup watchdog timer resets
2976itself or when a soft lockup is detected.  This ioctl can be called any time
2977after pausing the vcpu, but before it is resumed.
2978
2979
29804.71 KVM_SIGNAL_MSI
2981-------------------
2982
2983:Capability: KVM_CAP_SIGNAL_MSI
2984:Architectures: x86 arm64
2985:Type: vm ioctl
2986:Parameters: struct kvm_msi (in)
2987:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2988
2989Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2990MSI messages.
2991
2992::
2993
2994  struct kvm_msi {
2995	__u32 address_lo;
2996	__u32 address_hi;
2997	__u32 data;
2998	__u32 flags;
2999	__u32 devid;
3000	__u8  pad[12];
3001  };
3002
3003flags:
3004  KVM_MSI_VALID_DEVID: devid contains a valid value.  The per-VM
3005  KVM_CAP_MSI_DEVID capability advertises the requirement to provide
3006  the device ID.  If this capability is not available, userspace
3007  should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
3008
3009If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
3010for the device that wrote the MSI message.  For PCI, this is usually a
3011BDF identifier in the lower 16 bits.
3012
3013On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
3014feature of KVM_CAP_X2APIC_API capability is enabled.  If it is enabled,
3015address_hi bits 31-8 provide bits 31-8 of the destination id.  Bits 7-0 of
3016address_hi must be zero.
3017
3018
30194.71 KVM_CREATE_PIT2
3020--------------------
3021
3022:Capability: KVM_CAP_PIT2
3023:Architectures: x86
3024:Type: vm ioctl
3025:Parameters: struct kvm_pit_config (in)
3026:Returns: 0 on success, -1 on error
3027
3028Creates an in-kernel device model for the i8254 PIT. This call is only valid
3029after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
3030parameters have to be passed::
3031
3032  struct kvm_pit_config {
3033	__u32 flags;
3034	__u32 pad[15];
3035  };
3036
3037Valid flags are::
3038
3039  #define KVM_PIT_SPEAKER_DUMMY     1 /* emulate speaker port stub */
3040
3041PIT timer interrupts may use a per-VM kernel thread for injection. If it
3042exists, this thread will have a name of the following pattern::
3043
3044  kvm-pit/<owner-process-pid>
3045
3046When running a guest with elevated priorities, the scheduling parameters of
3047this thread may have to be adjusted accordingly.
3048
3049This IOCTL replaces the obsolete KVM_CREATE_PIT.
3050
3051
30524.72 KVM_GET_PIT2
3053-----------------
3054
3055:Capability: KVM_CAP_PIT_STATE2
3056:Architectures: x86
3057:Type: vm ioctl
3058:Parameters: struct kvm_pit_state2 (out)
3059:Returns: 0 on success, -1 on error
3060
3061Retrieves the state of the in-kernel PIT model. Only valid after
3062KVM_CREATE_PIT2. The state is returned in the following structure::
3063
3064  struct kvm_pit_state2 {
3065	struct kvm_pit_channel_state channels[3];
3066	__u32 flags;
3067	__u32 reserved[9];
3068  };
3069
3070Valid flags are::
3071
3072  /* disable PIT in HPET legacy mode */
3073  #define KVM_PIT_FLAGS_HPET_LEGACY     0x00000001
3074  /* speaker port data bit enabled */
3075  #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002
3076
3077This IOCTL replaces the obsolete KVM_GET_PIT.
3078
3079
30804.73 KVM_SET_PIT2
3081-----------------
3082
3083:Capability: KVM_CAP_PIT_STATE2
3084:Architectures: x86
3085:Type: vm ioctl
3086:Parameters: struct kvm_pit_state2 (in)
3087:Returns: 0 on success, -1 on error
3088
3089Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
3090See KVM_GET_PIT2 for details on struct kvm_pit_state2.
3091
3092.. Tip::
3093  ``KVM_SET_PIT2`` strictly adheres to the spec of Intel 8254 PIT.  For example,
3094  a ``count`` value of 0 in ``struct kvm_pit_channel_state`` is interpreted as
3095  65536, which is the maximum count value. Refer to `Intel 8254 programmable
3096  interval timer <https://www.scs.stanford.edu/10wi-cs140/pintos/specs/8254.pdf>`_.
3097
3098This IOCTL replaces the obsolete KVM_SET_PIT.
3099
3100
31014.74 KVM_PPC_GET_SMMU_INFO
3102--------------------------
3103
3104:Capability: KVM_CAP_PPC_GET_SMMU_INFO
3105:Architectures: powerpc
3106:Type: vm ioctl
3107:Parameters: None
3108:Returns: 0 on success, -1 on error
3109
3110This populates and returns a structure describing the features of
3111the "Server" class MMU emulation supported by KVM.
3112This can in turn be used by userspace to generate the appropriate
3113device-tree properties for the guest operating system.
3114
3115The structure contains some global information, followed by an
3116array of supported segment page sizes::
3117
3118      struct kvm_ppc_smmu_info {
3119	     __u64 flags;
3120	     __u32 slb_size;
3121	     __u32 pad;
3122	     struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
3123      };
3124
3125The supported flags are:
3126
3127    - KVM_PPC_PAGE_SIZES_REAL:
3128        When that flag is set, guest page sizes must "fit" the backing
3129        store page sizes. When not set, any page size in the list can
3130        be used regardless of how they are backed by userspace.
3131
3132    - KVM_PPC_1T_SEGMENTS
3133        The emulated MMU supports 1T segments in addition to the
3134        standard 256M ones.
3135
3136    - KVM_PPC_NO_HASH
3137	This flag indicates that HPT guests are not supported by KVM,
3138	thus all guests must use radix MMU mode.
3139
3140The "slb_size" field indicates how many SLB entries are supported
3141
3142The "sps" array contains 8 entries indicating the supported base
3143page sizes for a segment in increasing order. Each entry is defined
3144as follow::
3145
3146   struct kvm_ppc_one_seg_page_size {
3147	__u32 page_shift;	/* Base page shift of segment (or 0) */
3148	__u32 slb_enc;		/* SLB encoding for BookS */
3149	struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3150   };
3151
3152An entry with a "page_shift" of 0 is unused. Because the array is
3153organized in increasing order, a lookup can stop when encountering
3154such an entry.
3155
3156The "slb_enc" field provides the encoding to use in the SLB for the
3157page size. The bits are in positions such as the value can directly
3158be OR'ed into the "vsid" argument of the slbmte instruction.
3159
3160The "enc" array is a list which for each of those segment base page
3161size provides the list of supported actual page sizes (which can be
3162only larger or equal to the base page size), along with the
3163corresponding encoding in the hash PTE. Similarly, the array is
31648 entries sorted by increasing sizes and an entry with a "0" shift
3165is an empty entry and a terminator::
3166
3167   struct kvm_ppc_one_page_size {
3168	__u32 page_shift;	/* Page shift (or 0) */
3169	__u32 pte_enc;		/* Encoding in the HPTE (>>12) */
3170   };
3171
3172The "pte_enc" field provides a value that can OR'ed into the hash
3173PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3174into the hash PTE second double word).
3175
31764.75 KVM_IRQFD
3177--------------
3178
3179:Capability: KVM_CAP_IRQFD
3180:Architectures: x86 s390 arm64
3181:Type: vm ioctl
3182:Parameters: struct kvm_irqfd (in)
3183:Returns: 0 on success, -1 on error
3184
3185Allows setting an eventfd to directly trigger a guest interrupt.
3186kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3187kvm_irqfd.gsi specifies the irqchip pin toggled by this event.  When
3188an event is triggered on the eventfd, an interrupt is injected into
3189the guest using the specified gsi pin.  The irqfd is removed using
3190the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3191and kvm_irqfd.gsi.
3192
3193With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3194mechanism allowing emulation of level-triggered, irqfd-based
3195interrupts.  When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3196additional eventfd in the kvm_irqfd.resamplefd field.  When operating
3197in resample mode, posting of an interrupt through kvm_irq.fd asserts
3198the specified gsi in the irqchip.  When the irqchip is resampled, such
3199as from an EOI, the gsi is de-asserted and the user is notified via
3200kvm_irqfd.resamplefd.  It is the user's responsibility to re-queue
3201the interrupt if the device making use of it still requires service.
3202Note that closing the resamplefd is not sufficient to disable the
3203irqfd.  The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3204and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3205
3206On arm64, gsi routing being supported, the following can happen:
3207
3208- in case no routing entry is associated to this gsi, injection fails
3209- in case the gsi is associated to an irqchip routing entry,
3210  irqchip.pin + 32 corresponds to the injected SPI ID.
3211- in case the gsi is associated to an MSI routing entry, the MSI
3212  message and device ID are translated into an LPI (support restricted
3213  to GICv3 ITS in-kernel emulation).
3214
32154.76 KVM_PPC_ALLOCATE_HTAB
3216--------------------------
3217
3218:Capability: KVM_CAP_PPC_ALLOC_HTAB
3219:Architectures: powerpc
3220:Type: vm ioctl
3221:Parameters: Pointer to u32 containing hash table order (in/out)
3222:Returns: 0 on success, -1 on error
3223
3224This requests the host kernel to allocate an MMU hash table for a
3225guest using the PAPR paravirtualization interface.  This only does
3226anything if the kernel is configured to use the Book 3S HV style of
3227virtualization.  Otherwise the capability doesn't exist and the ioctl
3228returns an ENOTTY error.  The rest of this description assumes Book 3S
3229HV.
3230
3231There must be no vcpus running when this ioctl is called; if there
3232are, it will do nothing and return an EBUSY error.
3233
3234The parameter is a pointer to a 32-bit unsigned integer variable
3235containing the order (log base 2) of the desired size of the hash
3236table, which must be between 18 and 46.  On successful return from the
3237ioctl, the value will not be changed by the kernel.
3238
3239If no hash table has been allocated when any vcpu is asked to run
3240(with the KVM_RUN ioctl), the host kernel will allocate a
3241default-sized hash table (16 MB).
3242
3243If this ioctl is called when a hash table has already been allocated,
3244with a different order from the existing hash table, the existing hash
3245table will be freed and a new one allocated.  If this is ioctl is
3246called when a hash table has already been allocated of the same order
3247as specified, the kernel will clear out the existing hash table (zero
3248all HPTEs).  In either case, if the guest is using the virtualized
3249real-mode area (VRMA) facility, the kernel will re-create the VMRA
3250HPTEs on the next KVM_RUN of any vcpu.
3251
32524.77 KVM_S390_INTERRUPT
3253-----------------------
3254
3255:Capability: basic
3256:Architectures: s390
3257:Type: vm ioctl, vcpu ioctl
3258:Parameters: struct kvm_s390_interrupt (in)
3259:Returns: 0 on success, -1 on error
3260
3261Allows to inject an interrupt to the guest. Interrupts can be floating
3262(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3263
3264Interrupt parameters are passed via kvm_s390_interrupt::
3265
3266  struct kvm_s390_interrupt {
3267	__u32 type;
3268	__u32 parm;
3269	__u64 parm64;
3270  };
3271
3272type can be one of the following:
3273
3274KVM_S390_SIGP_STOP (vcpu)
3275    - sigp stop; optional flags in parm
3276KVM_S390_PROGRAM_INT (vcpu)
3277    - program check; code in parm
3278KVM_S390_SIGP_SET_PREFIX (vcpu)
3279    - sigp set prefix; prefix address in parm
3280KVM_S390_RESTART (vcpu)
3281    - restart
3282KVM_S390_INT_CLOCK_COMP (vcpu)
3283    - clock comparator interrupt
3284KVM_S390_INT_CPU_TIMER (vcpu)
3285    - CPU timer interrupt
3286KVM_S390_INT_VIRTIO (vm)
3287    - virtio external interrupt; external interrupt
3288      parameters in parm and parm64
3289KVM_S390_INT_SERVICE (vm)
3290    - sclp external interrupt; sclp parameter in parm
3291KVM_S390_INT_EMERGENCY (vcpu)
3292    - sigp emergency; source cpu in parm
3293KVM_S390_INT_EXTERNAL_CALL (vcpu)
3294    - sigp external call; source cpu in parm
3295KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3296    - compound value to indicate an
3297      I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3298      I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3299      interruption subclass)
3300KVM_S390_MCHK (vm, vcpu)
3301    - machine check interrupt; cr 14 bits in parm, machine check interrupt
3302      code in parm64 (note that machine checks needing further payload are not
3303      supported by this ioctl)
3304
3305This is an asynchronous vcpu ioctl and can be invoked from any thread.
3306
33074.78 KVM_PPC_GET_HTAB_FD
3308------------------------
3309
3310:Capability: KVM_CAP_PPC_HTAB_FD
3311:Architectures: powerpc
3312:Type: vm ioctl
3313:Parameters: Pointer to struct kvm_get_htab_fd (in)
3314:Returns: file descriptor number (>= 0) on success, -1 on error
3315
3316This returns a file descriptor that can be used either to read out the
3317entries in the guest's hashed page table (HPT), or to write entries to
3318initialize the HPT.  The returned fd can only be written to if the
3319KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3320can only be read if that bit is clear.  The argument struct looks like
3321this::
3322
3323  /* For KVM_PPC_GET_HTAB_FD */
3324  struct kvm_get_htab_fd {
3325	__u64	flags;
3326	__u64	start_index;
3327	__u64	reserved[2];
3328  };
3329
3330  /* Values for kvm_get_htab_fd.flags */
3331  #define KVM_GET_HTAB_BOLTED_ONLY	((__u64)0x1)
3332  #define KVM_GET_HTAB_WRITE		((__u64)0x2)
3333
3334The 'start_index' field gives the index in the HPT of the entry at
3335which to start reading.  It is ignored when writing.
3336
3337Reads on the fd will initially supply information about all
3338"interesting" HPT entries.  Interesting entries are those with the
3339bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3340all entries.  When the end of the HPT is reached, the read() will
3341return.  If read() is called again on the fd, it will start again from
3342the beginning of the HPT, but will only return HPT entries that have
3343changed since they were last read.
3344
3345Data read or written is structured as a header (8 bytes) followed by a
3346series of valid HPT entries (16 bytes) each.  The header indicates how
3347many valid HPT entries there are and how many invalid entries follow
3348the valid entries.  The invalid entries are not represented explicitly
3349in the stream.  The header format is::
3350
3351  struct kvm_get_htab_header {
3352	__u32	index;
3353	__u16	n_valid;
3354	__u16	n_invalid;
3355  };
3356
3357Writes to the fd create HPT entries starting at the index given in the
3358header; first 'n_valid' valid entries with contents from the data
3359written, then 'n_invalid' invalid entries, invalidating any previously
3360valid entries found.
3361
33624.79 KVM_CREATE_DEVICE
3363----------------------
3364
3365:Capability: KVM_CAP_DEVICE_CTRL
3366:Architectures: all
3367:Type: vm ioctl
3368:Parameters: struct kvm_create_device (in/out)
3369:Returns: 0 on success, -1 on error
3370
3371Errors:
3372
3373  ======  =======================================================
3374  ENODEV  The device type is unknown or unsupported
3375  EEXIST  Device already created, and this type of device may not
3376          be instantiated multiple times
3377  ======  =======================================================
3378
3379  Other error conditions may be defined by individual device types or
3380  have their standard meanings.
3381
3382Creates an emulated device in the kernel.  The file descriptor returned
3383in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3384
3385If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3386device type is supported (not necessarily whether it can be created
3387in the current vm).
3388
3389Individual devices should not define flags.  Attributes should be used
3390for specifying any behavior that is not implied by the device type
3391number.
3392
3393::
3394
3395  struct kvm_create_device {
3396	__u32	type;	/* in: KVM_DEV_TYPE_xxx */
3397	__u32	fd;	/* out: device handle */
3398	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
3399  };
3400
34014.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3402--------------------------------------------
3403
3404:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3405             KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3406             KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set)
3407:Architectures: x86, arm64, s390
3408:Type: device ioctl, vm ioctl, vcpu ioctl
3409:Parameters: struct kvm_device_attr
3410:Returns: 0 on success, -1 on error
3411
3412Errors:
3413
3414  =====   =============================================================
3415  ENXIO   The group or attribute is unknown/unsupported for this device
3416          or hardware support is missing.
3417  EPERM   The attribute cannot (currently) be accessed this way
3418          (e.g. read-only attribute, or attribute that only makes
3419          sense when the device is in a different state)
3420  =====   =============================================================
3421
3422  Other error conditions may be defined by individual device types.
3423
3424Gets/sets a specified piece of device configuration and/or state.  The
3425semantics are device-specific.  See individual device documentation in
3426the "devices" directory.  As with ONE_REG, the size of the data
3427transferred is defined by the particular attribute.
3428
3429::
3430
3431  struct kvm_device_attr {
3432	__u32	flags;		/* no flags currently defined */
3433	__u32	group;		/* device-defined */
3434	__u64	attr;		/* group-defined */
3435	__u64	addr;		/* userspace address of attr data */
3436  };
3437
34384.81 KVM_HAS_DEVICE_ATTR
3439------------------------
3440
3441:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3442             KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3443             KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device
3444:Type: device ioctl, vm ioctl, vcpu ioctl
3445:Parameters: struct kvm_device_attr
3446:Returns: 0 on success, -1 on error
3447
3448Errors:
3449
3450  =====   =============================================================
3451  ENXIO   The group or attribute is unknown/unsupported for this device
3452          or hardware support is missing.
3453  =====   =============================================================
3454
3455Tests whether a device supports a particular attribute.  A successful
3456return indicates the attribute is implemented.  It does not necessarily
3457indicate that the attribute can be read or written in the device's
3458current state.  "addr" is ignored.
3459
3460.. _KVM_ARM_VCPU_INIT:
3461
34624.82 KVM_ARM_VCPU_INIT
3463----------------------
3464
3465:Capability: basic
3466:Architectures: arm64
3467:Type: vcpu ioctl
3468:Parameters: struct kvm_vcpu_init (in)
3469:Returns: 0 on success; -1 on error
3470
3471Errors:
3472
3473  ======     =================================================================
3474  EINVAL     the target is unknown, or the combination of features is invalid.
3475  ENOENT     a features bit specified is unknown.
3476  ======     =================================================================
3477
3478This tells KVM what type of CPU to present to the guest, and what
3479optional features it should have.  This will cause a reset of the cpu
3480registers to their initial values.  If this is not called, KVM_RUN will
3481return ENOEXEC for that vcpu.
3482
3483The initial values are defined as:
3484	- Processor state:
3485		* AArch64: EL1h, D, A, I and F bits set. All other bits
3486		  are cleared.
3487		* AArch32: SVC, A, I and F bits set. All other bits are
3488		  cleared.
3489	- General Purpose registers, including PC and SP: set to 0
3490	- FPSIMD/NEON registers: set to 0
3491	- SVE registers: set to 0
3492	- System registers: Reset to their architecturally defined
3493	  values as for a warm reset to EL1 (resp. SVC) or EL2 (in the
3494	  case of EL2 being enabled).
3495
3496Note that because some registers reflect machine topology, all vcpus
3497should be created before this ioctl is invoked.
3498
3499Userspace can call this function multiple times for a given vcpu, including
3500after the vcpu has been run. This will reset the vcpu to its initial
3501state. All calls to this function after the initial call must use the same
3502target and same set of feature flags, otherwise EINVAL will be returned.
3503
3504Possible features:
3505
3506	- KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3507	  Depends on KVM_CAP_ARM_PSCI.  If not set, the CPU will be powered on
3508	  and execute guest code when KVM_RUN is called.
3509	- KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3510	  Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3511	- KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3512          backward compatible with v0.2) for the CPU.
3513	  Depends on KVM_CAP_ARM_PSCI_0_2.
3514	- KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3515	  Depends on KVM_CAP_ARM_PMU_V3.
3516
3517	- KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3518	  for arm64 only.
3519	  Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3520	  If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3521	  both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3522	  KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3523	  requested.
3524
3525	- KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3526	  for arm64 only.
3527	  Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3528	  If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3529	  both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3530	  KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3531	  requested.
3532
3533	- KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3534	  Depends on KVM_CAP_ARM_SVE.
3535	  Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3536
3537	   * After KVM_ARM_VCPU_INIT:
3538
3539	      - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3540	        initial value of this pseudo-register indicates the best set of
3541	        vector lengths possible for a vcpu on this host.
3542
3543	   * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3544
3545	      - KVM_RUN and KVM_GET_REG_LIST are not available;
3546
3547	      - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3548	        the scalable architectural SVE registers
3549	        KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3550	        KVM_REG_ARM64_SVE_FFR;
3551
3552	      - KVM_REG_ARM64_SVE_VLS may optionally be written using
3553	        KVM_SET_ONE_REG, to modify the set of vector lengths available
3554	        for the vcpu.
3555
3556	   * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3557
3558	      - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3559	        no longer be written using KVM_SET_ONE_REG.
3560
3561	- KVM_ARM_VCPU_HAS_EL2: Enable Nested Virtualisation support,
3562	  booting the guest from EL2 instead of EL1.
3563	  Depends on KVM_CAP_ARM_EL2.
3564	  The VM is running with HCR_EL2.E2H being RES1 (VHE) unless
3565	  KVM_ARM_VCPU_HAS_EL2_E2H0 is also set.
3566
3567	- KVM_ARM_VCPU_HAS_EL2_E2H0: Restrict Nested Virtualisation
3568	  support to HCR_EL2.E2H being RES0 (non-VHE).
3569	  Depends on KVM_CAP_ARM_EL2_E2H0.
3570	  KVM_ARM_VCPU_HAS_EL2 must also be set.
3571
35724.83 KVM_ARM_PREFERRED_TARGET
3573-----------------------------
3574
3575:Capability: basic
3576:Architectures: arm64
3577:Type: vm ioctl
3578:Parameters: struct kvm_vcpu_init (out)
3579:Returns: 0 on success; -1 on error
3580
3581Errors:
3582
3583  ======     ==========================================
3584  ENODEV     no preferred target available for the host
3585  ======     ==========================================
3586
3587This queries KVM for preferred CPU target type which can be emulated
3588by KVM on underlying host.
3589
3590The ioctl returns struct kvm_vcpu_init instance containing information
3591about preferred CPU target type and recommended features for it.  The
3592kvm_vcpu_init->features bitmap returned will have feature bits set if
3593the preferred target recommends setting these features, but this is
3594not mandatory.
3595
3596The information returned by this ioctl can be used to prepare an instance
3597of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3598VCPU matching underlying host.
3599
3600
36014.84 KVM_GET_REG_LIST
3602---------------------
3603
3604:Capability: basic
3605:Architectures: arm64, mips, riscv, x86 (if KVM_CAP_ONE_REG)
3606:Type: vcpu ioctl
3607:Parameters: struct kvm_reg_list (in/out)
3608:Returns: 0 on success; -1 on error
3609
3610Errors:
3611
3612  =====      ==============================================================
3613  E2BIG      the reg index list is too big to fit in the array specified by
3614             the user (the number required will be written into n).
3615  =====      ==============================================================
3616
3617::
3618
3619  struct kvm_reg_list {
3620	__u64 n; /* number of registers in reg[] */
3621	__u64 reg[0];
3622  };
3623
3624This ioctl returns the guest registers that are supported for the
3625KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3626
3627Note that s390 does not support KVM_GET_REG_LIST for historical reasons
3628(read: nobody cared).  The set of registers in kernels 4.x and newer is:
3629
3630- KVM_REG_S390_TODPR
3631
3632- KVM_REG_S390_EPOCHDIFF
3633
3634- KVM_REG_S390_CPU_TIMER
3635
3636- KVM_REG_S390_CLOCK_COMP
3637
3638- KVM_REG_S390_PFTOKEN
3639
3640- KVM_REG_S390_PFCOMPARE
3641
3642- KVM_REG_S390_PFSELECT
3643
3644- KVM_REG_S390_PP
3645
3646- KVM_REG_S390_GBEA
3647
3648Note, for x86, all MSRs enumerated by KVM_GET_MSR_INDEX_LIST are supported as
3649type KVM_X86_REG_TYPE_MSR, but are NOT enumerated via KVM_GET_REG_LIST.
3650
36514.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3652-----------------------------------------
3653
3654:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3655:Architectures: arm64
3656:Type: vm ioctl
3657:Parameters: struct kvm_arm_device_address (in)
3658:Returns: 0 on success, -1 on error
3659
3660Errors:
3661
3662  ======  ============================================
3663  ENODEV  The device id is unknown
3664  ENXIO   Device not supported on current system
3665  EEXIST  Address already set
3666  E2BIG   Address outside guest physical address space
3667  EBUSY   Address overlaps with other device range
3668  ======  ============================================
3669
3670::
3671
3672  struct kvm_arm_device_addr {
3673	__u64 id;
3674	__u64 addr;
3675  };
3676
3677Specify a device address in the guest's physical address space where guests
3678can access emulated or directly exposed devices, which the host kernel needs
3679to know about. The id field is an architecture specific identifier for a
3680specific device.
3681
3682arm64 divides the id field into two parts, a device id and an
3683address type id specific to the individual device::
3684
3685  bits:  | 63        ...       32 | 31    ...    16 | 15    ...    0 |
3686  field: |        0x00000000      |     device id   |  addr type id  |
3687
3688arm64 currently only require this when using the in-kernel GIC
3689support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3690as the device id.  When setting the base address for the guest's
3691mapping of the VGIC virtual CPU and distributor interface, the ioctl
3692must be called after calling KVM_CREATE_IRQCHIP, but before calling
3693KVM_RUN on any of the VCPUs.  Calling this ioctl twice for any of the
3694base addresses will return -EEXIST.
3695
3696Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3697should be used instead.
3698
3699
37004.86 KVM_PPC_RTAS_DEFINE_TOKEN
3701------------------------------
3702
3703:Capability: KVM_CAP_PPC_RTAS
3704:Architectures: ppc
3705:Type: vm ioctl
3706:Parameters: struct kvm_rtas_token_args
3707:Returns: 0 on success, -1 on error
3708
3709Defines a token value for a RTAS (Run Time Abstraction Services)
3710service in order to allow it to be handled in the kernel.  The
3711argument struct gives the name of the service, which must be the name
3712of a service that has a kernel-side implementation.  If the token
3713value is non-zero, it will be associated with that service, and
3714subsequent RTAS calls by the guest specifying that token will be
3715handled by the kernel.  If the token value is 0, then any token
3716associated with the service will be forgotten, and subsequent RTAS
3717calls by the guest for that service will be passed to userspace to be
3718handled.
3719
37204.87 KVM_SET_GUEST_DEBUG
3721------------------------
3722
3723:Capability: KVM_CAP_SET_GUEST_DEBUG
3724:Architectures: x86, s390, ppc, arm64
3725:Type: vcpu ioctl
3726:Parameters: struct kvm_guest_debug (in)
3727:Returns: 0 on success; -1 on error
3728
3729::
3730
3731  struct kvm_guest_debug {
3732       __u32 control;
3733       __u32 pad;
3734       struct kvm_guest_debug_arch arch;
3735  };
3736
3737Set up the processor specific debug registers and configure vcpu for
3738handling guest debug events. There are two parts to the structure, the
3739first a control bitfield indicates the type of debug events to handle
3740when running. Common control bits are:
3741
3742  - KVM_GUESTDBG_ENABLE:        guest debugging is enabled
3743  - KVM_GUESTDBG_SINGLESTEP:    the next run should single-step
3744
3745The top 16 bits of the control field are architecture specific control
3746flags which can include the following:
3747
3748  - KVM_GUESTDBG_USE_SW_BP:     using software breakpoints [x86, arm64]
3749  - KVM_GUESTDBG_USE_HW_BP:     using hardware breakpoints [x86, s390]
3750  - KVM_GUESTDBG_USE_HW:        using hardware debug events [arm64]
3751  - KVM_GUESTDBG_INJECT_DB:     inject DB type exception [x86]
3752  - KVM_GUESTDBG_INJECT_BP:     inject BP type exception [x86]
3753  - KVM_GUESTDBG_EXIT_PENDING:  trigger an immediate guest exit [s390]
3754  - KVM_GUESTDBG_BLOCKIRQ:      avoid injecting interrupts/NMI/SMI [x86]
3755
3756For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3757are enabled in memory so we need to ensure breakpoint exceptions are
3758correctly trapped and the KVM run loop exits at the breakpoint and not
3759running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3760we need to ensure the guest vCPUs architecture specific registers are
3761updated to the correct (supplied) values.
3762
3763The second part of the structure is architecture specific and
3764typically contains a set of debug registers.
3765
3766For arm64 the number of debug registers is implementation defined and
3767can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3768KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3769indicating the number of supported registers.
3770
3771For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3772the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3773
3774Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3775supported KVM_GUESTDBG_* bits in the control field.
3776
3777When debug events exit the main run loop with the reason
3778KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3779structure containing architecture specific debug information.
3780
37814.88 KVM_GET_EMULATED_CPUID
3782---------------------------
3783
3784:Capability: KVM_CAP_EXT_EMUL_CPUID
3785:Architectures: x86
3786:Type: system ioctl
3787:Parameters: struct kvm_cpuid2 (in/out)
3788:Returns: 0 on success, -1 on error
3789
3790::
3791
3792  struct kvm_cpuid2 {
3793	__u32 nent;
3794	__u32 flags;
3795	struct kvm_cpuid_entry2 entries[0];
3796  };
3797
3798The member 'flags' is used for passing flags from userspace.
3799
3800::
3801
3802  #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
3803  #define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1) /* deprecated */
3804  #define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2) /* deprecated */
3805
3806  struct kvm_cpuid_entry2 {
3807	__u32 function;
3808	__u32 index;
3809	__u32 flags;
3810	__u32 eax;
3811	__u32 ebx;
3812	__u32 ecx;
3813	__u32 edx;
3814	__u32 padding[3];
3815  };
3816
3817This ioctl returns x86 cpuid features which are emulated by
3818kvm.Userspace can use the information returned by this ioctl to query
3819which features are emulated by kvm instead of being present natively.
3820
3821Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3822structure with the 'nent' field indicating the number of entries in
3823the variable-size array 'entries'. If the number of entries is too low
3824to describe the cpu capabilities, an error (E2BIG) is returned. If the
3825number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3826is returned. If the number is just right, the 'nent' field is adjusted
3827to the number of valid entries in the 'entries' array, which is then
3828filled.
3829
3830The entries returned are the set CPUID bits of the respective features
3831which kvm emulates, as returned by the CPUID instruction, with unknown
3832or unsupported feature bits cleared.
3833
3834Features like x2apic, for example, may not be present in the host cpu
3835but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3836emulated efficiently and thus not included here.
3837
3838The fields in each entry are defined as follows:
3839
3840  function:
3841	 the eax value used to obtain the entry
3842  index:
3843	 the ecx value used to obtain the entry (for entries that are
3844         affected by ecx)
3845  flags:
3846    an OR of zero or more of the following:
3847
3848        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3849           if the index field is valid
3850
3851   eax, ebx, ecx, edx:
3852
3853         the values returned by the cpuid instruction for
3854         this function/index combination
3855
38564.89 KVM_S390_MEM_OP
3857--------------------
3858
3859:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION
3860:Architectures: s390
3861:Type: vm ioctl, vcpu ioctl
3862:Parameters: struct kvm_s390_mem_op (in)
3863:Returns: = 0 on success,
3864          < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3865          16 bit program exception code if the access causes such an exception
3866
3867Read or write data from/to the VM's memory.
3868The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is
3869supported.
3870
3871Parameters are specified via the following structure::
3872
3873  struct kvm_s390_mem_op {
3874	__u64 gaddr;		/* the guest address */
3875	__u64 flags;		/* flags */
3876	__u32 size;		/* amount of bytes */
3877	__u32 op;		/* type of operation */
3878	__u64 buf;		/* buffer in userspace */
3879	union {
3880		struct {
3881			__u8 ar;	/* the access register number */
3882			__u8 key;	/* access key, ignored if flag unset */
3883			__u8 pad1[6];	/* ignored */
3884			__u64 old_addr;	/* ignored if flag unset */
3885		};
3886		__u32 sida_offset; /* offset into the sida */
3887		__u8 reserved[32]; /* ignored */
3888	};
3889  };
3890
3891The start address of the memory region has to be specified in the "gaddr"
3892field, and the length of the region in the "size" field (which must not
3893be 0). The maximum value for "size" can be obtained by checking the
3894KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3895userspace application where the read data should be written to for
3896a read access, or where the data that should be written is stored for
3897a write access.  The "reserved" field is meant for future extensions.
3898Reserved and unused values are ignored. Future extension that add members must
3899introduce new flags.
3900
3901The type of operation is specified in the "op" field. Flags modifying
3902their behavior can be set in the "flags" field. Undefined flag bits must
3903be set to 0.
3904
3905Possible operations are:
3906  * ``KVM_S390_MEMOP_LOGICAL_READ``
3907  * ``KVM_S390_MEMOP_LOGICAL_WRITE``
3908  * ``KVM_S390_MEMOP_ABSOLUTE_READ``
3909  * ``KVM_S390_MEMOP_ABSOLUTE_WRITE``
3910  * ``KVM_S390_MEMOP_SIDA_READ``
3911  * ``KVM_S390_MEMOP_SIDA_WRITE``
3912  * ``KVM_S390_MEMOP_ABSOLUTE_CMPXCHG``
3913
3914Logical read/write:
3915^^^^^^^^^^^^^^^^^^^
3916
3917Access logical memory, i.e. translate the given guest address to an absolute
3918address given the state of the VCPU and use the absolute address as target of
3919the access. "ar" designates the access register number to be used; the valid
3920range is 0..15.
3921Logical accesses are permitted for the VCPU ioctl only.
3922Logical accesses are permitted for non-protected guests only.
3923
3924Supported flags:
3925  * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3926  * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION``
3927  * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3928
3929The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the
3930corresponding memory access would cause an access exception; however,
3931no actual access to the data in memory at the destination is performed.
3932In this case, "buf" is unused and can be NULL.
3933
3934In case an access exception occurred during the access (or would occur
3935in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive
3936error number indicating the type of exception. This exception is also
3937raised directly at the corresponding VCPU if the flag
3938KVM_S390_MEMOP_F_INJECT_EXCEPTION is set.
3939On protection exceptions, unless specified otherwise, the injected
3940translation-exception identifier (TEID) indicates suppression.
3941
3942If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key
3943protection is also in effect and may cause exceptions if accesses are
3944prohibited given the access key designated by "key"; the valid range is 0..15.
3945KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION
3946is > 0.
3947Since the accessed memory may span multiple pages and those pages might have
3948different storage keys, it is possible that a protection exception occurs
3949after memory has been modified. In this case, if the exception is injected,
3950the TEID does not indicate suppression.
3951
3952Absolute read/write:
3953^^^^^^^^^^^^^^^^^^^^
3954
3955Access absolute memory. This operation is intended to be used with the
3956KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing
3957the checks required for storage key protection as one operation (as opposed to
3958user space getting the storage keys, performing the checks, and accessing
3959memory thereafter, which could lead to a delay between check and access).
3960Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION
3961has the KVM_S390_MEMOP_EXTENSION_CAP_BASE bit set.
3962Currently absolute accesses are not permitted for VCPU ioctls.
3963Absolute accesses are permitted for non-protected guests only.
3964
3965Supported flags:
3966  * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3967  * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3968
3969The semantics of the flags common with logical accesses are as for logical
3970accesses.
3971
3972Absolute cmpxchg:
3973^^^^^^^^^^^^^^^^^
3974
3975Perform cmpxchg on absolute guest memory. Intended for use with the
3976KVM_S390_MEMOP_F_SKEY_PROTECTION flag.
3977Instead of doing an unconditional write, the access occurs only if the target
3978location contains the value pointed to by "old_addr".
3979This is performed as an atomic cmpxchg with the length specified by the "size"
3980parameter. "size" must be a power of two up to and including 16.
3981If the exchange did not take place because the target value doesn't match the
3982old value, the value "old_addr" points to is replaced by the target value.
3983User space can tell if an exchange took place by checking if this replacement
3984occurred. The cmpxchg op is permitted for the VM ioctl if
3985KVM_CAP_S390_MEM_OP_EXTENSION has flag KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG set.
3986
3987Supported flags:
3988  * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3989
3990SIDA read/write:
3991^^^^^^^^^^^^^^^^
3992
3993Access the secure instruction data area which contains memory operands necessary
3994for instruction emulation for protected guests.
3995SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available.
3996SIDA accesses are permitted for the VCPU ioctl only.
3997SIDA accesses are permitted for protected guests only.
3998
3999No flags are supported.
4000
40014.90 KVM_S390_GET_SKEYS
4002-----------------------
4003
4004:Capability: KVM_CAP_S390_SKEYS
4005:Architectures: s390
4006:Type: vm ioctl
4007:Parameters: struct kvm_s390_skeys
4008:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage
4009          keys, negative value on error
4010
4011This ioctl is used to get guest storage key values on the s390
4012architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
4013
4014  struct kvm_s390_skeys {
4015	__u64 start_gfn;
4016	__u64 count;
4017	__u64 skeydata_addr;
4018	__u32 flags;
4019	__u32 reserved[9];
4020  };
4021
4022The start_gfn field is the number of the first guest frame whose storage keys
4023you want to get.
4024
4025The count field is the number of consecutive frames (starting from start_gfn)
4026whose storage keys to get. The count field must be at least 1 and the maximum
4027allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
4028will cause the ioctl to return -EINVAL.
4029
4030The skeydata_addr field is the address to a buffer large enough to hold count
4031bytes. This buffer will be filled with storage key data by the ioctl.
4032
40334.91 KVM_S390_SET_SKEYS
4034-----------------------
4035
4036:Capability: KVM_CAP_S390_SKEYS
4037:Architectures: s390
4038:Type: vm ioctl
4039:Parameters: struct kvm_s390_skeys
4040:Returns: 0 on success, negative value on error
4041
4042This ioctl is used to set guest storage key values on the s390
4043architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
4044See section on KVM_S390_GET_SKEYS for struct definition.
4045
4046The start_gfn field is the number of the first guest frame whose storage keys
4047you want to set.
4048
4049The count field is the number of consecutive frames (starting from start_gfn)
4050whose storage keys to get. The count field must be at least 1 and the maximum
4051allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
4052will cause the ioctl to return -EINVAL.
4053
4054The skeydata_addr field is the address to a buffer containing count bytes of
4055storage keys. Each byte in the buffer will be set as the storage key for a
4056single frame starting at start_gfn for count frames.
4057
4058Note: If any architecturally invalid key value is found in the given data then
4059the ioctl will return -EINVAL.
4060
40614.92 KVM_S390_IRQ
4062-----------------
4063
4064:Capability: KVM_CAP_S390_INJECT_IRQ
4065:Architectures: s390
4066:Type: vcpu ioctl
4067:Parameters: struct kvm_s390_irq (in)
4068:Returns: 0 on success, -1 on error
4069
4070Errors:
4071
4072
4073  ======  =================================================================
4074  EINVAL  interrupt type is invalid
4075          type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
4076          type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
4077          than the maximum of VCPUs
4078  EBUSY   type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
4079          type is KVM_S390_SIGP_STOP and a stop irq is already pending,
4080          type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
4081          is already pending
4082  ======  =================================================================
4083
4084Allows to inject an interrupt to the guest.
4085
4086Using struct kvm_s390_irq as a parameter allows
4087to inject additional payload which is not
4088possible via KVM_S390_INTERRUPT.
4089
4090Interrupt parameters are passed via kvm_s390_irq::
4091
4092  struct kvm_s390_irq {
4093	__u64 type;
4094	union {
4095		struct kvm_s390_io_info io;
4096		struct kvm_s390_ext_info ext;
4097		struct kvm_s390_pgm_info pgm;
4098		struct kvm_s390_emerg_info emerg;
4099		struct kvm_s390_extcall_info extcall;
4100		struct kvm_s390_prefix_info prefix;
4101		struct kvm_s390_stop_info stop;
4102		struct kvm_s390_mchk_info mchk;
4103		char reserved[64];
4104	} u;
4105  };
4106
4107type can be one of the following:
4108
4109- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
4110- KVM_S390_PROGRAM_INT - program check; parameters in .pgm
4111- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
4112- KVM_S390_RESTART - restart; no parameters
4113- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
4114- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
4115- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
4116- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
4117- KVM_S390_MCHK - machine check interrupt; parameters in .mchk
4118
4119This is an asynchronous vcpu ioctl and can be invoked from any thread.
4120
41214.94 KVM_S390_GET_IRQ_STATE
4122---------------------------
4123
4124:Capability: KVM_CAP_S390_IRQ_STATE
4125:Architectures: s390
4126:Type: vcpu ioctl
4127:Parameters: struct kvm_s390_irq_state (out)
4128:Returns: >= number of bytes copied into buffer,
4129          -EINVAL if buffer size is 0,
4130          -ENOBUFS if buffer size is too small to fit all pending interrupts,
4131          -EFAULT if the buffer address was invalid
4132
4133This ioctl allows userspace to retrieve the complete state of all currently
4134pending interrupts in a single buffer. Use cases include migration
4135and introspection. The parameter structure contains the address of a
4136userspace buffer and its length::
4137
4138  struct kvm_s390_irq_state {
4139	__u64 buf;
4140	__u32 flags;        /* will stay unused for compatibility reasons */
4141	__u32 len;
4142	__u32 reserved[4];  /* will stay unused for compatibility reasons */
4143  };
4144
4145Userspace passes in the above struct and for each pending interrupt a
4146struct kvm_s390_irq is copied to the provided buffer.
4147
4148The structure contains a flags and a reserved field for future extensions. As
4149the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
4150reserved, these fields can not be used in the future without breaking
4151compatibility.
4152
4153If -ENOBUFS is returned the buffer provided was too small and userspace
4154may retry with a bigger buffer.
4155
41564.95 KVM_S390_SET_IRQ_STATE
4157---------------------------
4158
4159:Capability: KVM_CAP_S390_IRQ_STATE
4160:Architectures: s390
4161:Type: vcpu ioctl
4162:Parameters: struct kvm_s390_irq_state (in)
4163:Returns: 0 on success,
4164          -EFAULT if the buffer address was invalid,
4165          -EINVAL for an invalid buffer length (see below),
4166          -EBUSY if there were already interrupts pending,
4167          errors occurring when actually injecting the
4168          interrupt. See KVM_S390_IRQ.
4169
4170This ioctl allows userspace to set the complete state of all cpu-local
4171interrupts currently pending for the vcpu. It is intended for restoring
4172interrupt state after a migration. The input parameter is a userspace buffer
4173containing a struct kvm_s390_irq_state::
4174
4175  struct kvm_s390_irq_state {
4176	__u64 buf;
4177	__u32 flags;        /* will stay unused for compatibility reasons */
4178	__u32 len;
4179	__u32 reserved[4];  /* will stay unused for compatibility reasons */
4180  };
4181
4182The restrictions for flags and reserved apply as well.
4183(see KVM_S390_GET_IRQ_STATE)
4184
4185The userspace memory referenced by buf contains a struct kvm_s390_irq
4186for each interrupt to be injected into the guest.
4187If one of the interrupts could not be injected for some reason the
4188ioctl aborts.
4189
4190len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
4191and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
4192which is the maximum number of possibly pending cpu-local interrupts.
4193
41944.96 KVM_SMI
4195------------
4196
4197:Capability: KVM_CAP_X86_SMM
4198:Architectures: x86
4199:Type: vcpu ioctl
4200:Parameters: none
4201:Returns: 0 on success, -1 on error
4202
4203Queues an SMI on the thread's vcpu.
4204
42054.97 KVM_X86_SET_MSR_FILTER
4206----------------------------
4207
4208:Capability: KVM_CAP_X86_MSR_FILTER
4209:Architectures: x86
4210:Type: vm ioctl
4211:Parameters: struct kvm_msr_filter
4212:Returns: 0 on success, < 0 on error
4213
4214::
4215
4216  struct kvm_msr_filter_range {
4217  #define KVM_MSR_FILTER_READ  (1 << 0)
4218  #define KVM_MSR_FILTER_WRITE (1 << 1)
4219	__u32 flags;
4220	__u32 nmsrs; /* number of msrs in bitmap */
4221	__u32 base;  /* MSR index the bitmap starts at */
4222	__u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4223  };
4224
4225  #define KVM_MSR_FILTER_MAX_RANGES 16
4226  struct kvm_msr_filter {
4227  #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4228  #define KVM_MSR_FILTER_DEFAULT_DENY  (1 << 0)
4229	__u32 flags;
4230	struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4231  };
4232
4233flags values for ``struct kvm_msr_filter_range``:
4234
4235``KVM_MSR_FILTER_READ``
4236
4237  Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4238  indicates that read accesses should be denied, while a 1 indicates that
4239  a read for a particular MSR should be allowed regardless of the default
4240  filter action.
4241
4242``KVM_MSR_FILTER_WRITE``
4243
4244  Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4245  indicates that write accesses should be denied, while a 1 indicates that
4246  a write for a particular MSR should be allowed regardless of the default
4247  filter action.
4248
4249flags values for ``struct kvm_msr_filter``:
4250
4251``KVM_MSR_FILTER_DEFAULT_ALLOW``
4252
4253  If no filter range matches an MSR index that is getting accessed, KVM will
4254  allow accesses to all MSRs by default.
4255
4256``KVM_MSR_FILTER_DEFAULT_DENY``
4257
4258  If no filter range matches an MSR index that is getting accessed, KVM will
4259  deny accesses to all MSRs by default.
4260
4261This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny
4262guest MSR accesses that would normally be allowed by KVM.  If an MSR is not
4263covered by a specific range, the "default" filtering behavior applies.  Each
4264bitmap range covers MSRs from [base .. base+nmsrs).
4265
4266If an MSR access is denied by userspace, the resulting KVM behavior depends on
4267whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is
4268enabled.  If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace
4269on denied accesses, i.e. userspace effectively intercepts the MSR access.  If
4270KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest
4271on denied accesses.  Note, if an MSR access is denied during emulation of MSR
4272load/stores during VMX transitions, KVM ignores KVM_MSR_EXIT_REASON_FILTER.
4273See the below warning for full details.
4274
4275If an MSR access is allowed by userspace, KVM will emulate and/or virtualize
4276the access in accordance with the vCPU model.  Note, KVM may still ultimately
4277inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support
4278the MSR, or to follow architectural behavior for the MSR.
4279
4280By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range
4281filters.
4282
4283Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4284filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4285an error.
4286
4287.. warning::
4288   MSR accesses that are side effects of instruction execution (emulated or
4289   native) are not filtered as hardware does not honor MSR bitmaps outside of
4290   RDMSR and WRMSR, and KVM mimics that behavior when emulating instructions
4291   to avoid pointless divergence from hardware.  E.g. RDPID reads MSR_TSC_AUX,
4292   SYSENTER reads the SYSENTER MSRs, etc.
4293
4294   MSRs that are loaded/stored via dedicated VMCS fields are not filtered as
4295   part of VM-Enter/VM-Exit emulation.
4296
4297   MSRs that are loaded/store via VMX's load/store lists _are_ filtered as part
4298   of VM-Enter/VM-Exit emulation.  If an MSR access is denied on VM-Enter, KVM
4299   synthesizes a consistency check VM-Exit(EXIT_REASON_MSR_LOAD_FAIL).  If an
4300   MSR access is denied on VM-Exit, KVM synthesizes a VM-Abort.  In short, KVM
4301   extends Intel's architectural list of MSRs that cannot be loaded/saved via
4302   the VM-Enter/VM-Exit MSR list.  It is platform owner's responsibility to
4303   to communicate any such restrictions to their end users.
4304
4305   x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that
4306   cover any x2APIC MSRs).
4307
4308Note, invoking this ioctl while a vCPU is running is inherently racy.  However,
4309KVM does guarantee that vCPUs will see either the previous filter or the new
4310filter, e.g. MSRs with identical settings in both the old and new filter will
4311have deterministic behavior.
4312
4313Similarly, if userspace wishes to intercept on denied accesses,
4314KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and
4315left enabled until after all filters are deactivated.  Failure to do so may
4316result in KVM injecting a #GP instead of exiting to userspace.
4317
43184.98 KVM_CREATE_SPAPR_TCE_64
4319----------------------------
4320
4321:Capability: KVM_CAP_SPAPR_TCE_64
4322:Architectures: powerpc
4323:Type: vm ioctl
4324:Parameters: struct kvm_create_spapr_tce_64 (in)
4325:Returns: file descriptor for manipulating the created TCE table
4326
4327This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4328windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4329
4330This capability uses extended struct in ioctl interface::
4331
4332  /* for KVM_CAP_SPAPR_TCE_64 */
4333  struct kvm_create_spapr_tce_64 {
4334	__u64 liobn;
4335	__u32 page_shift;
4336	__u32 flags;
4337	__u64 offset;	/* in pages */
4338	__u64 size; 	/* in pages */
4339  };
4340
4341The aim of extension is to support an additional bigger DMA window with
4342a variable page size.
4343KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4344a bus offset of the corresponding DMA window, @size and @offset are numbers
4345of IOMMU pages.
4346
4347@flags are not used at the moment.
4348
4349The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4350
43514.99 KVM_REINJECT_CONTROL
4352-------------------------
4353
4354:Capability: KVM_CAP_REINJECT_CONTROL
4355:Architectures: x86
4356:Type: vm ioctl
4357:Parameters: struct kvm_reinject_control (in)
4358:Returns: 0 on success,
4359         -EFAULT if struct kvm_reinject_control cannot be read,
4360         -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4361
4362i8254 (PIT) has two modes, reinject and !reinject.  The default is reinject,
4363where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4364vector(s) that i8254 injects.  Reinject mode dequeues a tick and injects its
4365interrupt whenever there isn't a pending interrupt from i8254.
4366!reinject mode injects an interrupt as soon as a tick arrives.
4367
4368::
4369
4370  struct kvm_reinject_control {
4371	__u8 pit_reinject;
4372	__u8 reserved[31];
4373  };
4374
4375pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4376operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4377
43784.100 KVM_PPC_CONFIGURE_V3_MMU
4379------------------------------
4380
4381:Capability: KVM_CAP_PPC_MMU_RADIX or KVM_CAP_PPC_MMU_HASH_V3
4382:Architectures: ppc
4383:Type: vm ioctl
4384:Parameters: struct kvm_ppc_mmuv3_cfg (in)
4385:Returns: 0 on success,
4386         -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4387         -EINVAL if the configuration is invalid
4388
4389This ioctl controls whether the guest will use radix or HPT (hashed
4390page table) translation, and sets the pointer to the process table for
4391the guest.
4392
4393::
4394
4395  struct kvm_ppc_mmuv3_cfg {
4396	__u64	flags;
4397	__u64	process_table;
4398  };
4399
4400There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4401KVM_PPC_MMUV3_GTSE.  KVM_PPC_MMUV3_RADIX, if set, configures the guest
4402to use radix tree translation, and if clear, to use HPT translation.
4403KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4404to be able to use the global TLB and SLB invalidation instructions;
4405if clear, the guest may not use these instructions.
4406
4407The process_table field specifies the address and size of the guest
4408process table, which is in the guest's space.  This field is formatted
4409as the second doubleword of the partition table entry, as defined in
4410the Power ISA V3.00, Book III section 5.7.6.1.
4411
44124.101 KVM_PPC_GET_RMMU_INFO
4413---------------------------
4414
4415:Capability: KVM_CAP_PPC_MMU_RADIX
4416:Architectures: ppc
4417:Type: vm ioctl
4418:Parameters: struct kvm_ppc_rmmu_info (out)
4419:Returns: 0 on success,
4420	 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4421	 -EINVAL if no useful information can be returned
4422
4423This ioctl returns a structure containing two things: (a) a list
4424containing supported radix tree geometries, and (b) a list that maps
4425page sizes to put in the "AP" (actual page size) field for the tlbie
4426(TLB invalidate entry) instruction.
4427
4428::
4429
4430  struct kvm_ppc_rmmu_info {
4431	struct kvm_ppc_radix_geom {
4432		__u8	page_shift;
4433		__u8	level_bits[4];
4434		__u8	pad[3];
4435	}	geometries[8];
4436	__u32	ap_encodings[8];
4437  };
4438
4439The geometries[] field gives up to 8 supported geometries for the
4440radix page table, in terms of the log base 2 of the smallest page
4441size, and the number of bits indexed at each level of the tree, from
4442the PTE level up to the PGD level in that order.  Any unused entries
4443will have 0 in the page_shift field.
4444
4445The ap_encodings gives the supported page sizes and their AP field
4446encodings, encoded with the AP value in the top 3 bits and the log
4447base 2 of the page size in the bottom 6 bits.
4448
44494.102 KVM_PPC_RESIZE_HPT_PREPARE
4450--------------------------------
4451
4452:Capability: KVM_CAP_SPAPR_RESIZE_HPT
4453:Architectures: powerpc
4454:Type: vm ioctl
4455:Parameters: struct kvm_ppc_resize_hpt (in)
4456:Returns: 0 on successful completion,
4457	 >0 if a new HPT is being prepared, the value is an estimated
4458         number of milliseconds until preparation is complete,
4459         -EFAULT if struct kvm_reinject_control cannot be read,
4460	 -EINVAL if the supplied shift or flags are invalid,
4461	 -ENOMEM if unable to allocate the new HPT,
4462
4463Used to implement the PAPR extension for runtime resizing of a guest's
4464Hashed Page Table (HPT).  Specifically this starts, stops or monitors
4465the preparation of a new potential HPT for the guest, essentially
4466implementing the H_RESIZE_HPT_PREPARE hypercall.
4467
4468::
4469
4470  struct kvm_ppc_resize_hpt {
4471	__u64 flags;
4472	__u32 shift;
4473	__u32 pad;
4474  };
4475
4476If called with shift > 0 when there is no pending HPT for the guest,
4477this begins preparation of a new pending HPT of size 2^(shift) bytes.
4478It then returns a positive integer with the estimated number of
4479milliseconds until preparation is complete.
4480
4481If called when there is a pending HPT whose size does not match that
4482requested in the parameters, discards the existing pending HPT and
4483creates a new one as above.
4484
4485If called when there is a pending HPT of the size requested, will:
4486
4487  * If preparation of the pending HPT is already complete, return 0
4488  * If preparation of the pending HPT has failed, return an error
4489    code, then discard the pending HPT.
4490  * If preparation of the pending HPT is still in progress, return an
4491    estimated number of milliseconds until preparation is complete.
4492
4493If called with shift == 0, discards any currently pending HPT and
4494returns 0 (i.e. cancels any in-progress preparation).
4495
4496flags is reserved for future expansion, currently setting any bits in
4497flags will result in an -EINVAL.
4498
4499Normally this will be called repeatedly with the same parameters until
4500it returns <= 0.  The first call will initiate preparation, subsequent
4501ones will monitor preparation until it completes or fails.
4502
45034.103 KVM_PPC_RESIZE_HPT_COMMIT
4504-------------------------------
4505
4506:Capability: KVM_CAP_SPAPR_RESIZE_HPT
4507:Architectures: powerpc
4508:Type: vm ioctl
4509:Parameters: struct kvm_ppc_resize_hpt (in)
4510:Returns: 0 on successful completion,
4511         -EFAULT if struct kvm_reinject_control cannot be read,
4512	 -EINVAL if the supplied shift or flags are invalid,
4513	 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4514         have the requested size,
4515	 -EBUSY if the pending HPT is not fully prepared,
4516	 -ENOSPC if there was a hash collision when moving existing
4517         HPT entries to the new HPT,
4518	 -EIO on other error conditions
4519
4520Used to implement the PAPR extension for runtime resizing of a guest's
4521Hashed Page Table (HPT).  Specifically this requests that the guest be
4522transferred to working with the new HPT, essentially implementing the
4523H_RESIZE_HPT_COMMIT hypercall.
4524
4525::
4526
4527  struct kvm_ppc_resize_hpt {
4528	__u64 flags;
4529	__u32 shift;
4530	__u32 pad;
4531  };
4532
4533This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4534returned 0 with the same parameters.  In other cases
4535KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4536-EBUSY, though others may be possible if the preparation was started,
4537but failed).
4538
4539This will have undefined effects on the guest if it has not already
4540placed itself in a quiescent state where no vcpu will make MMU enabled
4541memory accesses.
4542
4543On successful completion, the pending HPT will become the guest's active
4544HPT and the previous HPT will be discarded.
4545
4546On failure, the guest will still be operating on its previous HPT.
4547
45484.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4549-----------------------------------
4550
4551:Capability: KVM_CAP_MCE
4552:Architectures: x86
4553:Type: system ioctl
4554:Parameters: u64 mce_cap (out)
4555:Returns: 0 on success, -1 on error
4556
4557Returns supported MCE capabilities. The u64 mce_cap parameter
4558has the same format as the MSR_IA32_MCG_CAP register. Supported
4559capabilities will have the corresponding bits set.
4560
45614.105 KVM_X86_SETUP_MCE
4562-----------------------
4563
4564:Capability: KVM_CAP_MCE
4565:Architectures: x86
4566:Type: vcpu ioctl
4567:Parameters: u64 mcg_cap (in)
4568:Returns: 0 on success,
4569         -EFAULT if u64 mcg_cap cannot be read,
4570         -EINVAL if the requested number of banks is invalid,
4571         -EINVAL if requested MCE capability is not supported.
4572
4573Initializes MCE support for use. The u64 mcg_cap parameter
4574has the same format as the MSR_IA32_MCG_CAP register and
4575specifies which capabilities should be enabled. The maximum
4576supported number of error-reporting banks can be retrieved when
4577checking for KVM_CAP_MCE. The supported capabilities can be
4578retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4579
45804.106 KVM_X86_SET_MCE
4581---------------------
4582
4583:Capability: KVM_CAP_MCE
4584:Architectures: x86
4585:Type: vcpu ioctl
4586:Parameters: struct kvm_x86_mce (in)
4587:Returns: 0 on success,
4588         -EFAULT if struct kvm_x86_mce cannot be read,
4589         -EINVAL if the bank number is invalid,
4590         -EINVAL if VAL bit is not set in status field.
4591
4592Inject a machine check error (MCE) into the guest. The input
4593parameter is::
4594
4595  struct kvm_x86_mce {
4596	__u64 status;
4597	__u64 addr;
4598	__u64 misc;
4599	__u64 mcg_status;
4600	__u8 bank;
4601	__u8 pad1[7];
4602	__u64 pad2[3];
4603  };
4604
4605If the MCE being reported is an uncorrected error, KVM will
4606inject it as an MCE exception into the guest. If the guest
4607MCG_STATUS register reports that an MCE is in progress, KVM
4608causes an KVM_EXIT_SHUTDOWN vmexit.
4609
4610Otherwise, if the MCE is a corrected error, KVM will just
4611store it in the corresponding bank (provided this bank is
4612not holding a previously reported uncorrected error).
4613
46144.107 KVM_S390_GET_CMMA_BITS
4615----------------------------
4616
4617:Capability: KVM_CAP_S390_CMMA_MIGRATION
4618:Architectures: s390
4619:Type: vm ioctl
4620:Parameters: struct kvm_s390_cmma_log (in, out)
4621:Returns: 0 on success, a negative value on error
4622
4623Errors:
4624
4625  ======     =============================================================
4626  ENOMEM     not enough memory can be allocated to complete the task
4627  ENXIO      if CMMA is not enabled
4628  EINVAL     if KVM_S390_CMMA_PEEK is not set but migration mode was not enabled
4629  EINVAL     if KVM_S390_CMMA_PEEK is not set but dirty tracking has been
4630             disabled (and thus migration mode was automatically disabled)
4631  EFAULT     if the userspace address is invalid or if no page table is
4632             present for the addresses (e.g. when using hugepages).
4633  ======     =============================================================
4634
4635This ioctl is used to get the values of the CMMA bits on the s390
4636architecture. It is meant to be used in two scenarios:
4637
4638- During live migration to save the CMMA values. Live migration needs
4639  to be enabled via the KVM_REQ_START_MIGRATION VM property.
4640- To non-destructively peek at the CMMA values, with the flag
4641  KVM_S390_CMMA_PEEK set.
4642
4643The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4644values are written to a buffer whose location is indicated via the "values"
4645member in the kvm_s390_cmma_log struct.  The values in the input struct are
4646also updated as needed.
4647
4648Each CMMA value takes up one byte.
4649
4650::
4651
4652  struct kvm_s390_cmma_log {
4653	__u64 start_gfn;
4654	__u32 count;
4655	__u32 flags;
4656	union {
4657		__u64 remaining;
4658		__u64 mask;
4659	};
4660	__u64 values;
4661  };
4662
4663start_gfn is the number of the first guest frame whose CMMA values are
4664to be retrieved,
4665
4666count is the length of the buffer in bytes,
4667
4668values points to the buffer where the result will be written to.
4669
4670If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4671KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4672other ioctls.
4673
4674The result is written in the buffer pointed to by the field values, and
4675the values of the input parameter are updated as follows.
4676
4677Depending on the flags, different actions are performed. The only
4678supported flag so far is KVM_S390_CMMA_PEEK.
4679
4680The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4681start_gfn will indicate the first page frame whose CMMA bits were dirty.
4682It is not necessarily the same as the one passed as input, as clean pages
4683are skipped.
4684
4685count will indicate the number of bytes actually written in the buffer.
4686It can (and very often will) be smaller than the input value, since the
4687buffer is only filled until 16 bytes of clean values are found (which
4688are then not copied in the buffer). Since a CMMA migration block needs
4689the base address and the length, for a total of 16 bytes, we will send
4690back some clean data if there is some dirty data afterwards, as long as
4691the size of the clean data does not exceed the size of the header. This
4692allows to minimize the amount of data to be saved or transferred over
4693the network at the expense of more roundtrips to userspace. The next
4694invocation of the ioctl will skip over all the clean values, saving
4695potentially more than just the 16 bytes we found.
4696
4697If KVM_S390_CMMA_PEEK is set:
4698the existing storage attributes are read even when not in migration
4699mode, and no other action is performed;
4700
4701the output start_gfn will be equal to the input start_gfn,
4702
4703the output count will be equal to the input count, except if the end of
4704memory has been reached.
4705
4706In both cases:
4707the field "remaining" will indicate the total number of dirty CMMA values
4708still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4709not enabled.
4710
4711mask is unused.
4712
4713values points to the userspace buffer where the result will be stored.
4714
47154.108 KVM_S390_SET_CMMA_BITS
4716----------------------------
4717
4718:Capability: KVM_CAP_S390_CMMA_MIGRATION
4719:Architectures: s390
4720:Type: vm ioctl
4721:Parameters: struct kvm_s390_cmma_log (in)
4722:Returns: 0 on success, a negative value on error
4723
4724This ioctl is used to set the values of the CMMA bits on the s390
4725architecture. It is meant to be used during live migration to restore
4726the CMMA values, but there are no restrictions on its use.
4727The ioctl takes parameters via the kvm_s390_cmma_values struct.
4728Each CMMA value takes up one byte.
4729
4730::
4731
4732  struct kvm_s390_cmma_log {
4733	__u64 start_gfn;
4734	__u32 count;
4735	__u32 flags;
4736	union {
4737		__u64 remaining;
4738		__u64 mask;
4739 	};
4740	__u64 values;
4741  };
4742
4743start_gfn indicates the starting guest frame number,
4744
4745count indicates how many values are to be considered in the buffer,
4746
4747flags is not used and must be 0.
4748
4749mask indicates which PGSTE bits are to be considered.
4750
4751remaining is not used.
4752
4753values points to the buffer in userspace where to store the values.
4754
4755This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4756complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4757the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4758if the flags field was not 0, with -EFAULT if the userspace address is
4759invalid, if invalid pages are written to (e.g. after the end of memory)
4760or if no page table is present for the addresses (e.g. when using
4761hugepages).
4762
47634.109 KVM_PPC_GET_CPU_CHAR
4764--------------------------
4765
4766:Capability: KVM_CAP_PPC_GET_CPU_CHAR
4767:Architectures: powerpc
4768:Type: vm ioctl
4769:Parameters: struct kvm_ppc_cpu_char (out)
4770:Returns: 0 on successful completion,
4771	 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4772
4773This ioctl gives userspace information about certain characteristics
4774of the CPU relating to speculative execution of instructions and
4775possible information leakage resulting from speculative execution (see
4776CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754).  The information is
4777returned in struct kvm_ppc_cpu_char, which looks like this::
4778
4779  struct kvm_ppc_cpu_char {
4780	__u64	character;		/* characteristics of the CPU */
4781	__u64	behaviour;		/* recommended software behaviour */
4782	__u64	character_mask;		/* valid bits in character */
4783	__u64	behaviour_mask;		/* valid bits in behaviour */
4784  };
4785
4786For extensibility, the character_mask and behaviour_mask fields
4787indicate which bits of character and behaviour have been filled in by
4788the kernel.  If the set of defined bits is extended in future then
4789userspace will be able to tell whether it is running on a kernel that
4790knows about the new bits.
4791
4792The character field describes attributes of the CPU which can help
4793with preventing inadvertent information disclosure - specifically,
4794whether there is an instruction to flash-invalidate the L1 data cache
4795(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4796to a mode where entries can only be used by the thread that created
4797them, whether the bcctr[l] instruction prevents speculation, and
4798whether a speculation barrier instruction (ori 31,31,0) is provided.
4799
4800The behaviour field describes actions that software should take to
4801prevent inadvertent information disclosure, and thus describes which
4802vulnerabilities the hardware is subject to; specifically whether the
4803L1 data cache should be flushed when returning to user mode from the
4804kernel, and whether a speculation barrier should be placed between an
4805array bounds check and the array access.
4806
4807These fields use the same bit definitions as the new
4808H_GET_CPU_CHARACTERISTICS hypercall.
4809
48104.110 KVM_MEMORY_ENCRYPT_OP
4811---------------------------
4812
4813:Capability: basic
4814:Architectures: x86
4815:Type: vm ioctl, vcpu ioctl
4816:Parameters: an opaque platform specific structure (in/out)
4817:Returns: 0 on success; -1 on error
4818
4819If the platform supports creating encrypted VMs then this ioctl can be used
4820for issuing platform-specific memory encryption commands to manage those
4821encrypted VMs.
4822
4823Currently, this ioctl is used for issuing both Secure Encrypted Virtualization
4824(SEV) commands on AMD Processors and Trusted Domain Extensions (TDX) commands
4825on Intel Processors.  The detailed commands are defined in
4826Documentation/virt/kvm/x86/amd-memory-encryption.rst and
4827Documentation/virt/kvm/x86/intel-tdx.rst.
4828
48294.111 KVM_MEMORY_ENCRYPT_REG_REGION
4830-----------------------------------
4831
4832:Capability: basic
4833:Architectures: x86
4834:Type: system
4835:Parameters: struct kvm_enc_region (in)
4836:Returns: 0 on success; -1 on error
4837
4838This ioctl can be used to register a guest memory region which may
4839contain encrypted data (e.g. guest RAM, SMRAM etc).
4840
4841It is used in the SEV-enabled guest. When encryption is enabled, a guest
4842memory region may contain encrypted data. The SEV memory encryption
4843engine uses a tweak such that two identical plaintext pages, each at
4844different locations will have differing ciphertexts. So swapping or
4845moving ciphertext of those pages will not result in plaintext being
4846swapped. So relocating (or migrating) physical backing pages for the SEV
4847guest will require some additional steps.
4848
4849Note: The current SEV key management spec does not provide commands to
4850swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4851memory region registered with the ioctl.
4852
48534.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4854-------------------------------------
4855
4856:Capability: basic
4857:Architectures: x86
4858:Type: system
4859:Parameters: struct kvm_enc_region (in)
4860:Returns: 0 on success; -1 on error
4861
4862This ioctl can be used to unregister the guest memory region registered
4863with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4864
48654.113 KVM_HYPERV_EVENTFD
4866------------------------
4867
4868:Capability: KVM_CAP_HYPERV_EVENTFD
4869:Architectures: x86
4870:Type: vm ioctl
4871:Parameters: struct kvm_hyperv_eventfd (in)
4872
4873This ioctl (un)registers an eventfd to receive notifications from the guest on
4874the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4875causing a user exit.  SIGNAL_EVENT hypercall with non-zero event flag number
4876(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4877
4878::
4879
4880  struct kvm_hyperv_eventfd {
4881	__u32 conn_id;
4882	__s32 fd;
4883	__u32 flags;
4884	__u32 padding[3];
4885  };
4886
4887The conn_id field should fit within 24 bits::
4888
4889  #define KVM_HYPERV_CONN_ID_MASK		0x00ffffff
4890
4891The acceptable values for the flags field are::
4892
4893  #define KVM_HYPERV_EVENTFD_DEASSIGN	(1 << 0)
4894
4895:Returns: 0 on success,
4896 	  -EINVAL if conn_id or flags is outside the allowed range,
4897	  -ENOENT on deassign if the conn_id isn't registered,
4898	  -EEXIST on assign if the conn_id is already registered
4899
49004.114 KVM_GET_NESTED_STATE
4901--------------------------
4902
4903:Capability: KVM_CAP_NESTED_STATE
4904:Architectures: x86
4905:Type: vcpu ioctl
4906:Parameters: struct kvm_nested_state (in/out)
4907:Returns: 0 on success, -1 on error
4908
4909Errors:
4910
4911  =====      =============================================================
4912  E2BIG      the total state size exceeds the value of 'size' specified by
4913             the user; the size required will be written into size.
4914  =====      =============================================================
4915
4916::
4917
4918  struct kvm_nested_state {
4919	__u16 flags;
4920	__u16 format;
4921	__u32 size;
4922
4923	union {
4924		struct kvm_vmx_nested_state_hdr vmx;
4925		struct kvm_svm_nested_state_hdr svm;
4926
4927		/* Pad the header to 128 bytes.  */
4928		__u8 pad[120];
4929	} hdr;
4930
4931	union {
4932		struct kvm_vmx_nested_state_data vmx[0];
4933		struct kvm_svm_nested_state_data svm[0];
4934	} data;
4935  };
4936
4937  #define KVM_STATE_NESTED_GUEST_MODE		0x00000001
4938  #define KVM_STATE_NESTED_RUN_PENDING		0x00000002
4939  #define KVM_STATE_NESTED_EVMCS		0x00000004
4940
4941  #define KVM_STATE_NESTED_FORMAT_VMX		0
4942  #define KVM_STATE_NESTED_FORMAT_SVM		1
4943
4944  #define KVM_STATE_NESTED_VMX_VMCS_SIZE	0x1000
4945
4946  #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE	0x00000001
4947  #define KVM_STATE_NESTED_VMX_SMM_VMXON	0x00000002
4948
4949  #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4950
4951  struct kvm_vmx_nested_state_hdr {
4952	__u64 vmxon_pa;
4953	__u64 vmcs12_pa;
4954
4955	struct {
4956		__u16 flags;
4957	} smm;
4958
4959	__u32 flags;
4960	__u64 preemption_timer_deadline;
4961  };
4962
4963  struct kvm_vmx_nested_state_data {
4964	__u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4965	__u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4966  };
4967
4968This ioctl copies the vcpu's nested virtualization state from the kernel to
4969userspace.
4970
4971The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4972to the KVM_CHECK_EXTENSION ioctl().
4973
49744.115 KVM_SET_NESTED_STATE
4975--------------------------
4976
4977:Capability: KVM_CAP_NESTED_STATE
4978:Architectures: x86
4979:Type: vcpu ioctl
4980:Parameters: struct kvm_nested_state (in)
4981:Returns: 0 on success, -1 on error
4982
4983This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4984For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4985
49864.116 KVM_(UN)REGISTER_COALESCED_MMIO
4987-------------------------------------
4988
4989:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4990	     KVM_CAP_COALESCED_PIO (for coalesced pio)
4991:Architectures: all
4992:Type: vm ioctl
4993:Parameters: struct kvm_coalesced_mmio_zone
4994:Returns: 0 on success, < 0 on error
4995
4996Coalesced I/O is a performance optimization that defers hardware
4997register write emulation so that userspace exits are avoided.  It is
4998typically used to reduce the overhead of emulating frequently accessed
4999hardware registers.
5000
5001When a hardware register is configured for coalesced I/O, write accesses
5002do not exit to userspace and their value is recorded in a ring buffer
5003that is shared between kernel and userspace.
5004
5005Coalesced I/O is used if one or more write accesses to a hardware
5006register can be deferred until a read or a write to another hardware
5007register on the same device.  This last access will cause a vmexit and
5008userspace will process accesses from the ring buffer before emulating
5009it. That will avoid exiting to userspace on repeated writes.
5010
5011Coalesced pio is based on coalesced mmio. There is little difference
5012between coalesced mmio and pio except that coalesced pio records accesses
5013to I/O ports.
5014
50154.117 KVM_CLEAR_DIRTY_LOG
5016-------------------------
5017
5018:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
5019:Architectures: x86, arm64, mips
5020:Type: vm ioctl
5021:Parameters: struct kvm_clear_dirty_log (in)
5022:Returns: 0 on success, -1 on error
5023
5024::
5025
5026  /* for KVM_CLEAR_DIRTY_LOG */
5027  struct kvm_clear_dirty_log {
5028	__u32 slot;
5029	__u32 num_pages;
5030	__u64 first_page;
5031	union {
5032		void __user *dirty_bitmap; /* one bit per page */
5033		__u64 padding;
5034	};
5035  };
5036
5037The ioctl clears the dirty status of pages in a memory slot, according to
5038the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
5039field.  Bit 0 of the bitmap corresponds to page "first_page" in the
5040memory slot, and num_pages is the size in bits of the input bitmap.
5041first_page must be a multiple of 64; num_pages must also be a multiple of
504264 unless first_page + num_pages is the size of the memory slot.  For each
5043bit that is set in the input bitmap, the corresponding page is marked "clean"
5044in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
5045(for example via write-protection, or by clearing the dirty bit in
5046a page table entry).
5047
5048If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
5049the address space for which you want to clear the dirty status.  See
5050KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
5051
5052This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
5053is enabled; for more information, see the description of the capability.
5054However, it can always be used as long as KVM_CHECK_EXTENSION confirms
5055that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
5056
50574.118 KVM_GET_SUPPORTED_HV_CPUID
5058--------------------------------
5059
5060:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
5061:Architectures: x86
5062:Type: system ioctl, vcpu ioctl
5063:Parameters: struct kvm_cpuid2 (in/out)
5064:Returns: 0 on success, -1 on error
5065
5066::
5067
5068  struct kvm_cpuid2 {
5069	__u32 nent;
5070	__u32 padding;
5071	struct kvm_cpuid_entry2 entries[0];
5072  };
5073
5074  struct kvm_cpuid_entry2 {
5075	__u32 function;
5076	__u32 index;
5077	__u32 flags;
5078	__u32 eax;
5079	__u32 ebx;
5080	__u32 ecx;
5081	__u32 edx;
5082	__u32 padding[3];
5083  };
5084
5085This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
5086KVM.  Userspace can use the information returned by this ioctl to construct
5087cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
5088Windows or Hyper-V guests).
5089
5090CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
5091Functional Specification (TLFS). These leaves can't be obtained with
5092KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
5093leaves (0x40000000, 0x40000001).
5094
5095Currently, the following list of CPUID leaves are returned:
5096
5097 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
5098 - HYPERV_CPUID_INTERFACE
5099 - HYPERV_CPUID_VERSION
5100 - HYPERV_CPUID_FEATURES
5101 - HYPERV_CPUID_ENLIGHTMENT_INFO
5102 - HYPERV_CPUID_IMPLEMENT_LIMITS
5103 - HYPERV_CPUID_NESTED_FEATURES
5104 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
5105 - HYPERV_CPUID_SYNDBG_INTERFACE
5106 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
5107
5108Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
5109with the 'nent' field indicating the number of entries in the variable-size
5110array 'entries'.  If the number of entries is too low to describe all Hyper-V
5111feature leaves, an error (E2BIG) is returned. If the number is more or equal
5112to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
5113number of valid entries in the 'entries' array, which is then filled.
5114
5115'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
5116userspace should not expect to get any particular value there.
5117
5118Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
5119system ioctl which exposes all supported feature bits unconditionally, vcpu
5120version has the following quirks:
5121
5122- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
5123  feature bit are only exposed when Enlightened VMCS was previously enabled
5124  on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
5125- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
5126  (presumes KVM_CREATE_IRQCHIP has already been called).
5127
51284.119 KVM_ARM_VCPU_FINALIZE
5129---------------------------
5130
5131:Architectures: arm64
5132:Type: vcpu ioctl
5133:Parameters: int feature (in)
5134:Returns: 0 on success, -1 on error
5135
5136Errors:
5137
5138  ======     ==============================================================
5139  EPERM      feature not enabled, needs configuration, or already finalized
5140  EINVAL     feature unknown or not present
5141  ======     ==============================================================
5142
5143Recognised values for feature:
5144
5145  =====      ===========================================
5146  arm64      KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
5147  =====      ===========================================
5148
5149Finalizes the configuration of the specified vcpu feature.
5150
5151The vcpu must already have been initialised, enabling the affected feature, by
5152means of a successful :ref:`KVM_ARM_VCPU_INIT <KVM_ARM_VCPU_INIT>` call with the
5153appropriate flag set in features[].
5154
5155For affected vcpu features, this is a mandatory step that must be performed
5156before the vcpu is fully usable.
5157
5158Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
5159configured by use of ioctls such as KVM_SET_ONE_REG.  The exact configuration
5160that should be performed and how to do it are feature-dependent.
5161
5162Other calls that depend on a particular feature being finalized, such as
5163KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
5164-EPERM unless the feature has already been finalized by means of a
5165KVM_ARM_VCPU_FINALIZE call.
5166
5167See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
5168using this ioctl.
5169
51704.120 KVM_SET_PMU_EVENT_FILTER
5171------------------------------
5172
5173:Capability: KVM_CAP_PMU_EVENT_FILTER
5174:Architectures: x86
5175:Type: vm ioctl
5176:Parameters: struct kvm_pmu_event_filter (in)
5177:Returns: 0 on success, -1 on error
5178
5179Errors:
5180
5181  ======     ============================================================
5182  EFAULT     args[0] cannot be accessed
5183  EINVAL     args[0] contains invalid data in the filter or filter events
5184  E2BIG      nevents is too large
5185  EBUSY      not enough memory to allocate the filter
5186  ======     ============================================================
5187
5188::
5189
5190  struct kvm_pmu_event_filter {
5191	__u32 action;
5192	__u32 nevents;
5193	__u32 fixed_counter_bitmap;
5194	__u32 flags;
5195	__u32 pad[4];
5196	__u64 events[0];
5197  };
5198
5199This ioctl restricts the set of PMU events the guest can program by limiting
5200which event select and unit mask combinations are permitted.
5201
5202The argument holds a list of filter events which will be allowed or denied.
5203
5204Filter events only control general purpose counters; fixed purpose counters
5205are controlled by the fixed_counter_bitmap.
5206
5207Valid values for 'flags'::
5208
5209``0``
5210
5211To use this mode, clear the 'flags' field.
5212
5213In this mode each event will contain an event select + unit mask.
5214
5215When the guest attempts to program the PMU the guest's event select +
5216unit mask is compared against the filter events to determine whether the
5217guest should have access.
5218
5219``KVM_PMU_EVENT_FLAG_MASKED_EVENTS``
5220:Capability: KVM_CAP_PMU_EVENT_MASKED_EVENTS
5221
5222In this mode each filter event will contain an event select, mask, match, and
5223exclude value.  To encode a masked event use::
5224
5225  KVM_PMU_ENCODE_MASKED_ENTRY()
5226
5227An encoded event will follow this layout::
5228
5229  Bits   Description
5230  ----   -----------
5231  7:0    event select (low bits)
5232  15:8   umask match
5233  31:16  unused
5234  35:32  event select (high bits)
5235  36:54  unused
5236  55     exclude bit
5237  63:56  umask mask
5238
5239When the guest attempts to program the PMU, these steps are followed in
5240determining if the guest should have access:
5241
5242 1. Match the event select from the guest against the filter events.
5243 2. If a match is found, match the guest's unit mask to the mask and match
5244    values of the included filter events.
5245    I.e. (unit mask & mask) == match && !exclude.
5246 3. If a match is found, match the guest's unit mask to the mask and match
5247    values of the excluded filter events.
5248    I.e. (unit mask & mask) == match && exclude.
5249 4.
5250   a. If an included match is found and an excluded match is not found, filter
5251      the event.
5252   b. For everything else, do not filter the event.
5253 5.
5254   a. If the event is filtered and it's an allow list, allow the guest to
5255      program the event.
5256   b. If the event is filtered and it's a deny list, do not allow the guest to
5257      program the event.
5258
5259When setting a new pmu event filter, -EINVAL will be returned if any of the
5260unused fields are set or if any of the high bits (35:32) in the event
5261select are set when called on Intel.
5262
5263Valid values for 'action'::
5264
5265  #define KVM_PMU_EVENT_ALLOW 0
5266  #define KVM_PMU_EVENT_DENY 1
5267
5268Via this API, KVM userspace can also control the behavior of the VM's fixed
5269counters (if any) by configuring the "action" and "fixed_counter_bitmap" fields.
5270
5271Specifically, KVM follows the following pseudo-code when determining whether to
5272allow the guest FixCtr[i] to count its pre-defined fixed event::
5273
5274  FixCtr[i]_is_allowed = (action == ALLOW) && (bitmap & BIT(i)) ||
5275    (action == DENY) && !(bitmap & BIT(i));
5276  FixCtr[i]_is_denied = !FixCtr[i]_is_allowed;
5277
5278KVM always consumes fixed_counter_bitmap, it's userspace's responsibility to
5279ensure fixed_counter_bitmap is set correctly, e.g. if userspace wants to define
5280a filter that only affects general purpose counters.
5281
5282Note, the "events" field also applies to fixed counters' hardcoded event_select
5283and unit_mask values.  "fixed_counter_bitmap" has higher priority than "events"
5284if there is a contradiction between the two.
5285
52864.121 KVM_PPC_SVM_OFF
5287---------------------
5288
5289:Capability: basic
5290:Architectures: powerpc
5291:Type: vm ioctl
5292:Parameters: none
5293:Returns: 0 on successful completion,
5294
5295Errors:
5296
5297  ======     ================================================================
5298  EINVAL     if ultravisor failed to terminate the secure guest
5299  ENOMEM     if hypervisor failed to allocate new radix page tables for guest
5300  ======     ================================================================
5301
5302This ioctl is used to turn off the secure mode of the guest or transition
5303the guest from secure mode to normal mode. This is invoked when the guest
5304is reset. This has no effect if called for a normal guest.
5305
5306This ioctl issues an ultravisor call to terminate the secure guest,
5307unpins the VPA pages and releases all the device pages that are used to
5308track the secure pages by hypervisor.
5309
53104.122 KVM_S390_NORMAL_RESET
5311---------------------------
5312
5313:Capability: KVM_CAP_S390_VCPU_RESETS
5314:Architectures: s390
5315:Type: vcpu ioctl
5316:Parameters: none
5317:Returns: 0
5318
5319This ioctl resets VCPU registers and control structures according to
5320the cpu reset definition in the POP (Principles Of Operation).
5321
53224.123 KVM_S390_INITIAL_RESET
5323----------------------------
5324
5325:Capability: basic
5326:Architectures: s390
5327:Type: vcpu ioctl
5328:Parameters: none
5329:Returns: 0
5330
5331This ioctl resets VCPU registers and control structures according to
5332the initial cpu reset definition in the POP. However, the cpu is not
5333put into ESA mode. This reset is a superset of the normal reset.
5334
53354.124 KVM_S390_CLEAR_RESET
5336--------------------------
5337
5338:Capability: KVM_CAP_S390_VCPU_RESETS
5339:Architectures: s390
5340:Type: vcpu ioctl
5341:Parameters: none
5342:Returns: 0
5343
5344This ioctl resets VCPU registers and control structures according to
5345the clear cpu reset definition in the POP. However, the cpu is not put
5346into ESA mode. This reset is a superset of the initial reset.
5347
5348
53494.125 KVM_S390_PV_COMMAND
5350-------------------------
5351
5352:Capability: KVM_CAP_S390_PROTECTED
5353:Architectures: s390
5354:Type: vm ioctl
5355:Parameters: struct kvm_pv_cmd
5356:Returns: 0 on success, < 0 on error
5357
5358::
5359
5360  struct kvm_pv_cmd {
5361	__u32 cmd;	/* Command to be executed */
5362	__u16 rc;	/* Ultravisor return code */
5363	__u16 rrc;	/* Ultravisor return reason code */
5364	__u64 data;	/* Data or address */
5365	__u32 flags;    /* flags for future extensions. Must be 0 for now */
5366	__u32 reserved[3];
5367  };
5368
5369**Ultravisor return codes**
5370The Ultravisor return (reason) codes are provided by the kernel if a
5371Ultravisor call has been executed to achieve the results expected by
5372the command. Therefore they are independent of the IOCTL return
5373code. If KVM changes `rc`, its value will always be greater than 0
5374hence setting it to 0 before issuing a PV command is advised to be
5375able to detect a change of `rc`.
5376
5377**cmd values:**
5378
5379KVM_PV_ENABLE
5380  Allocate memory and register the VM with the Ultravisor, thereby
5381  donating memory to the Ultravisor that will become inaccessible to
5382  KVM. All existing CPUs are converted to protected ones. After this
5383  command has succeeded, any CPU added via hotplug will become
5384  protected during its creation as well.
5385
5386  Errors:
5387
5388  =====      =============================
5389  EINTR      an unmasked signal is pending
5390  =====      =============================
5391
5392KVM_PV_DISABLE
5393  Deregister the VM from the Ultravisor and reclaim the memory that had
5394  been donated to the Ultravisor, making it usable by the kernel again.
5395  All registered VCPUs are converted back to non-protected ones. If a
5396  previous protected VM had been prepared for asynchronous teardown with
5397  KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with
5398  KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call
5399  together with the current protected VM.
5400
5401KVM_PV_VM_SET_SEC_PARMS
5402  Pass the image header from VM memory to the Ultravisor in
5403  preparation of image unpacking and verification.
5404
5405KVM_PV_VM_UNPACK
5406  Unpack (protect and decrypt) a page of the encrypted boot image.
5407
5408KVM_PV_VM_VERIFY
5409  Verify the integrity of the unpacked image. Only if this succeeds,
5410  KVM is allowed to start protected VCPUs.
5411
5412KVM_PV_INFO
5413  :Capability: KVM_CAP_S390_PROTECTED_DUMP
5414
5415  Presents an API that provides Ultravisor related data to userspace
5416  via subcommands. len_max is the size of the user space buffer,
5417  len_written is KVM's indication of how much bytes of that buffer
5418  were actually written to. len_written can be used to determine the
5419  valid fields if more response fields are added in the future.
5420
5421  ::
5422
5423     enum pv_cmd_info_id {
5424	KVM_PV_INFO_VM,
5425	KVM_PV_INFO_DUMP,
5426     };
5427
5428     struct kvm_s390_pv_info_header {
5429	__u32 id;
5430	__u32 len_max;
5431	__u32 len_written;
5432	__u32 reserved;
5433     };
5434
5435     struct kvm_s390_pv_info {
5436	struct kvm_s390_pv_info_header header;
5437	struct kvm_s390_pv_info_dump dump;
5438	struct kvm_s390_pv_info_vm vm;
5439     };
5440
5441**subcommands:**
5442
5443  KVM_PV_INFO_VM
5444    This subcommand provides basic Ultravisor information for PV
5445    hosts. These values are likely also exported as files in the sysfs
5446    firmware UV query interface but they are more easily available to
5447    programs in this API.
5448
5449    The installed calls and feature_indication members provide the
5450    installed UV calls and the UV's other feature indications.
5451
5452    The max_* members provide information about the maximum number of PV
5453    vcpus, PV guests and PV guest memory size.
5454
5455    ::
5456
5457      struct kvm_s390_pv_info_vm {
5458	__u64 inst_calls_list[4];
5459	__u64 max_cpus;
5460	__u64 max_guests;
5461	__u64 max_guest_addr;
5462	__u64 feature_indication;
5463      };
5464
5465
5466  KVM_PV_INFO_DUMP
5467    This subcommand provides information related to dumping PV guests.
5468
5469    ::
5470
5471      struct kvm_s390_pv_info_dump {
5472	__u64 dump_cpu_buffer_len;
5473	__u64 dump_config_mem_buffer_per_1m;
5474	__u64 dump_config_finalize_len;
5475      };
5476
5477KVM_PV_DUMP
5478  :Capability: KVM_CAP_S390_PROTECTED_DUMP
5479
5480  Presents an API that provides calls which facilitate dumping a
5481  protected VM.
5482
5483  ::
5484
5485    struct kvm_s390_pv_dmp {
5486      __u64 subcmd;
5487      __u64 buff_addr;
5488      __u64 buff_len;
5489      __u64 gaddr;		/* For dump storage state */
5490    };
5491
5492  **subcommands:**
5493
5494  KVM_PV_DUMP_INIT
5495    Initializes the dump process of a protected VM. If this call does
5496    not succeed all other subcommands will fail with -EINVAL. This
5497    subcommand will return -EINVAL if a dump process has not yet been
5498    completed.
5499
5500    Not all PV vms can be dumped, the owner needs to set `dump
5501    allowed` PCF bit 34 in the SE header to allow dumping.
5502
5503  KVM_PV_DUMP_CONFIG_STOR_STATE
5504     Stores `buff_len` bytes of tweak component values starting with
5505     the 1MB block specified by the absolute guest address
5506     (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len`
5507     aligned and at least >= the `conf_dump_storage_state_len` value
5508     provided by the dump uv_info data. buff_user might be written to
5509     even if an error rc is returned. For instance if we encounter a
5510     fault after writing the first page of data.
5511
5512  KVM_PV_DUMP_COMPLETE
5513    If the subcommand succeeds it completes the dump process and lets
5514    KVM_PV_DUMP_INIT be called again.
5515
5516    On success `conf_dump_finalize_len` bytes of completion data will be
5517    stored to the `buff_addr`. The completion data contains a key
5518    derivation seed, IV, tweak nonce and encryption keys as well as an
5519    authentication tag all of which are needed to decrypt the dump at a
5520    later time.
5521
5522KVM_PV_ASYNC_CLEANUP_PREPARE
5523  :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE
5524
5525  Prepare the current protected VM for asynchronous teardown. Most
5526  resources used by the current protected VM will be set aside for a
5527  subsequent asynchronous teardown. The current protected VM will then
5528  resume execution immediately as non-protected. There can be at most
5529  one protected VM prepared for asynchronous teardown at any time. If
5530  a protected VM had already been prepared for teardown without
5531  subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will
5532  fail. In that case, the userspace process should issue a normal
5533  KVM_PV_DISABLE. The resources set aside with this call will need to
5534  be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM
5535  or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM
5536  terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon
5537  as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes.
5538
5539KVM_PV_ASYNC_CLEANUP_PERFORM
5540  :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE
5541
5542  Tear down the protected VM previously prepared for teardown with
5543  KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside
5544  will be freed during the execution of this command. This PV command
5545  should ideally be issued by userspace from a separate thread. If a
5546  fatal signal is received (or the process terminates naturally), the
5547  command will terminate immediately without completing, and the normal
5548  KVM shutdown procedure will take care of cleaning up all remaining
5549  protected VMs, including the ones whose teardown was interrupted by
5550  process termination.
5551
55524.126 KVM_XEN_HVM_SET_ATTR
5553--------------------------
5554
5555:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5556:Architectures: x86
5557:Type: vm ioctl
5558:Parameters: struct kvm_xen_hvm_attr
5559:Returns: 0 on success, < 0 on error
5560
5561::
5562
5563  struct kvm_xen_hvm_attr {
5564	__u16 type;
5565	__u16 pad[3];
5566	union {
5567		__u8 long_mode;
5568		__u8 vector;
5569		__u8 runstate_update_flag;
5570		union {
5571			__u64 gfn;
5572			__u64 hva;
5573		} shared_info;
5574		struct {
5575			__u32 send_port;
5576			__u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */
5577			__u32 flags;
5578			union {
5579				struct {
5580					__u32 port;
5581					__u32 vcpu;
5582					__u32 priority;
5583				} port;
5584				struct {
5585					__u32 port; /* Zero for eventfd */
5586					__s32 fd;
5587				} eventfd;
5588				__u32 padding[4];
5589			} deliver;
5590		} evtchn;
5591		__u32 xen_version;
5592		__u64 pad[8];
5593	} u;
5594  };
5595
5596type values:
5597
5598KVM_XEN_ATTR_TYPE_LONG_MODE
5599  Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
5600  determines the layout of the shared_info page exposed to the VM.
5601
5602KVM_XEN_ATTR_TYPE_SHARED_INFO
5603  Sets the guest physical frame number at which the Xen shared_info
5604  page resides. Note that although Xen places vcpu_info for the first
5605  32 vCPUs in the shared_info page, KVM does not automatically do so
5606  and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO or
5607  KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA be used explicitly even when
5608  the vcpu_info for a given vCPU resides at the "default" location
5609  in the shared_info page. This is because KVM may not be aware of
5610  the Xen CPU id which is used as the index into the vcpu_info[]
5611  array, so may know the correct default location.
5612
5613  Note that the shared_info page may be constantly written to by KVM;
5614  it contains the event channel bitmap used to deliver interrupts to
5615  a Xen guest, amongst other things. It is exempt from dirty tracking
5616  mechanisms — KVM will not explicitly mark the page as dirty each
5617  time an event channel interrupt is delivered to the guest! Thus,
5618  userspace should always assume that the designated GFN is dirty if
5619  any vCPU has been running or any event channel interrupts can be
5620  routed to the guest.
5621
5622  Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared_info
5623  page.
5624
5625KVM_XEN_ATTR_TYPE_SHARED_INFO_HVA
5626  If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the
5627  Xen capabilities, then this attribute may be used to set the
5628  userspace address at which the shared_info page resides, which
5629  will always be fixed in the VMM regardless of where it is mapped
5630  in guest physical address space. This attribute should be used in
5631  preference to KVM_XEN_ATTR_TYPE_SHARED_INFO as it avoids
5632  unnecessary invalidation of an internal cache when the page is
5633  re-mapped in guest physical address space.
5634
5635  Setting the hva to zero will disable the shared_info page.
5636
5637KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
5638  Sets the exception vector used to deliver Xen event channel upcalls.
5639  This is the HVM-wide vector injected directly by the hypervisor
5640  (not through the local APIC), typically configured by a guest via
5641  HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest
5642  SHUTDOWN_soft_reset) by setting it to zero.
5643
5644KVM_XEN_ATTR_TYPE_EVTCHN
5645  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5646  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5647  an outbound port number for interception of EVTCHNOP_send requests
5648  from the guest. A given sending port number may be directed back to
5649  a specified vCPU (by APIC ID) / port / priority on the guest, or to
5650  trigger events on an eventfd. The vCPU and priority can be changed
5651  by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but other
5652  fields cannot change for a given sending port. A port mapping is
5653  removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing
5654  KVM_XEN_EVTCHN_RESET in the flags field removes all interception of
5655  outbound event channels. The values of the flags field are mutually
5656  exclusive and cannot be combined as a bitmask.
5657
5658KVM_XEN_ATTR_TYPE_XEN_VERSION
5659  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5660  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5661  the 32-bit version code returned to the guest when it invokes the
5662  XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV
5663  Xen guests will often use this to as a dummy hypercall to trigger
5664  event channel delivery, so responding within the kernel without
5665  exiting to userspace is beneficial.
5666
5667KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG
5668  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5669  support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the
5670  XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read
5671  other vCPUs' vcpu_runstate_info. Xen guests enable this feature via
5672  the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist
5673  hypercall.
5674
56754.127 KVM_XEN_HVM_GET_ATTR
5676--------------------------
5677
5678:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5679:Architectures: x86
5680:Type: vm ioctl
5681:Parameters: struct kvm_xen_hvm_attr
5682:Returns: 0 on success, < 0 on error
5683
5684Allows Xen VM attributes to be read. For the structure and types,
5685see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN
5686attribute cannot be read.
5687
56884.128 KVM_XEN_VCPU_SET_ATTR
5689---------------------------
5690
5691:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5692:Architectures: x86
5693:Type: vcpu ioctl
5694:Parameters: struct kvm_xen_vcpu_attr
5695:Returns: 0 on success, < 0 on error
5696
5697::
5698
5699  struct kvm_xen_vcpu_attr {
5700	__u16 type;
5701	__u16 pad[3];
5702	union {
5703		__u64 gpa;
5704		__u64 pad[4];
5705		struct {
5706			__u64 state;
5707			__u64 state_entry_time;
5708			__u64 time_running;
5709			__u64 time_runnable;
5710			__u64 time_blocked;
5711			__u64 time_offline;
5712		} runstate;
5713		__u32 vcpu_id;
5714		struct {
5715			__u32 port;
5716			__u32 priority;
5717			__u64 expires_ns;
5718		} timer;
5719		__u8 vector;
5720	} u;
5721  };
5722
5723type values:
5724
5725KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
5726  Sets the guest physical address of the vcpu_info for a given vCPU.
5727  As with the shared_info page for the VM, the corresponding page may be
5728  dirtied at any time if event channel interrupt delivery is enabled, so
5729  userspace should always assume that the page is dirty without relying
5730  on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable
5731  the vcpu_info.
5732
5733KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO_HVA
5734  If the KVM_XEN_HVM_CONFIG_SHARED_INFO_HVA flag is also set in the
5735  Xen capabilities, then this attribute may be used to set the
5736  userspace address of the vcpu_info for a given vCPU. It should
5737  only be used when the vcpu_info resides at the "default" location
5738  in the shared_info page. In this case it is safe to assume the
5739  userspace address will not change, because the shared_info page is
5740  an overlay on guest memory and remains at a fixed host address
5741  regardless of where it is mapped in guest physical address space
5742  and hence unnecessary invalidation of an internal cache may be
5743  avoided if the guest memory layout is modified.
5744  If the vcpu_info does not reside at the "default" location then
5745  it is not guaranteed to remain at the same host address and
5746  hence the aforementioned cache invalidation is required.
5747
5748KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
5749  Sets the guest physical address of an additional pvclock structure
5750  for a given vCPU. This is typically used for guest vsyscall support.
5751  Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure.
5752
5753KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5754  Sets the guest physical address of the vcpu_runstate_info for a given
5755  vCPU. This is how a Xen guest tracks CPU state such as steal time.
5756  Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area.
5757
5758KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5759  Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5760  the given vCPU from the .u.runstate.state member of the structure.
5761  KVM automatically accounts running and runnable time but blocked
5762  and offline states are only entered explicitly.
5763
5764KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5765  Sets all fields of the vCPU runstate data from the .u.runstate member
5766  of the structure, including the current runstate. The state_entry_time
5767  must equal the sum of the other four times.
5768
5769KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5770  This *adds* the contents of the .u.runstate members of the structure
5771  to the corresponding members of the given vCPU's runstate data, thus
5772  permitting atomic adjustments to the runstate times. The adjustment
5773  to the state_entry_time must equal the sum of the adjustments to the
5774  other four times. The state field must be set to -1, or to a valid
5775  runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5776  or RUNSTATE_offline) to set the current accounted state as of the
5777  adjusted state_entry_time.
5778
5779KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID
5780  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5781  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen
5782  vCPU ID of the given vCPU, to allow timer-related VCPU operations to
5783  be intercepted by KVM.
5784
5785KVM_XEN_VCPU_ATTR_TYPE_TIMER
5786  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5787  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5788  event channel port/priority for the VIRQ_TIMER of the vCPU, as well
5789  as allowing a pending timer to be saved/restored. Setting the timer
5790  port to zero disables kernel handling of the singleshot timer.
5791
5792KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR
5793  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5794  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5795  per-vCPU local APIC upcall vector, configured by a Xen guest with
5796  the HVMOP_set_evtchn_upcall_vector hypercall. This is typically
5797  used by Windows guests, and is distinct from the HVM-wide upcall
5798  vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by
5799  setting the vector to zero.
5800
5801
58024.129 KVM_XEN_VCPU_GET_ATTR
5803---------------------------
5804
5805:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5806:Architectures: x86
5807:Type: vcpu ioctl
5808:Parameters: struct kvm_xen_vcpu_attr
5809:Returns: 0 on success, < 0 on error
5810
5811Allows Xen vCPU attributes to be read. For the structure and types,
5812see KVM_XEN_VCPU_SET_ATTR above.
5813
5814The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5815with the KVM_XEN_VCPU_GET_ATTR ioctl.
5816
58174.130 KVM_ARM_MTE_COPY_TAGS
5818---------------------------
5819
5820:Capability: KVM_CAP_ARM_MTE
5821:Architectures: arm64
5822:Type: vm ioctl
5823:Parameters: struct kvm_arm_copy_mte_tags
5824:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5825          arguments, -EFAULT if memory cannot be accessed).
5826
5827::
5828
5829  struct kvm_arm_copy_mte_tags {
5830	__u64 guest_ipa;
5831	__u64 length;
5832	void __user *addr;
5833	__u64 flags;
5834	__u64 reserved[2];
5835  };
5836
5837Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5838``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned.
5839``length`` must not be bigger than 2^31 - PAGE_SIZE bytes. The ``addr``
5840field must point to a buffer which the tags will be copied to or from.
5841
5842``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5843``KVM_ARM_TAGS_FROM_GUEST``.
5844
5845The size of the buffer to store the tags is ``(length / 16)`` bytes
5846(granules in MTE are 16 bytes long). Each byte contains a single tag
5847value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5848``PTRACE_POKEMTETAGS``.
5849
5850If an error occurs before any data is copied then a negative error code is
5851returned. If some tags have been copied before an error occurs then the number
5852of bytes successfully copied is returned. If the call completes successfully
5853then ``length`` is returned.
5854
58554.131 KVM_GET_SREGS2
5856--------------------
5857
5858:Capability: KVM_CAP_SREGS2
5859:Architectures: x86
5860:Type: vcpu ioctl
5861:Parameters: struct kvm_sregs2 (out)
5862:Returns: 0 on success, -1 on error
5863
5864Reads special registers from the vcpu.
5865This ioctl (when supported) replaces the KVM_GET_SREGS.
5866
5867::
5868
5869        struct kvm_sregs2 {
5870                /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5871                struct kvm_segment cs, ds, es, fs, gs, ss;
5872                struct kvm_segment tr, ldt;
5873                struct kvm_dtable gdt, idt;
5874                __u64 cr0, cr2, cr3, cr4, cr8;
5875                __u64 efer;
5876                __u64 apic_base;
5877                __u64 flags;
5878                __u64 pdptrs[4];
5879        };
5880
5881flags values for ``kvm_sregs2``:
5882
5883``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5884
5885  Indicates that the struct contains valid PDPTR values.
5886
5887
58884.132 KVM_SET_SREGS2
5889--------------------
5890
5891:Capability: KVM_CAP_SREGS2
5892:Architectures: x86
5893:Type: vcpu ioctl
5894:Parameters: struct kvm_sregs2 (in)
5895:Returns: 0 on success, -1 on error
5896
5897Writes special registers into the vcpu.
5898See KVM_GET_SREGS2 for the data structures.
5899This ioctl (when supported) replaces the KVM_SET_SREGS.
5900
59014.133 KVM_GET_STATS_FD
5902----------------------
5903
5904:Capability: KVM_CAP_STATS_BINARY_FD
5905:Architectures: all
5906:Type: vm ioctl, vcpu ioctl
5907:Parameters: none
5908:Returns: statistics file descriptor on success, < 0 on error
5909
5910Errors:
5911
5912  ======     ======================================================
5913  ENOMEM     if the fd could not be created due to lack of memory
5914  EMFILE     if the number of opened files exceeds the limit
5915  ======     ======================================================
5916
5917The returned file descriptor can be used to read VM/vCPU statistics data in
5918binary format. The data in the file descriptor consists of four blocks
5919organized as follows:
5920
5921+-------------+
5922|   Header    |
5923+-------------+
5924|  id string  |
5925+-------------+
5926| Descriptors |
5927+-------------+
5928| Stats Data  |
5929+-------------+
5930
5931Apart from the header starting at offset 0, please be aware that it is
5932not guaranteed that the four blocks are adjacent or in the above order;
5933the offsets of the id, descriptors and data blocks are found in the
5934header.  However, all four blocks are aligned to 64 bit offsets in the
5935file and they do not overlap.
5936
5937All blocks except the data block are immutable.  Userspace can read them
5938only one time after retrieving the file descriptor, and then use ``pread`` or
5939``lseek`` to read the statistics repeatedly.
5940
5941All data is in system endianness.
5942
5943The format of the header is as follows::
5944
5945	struct kvm_stats_header {
5946		__u32 flags;
5947		__u32 name_size;
5948		__u32 num_desc;
5949		__u32 id_offset;
5950		__u32 desc_offset;
5951		__u32 data_offset;
5952	};
5953
5954The ``flags`` field is not used at the moment. It is always read as 0.
5955
5956The ``name_size`` field is the size (in byte) of the statistics name string
5957(including trailing '\0') which is contained in the "id string" block and
5958appended at the end of every descriptor.
5959
5960The ``num_desc`` field is the number of descriptors that are included in the
5961descriptor block.  (The actual number of values in the data block may be
5962larger, since each descriptor may comprise more than one value).
5963
5964The ``id_offset`` field is the offset of the id string from the start of the
5965file indicated by the file descriptor. It is a multiple of 8.
5966
5967The ``desc_offset`` field is the offset of the Descriptors block from the start
5968of the file indicated by the file descriptor. It is a multiple of 8.
5969
5970The ``data_offset`` field is the offset of the Stats Data block from the start
5971of the file indicated by the file descriptor. It is a multiple of 8.
5972
5973The id string block contains a string which identifies the file descriptor on
5974which KVM_GET_STATS_FD was invoked.  The size of the block, including the
5975trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5976
5977The descriptors block is only needed to be read once for the lifetime of the
5978file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5979by a string of size ``name_size``.
5980::
5981
5982	#define KVM_STATS_TYPE_SHIFT		0
5983	#define KVM_STATS_TYPE_MASK		(0xF << KVM_STATS_TYPE_SHIFT)
5984	#define KVM_STATS_TYPE_CUMULATIVE	(0x0 << KVM_STATS_TYPE_SHIFT)
5985	#define KVM_STATS_TYPE_INSTANT		(0x1 << KVM_STATS_TYPE_SHIFT)
5986	#define KVM_STATS_TYPE_PEAK		(0x2 << KVM_STATS_TYPE_SHIFT)
5987	#define KVM_STATS_TYPE_LINEAR_HIST	(0x3 << KVM_STATS_TYPE_SHIFT)
5988	#define KVM_STATS_TYPE_LOG_HIST		(0x4 << KVM_STATS_TYPE_SHIFT)
5989	#define KVM_STATS_TYPE_MAX		KVM_STATS_TYPE_LOG_HIST
5990
5991	#define KVM_STATS_UNIT_SHIFT		4
5992	#define KVM_STATS_UNIT_MASK		(0xF << KVM_STATS_UNIT_SHIFT)
5993	#define KVM_STATS_UNIT_NONE		(0x0 << KVM_STATS_UNIT_SHIFT)
5994	#define KVM_STATS_UNIT_BYTES		(0x1 << KVM_STATS_UNIT_SHIFT)
5995	#define KVM_STATS_UNIT_SECONDS		(0x2 << KVM_STATS_UNIT_SHIFT)
5996	#define KVM_STATS_UNIT_CYCLES		(0x3 << KVM_STATS_UNIT_SHIFT)
5997	#define KVM_STATS_UNIT_BOOLEAN		(0x4 << KVM_STATS_UNIT_SHIFT)
5998	#define KVM_STATS_UNIT_MAX		KVM_STATS_UNIT_BOOLEAN
5999
6000	#define KVM_STATS_BASE_SHIFT		8
6001	#define KVM_STATS_BASE_MASK		(0xF << KVM_STATS_BASE_SHIFT)
6002	#define KVM_STATS_BASE_POW10		(0x0 << KVM_STATS_BASE_SHIFT)
6003	#define KVM_STATS_BASE_POW2		(0x1 << KVM_STATS_BASE_SHIFT)
6004	#define KVM_STATS_BASE_MAX		KVM_STATS_BASE_POW2
6005
6006	struct kvm_stats_desc {
6007		__u32 flags;
6008		__s16 exponent;
6009		__u16 size;
6010		__u32 offset;
6011		__u32 bucket_size;
6012		char name[];
6013	};
6014
6015The ``flags`` field contains the type and unit of the statistics data described
6016by this descriptor. Its endianness is CPU native.
6017The following flags are supported:
6018
6019Bits 0-3 of ``flags`` encode the type:
6020
6021  * ``KVM_STATS_TYPE_CUMULATIVE``
6022    The statistics reports a cumulative count. The value of data can only be increased.
6023    Most of the counters used in KVM are of this type.
6024    The corresponding ``size`` field for this type is always 1.
6025    All cumulative statistics data are read/write.
6026  * ``KVM_STATS_TYPE_INSTANT``
6027    The statistics reports an instantaneous value. Its value can be increased or
6028    decreased. This type is usually used as a measurement of some resources,
6029    like the number of dirty pages, the number of large pages, etc.
6030    All instant statistics are read only.
6031    The corresponding ``size`` field for this type is always 1.
6032  * ``KVM_STATS_TYPE_PEAK``
6033    The statistics data reports a peak value, for example the maximum number
6034    of items in a hash table bucket, the longest time waited and so on.
6035    The value of data can only be increased.
6036    The corresponding ``size`` field for this type is always 1.
6037  * ``KVM_STATS_TYPE_LINEAR_HIST``
6038    The statistic is reported as a linear histogram. The number of
6039    buckets is specified by the ``size`` field. The size of buckets is specified
6040    by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``)
6041    is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last
6042    bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity
6043    value.)
6044  * ``KVM_STATS_TYPE_LOG_HIST``
6045    The statistic is reported as a logarithmic histogram. The number of
6046    buckets is specified by the ``size`` field. The range of the first bucket is
6047    [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF).
6048    Otherwise, The Nth bucket (1 < N < ``size``) covers
6049    [pow(2, N-2), pow(2, N-1)).
6050
6051Bits 4-7 of ``flags`` encode the unit:
6052
6053  * ``KVM_STATS_UNIT_NONE``
6054    There is no unit for the value of statistics data. This usually means that
6055    the value is a simple counter of an event.
6056  * ``KVM_STATS_UNIT_BYTES``
6057    It indicates that the statistics data is used to measure memory size, in the
6058    unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
6059    determined by the ``exponent`` field in the descriptor.
6060  * ``KVM_STATS_UNIT_SECONDS``
6061    It indicates that the statistics data is used to measure time or latency.
6062  * ``KVM_STATS_UNIT_CYCLES``
6063    It indicates that the statistics data is used to measure CPU clock cycles.
6064  * ``KVM_STATS_UNIT_BOOLEAN``
6065    It indicates that the statistic will always be either 0 or 1.  Boolean
6066    statistics of "peak" type will never go back from 1 to 0.  Boolean
6067    statistics can be linear histograms (with two buckets) but not logarithmic
6068    histograms.
6069
6070Note that, in the case of histograms, the unit applies to the bucket
6071ranges, while the bucket value indicates how many samples fell in the
6072bucket's range.
6073
6074Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
6075unit:
6076
6077  * ``KVM_STATS_BASE_POW10``
6078    The scale is based on power of 10. It is used for measurement of time and
6079    CPU clock cycles.  For example, an exponent of -9 can be used with
6080    ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
6081  * ``KVM_STATS_BASE_POW2``
6082    The scale is based on power of 2. It is used for measurement of memory size.
6083    For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
6084    express that the unit is MiB.
6085
6086The ``size`` field is the number of values of this statistics data. Its
6087value is usually 1 for most of simple statistics. 1 means it contains an
6088unsigned 64bit data.
6089
6090The ``offset`` field is the offset from the start of Data Block to the start of
6091the corresponding statistics data.
6092
6093The ``bucket_size`` field is used as a parameter for histogram statistics data.
6094It is only used by linear histogram statistics data, specifying the size of a
6095bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``.
6096
6097The ``name`` field is the name string of the statistics data. The name string
6098starts at the end of ``struct kvm_stats_desc``.  The maximum length including
6099the trailing ``'\0'``, is indicated by ``name_size`` in the header.
6100
6101The Stats Data block contains an array of 64-bit values in the same order
6102as the descriptors in Descriptors block.
6103
61044.134 KVM_GET_XSAVE2
6105--------------------
6106
6107:Capability: KVM_CAP_XSAVE2
6108:Architectures: x86
6109:Type: vcpu ioctl
6110:Parameters: struct kvm_xsave (out)
6111:Returns: 0 on success, -1 on error
6112
6113
6114::
6115
6116  struct kvm_xsave {
6117	__u32 region[1024];
6118	__u32 extra[0];
6119  };
6120
6121This ioctl would copy current vcpu's xsave struct to the userspace. It
6122copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2)
6123when invoked on the vm file descriptor. The size value returned by
6124KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
6125Currently, it is only greater than 4096 if a dynamic feature has been
6126enabled with ``arch_prctl()``, but this may change in the future.
6127
6128The offsets of the state save areas in struct kvm_xsave follow the contents
6129of CPUID leaf 0xD on the host.
6130
61314.135 KVM_XEN_HVM_EVTCHN_SEND
6132-----------------------------
6133
6134:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND
6135:Architectures: x86
6136:Type: vm ioctl
6137:Parameters: struct kvm_irq_routing_xen_evtchn
6138:Returns: 0 on success, < 0 on error
6139
6140
6141::
6142
6143   struct kvm_irq_routing_xen_evtchn {
6144	__u32 port;
6145	__u32 vcpu;
6146	__u32 priority;
6147   };
6148
6149This ioctl injects an event channel interrupt directly to the guest vCPU.
6150
61514.136 KVM_S390_PV_CPU_COMMAND
6152-----------------------------
6153
6154:Capability: KVM_CAP_S390_PROTECTED_DUMP
6155:Architectures: s390
6156:Type: vcpu ioctl
6157:Parameters: none
6158:Returns: 0 on success, < 0 on error
6159
6160This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests
6161for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares
6162the command ids.
6163
6164**command:**
6165
6166KVM_PV_DUMP
6167  Presents an API that provides calls which facilitate dumping a vcpu
6168  of a protected VM.
6169
6170**subcommand:**
6171
6172KVM_PV_DUMP_CPU
6173  Provides encrypted dump data like register values.
6174  The length of the returned data is provided by uv_info.guest_cpu_stor_len.
6175
61764.137 KVM_S390_ZPCI_OP
6177----------------------
6178
6179:Capability: KVM_CAP_S390_ZPCI_OP
6180:Architectures: s390
6181:Type: vm ioctl
6182:Parameters: struct kvm_s390_zpci_op (in)
6183:Returns: 0 on success, <0 on error
6184
6185Used to manage hardware-assisted virtualization features for zPCI devices.
6186
6187Parameters are specified via the following structure::
6188
6189  struct kvm_s390_zpci_op {
6190	/* in */
6191	__u32 fh;		/* target device */
6192	__u8  op;		/* operation to perform */
6193	__u8  pad[3];
6194	union {
6195		/* for KVM_S390_ZPCIOP_REG_AEN */
6196		struct {
6197			__u64 ibv;	/* Guest addr of interrupt bit vector */
6198			__u64 sb;	/* Guest addr of summary bit */
6199			__u32 flags;
6200			__u32 noi;	/* Number of interrupts */
6201			__u8 isc;	/* Guest interrupt subclass */
6202			__u8 sbo;	/* Offset of guest summary bit vector */
6203			__u16 pad;
6204		} reg_aen;
6205		__u64 reserved[8];
6206	} u;
6207  };
6208
6209The type of operation is specified in the "op" field.
6210KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event
6211notification interpretation, which will allow firmware delivery of adapter
6212events directly to the vm, with KVM providing a backup delivery mechanism;
6213KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of
6214adapter event notifications.
6215
6216The target zPCI function must also be specified via the "fh" field.  For the
6217KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware
6218delivery must be provided via the "reg_aen" struct.
6219
6220The "pad" and "reserved" fields may be used for future extensions and should be
6221set to 0s by userspace.
6222
62234.138 KVM_ARM_SET_COUNTER_OFFSET
6224--------------------------------
6225
6226:Capability: KVM_CAP_COUNTER_OFFSET
6227:Architectures: arm64
6228:Type: vm ioctl
6229:Parameters: struct kvm_arm_counter_offset (in)
6230:Returns: 0 on success, < 0 on error
6231
6232This capability indicates that userspace is able to apply a single VM-wide
6233offset to both the virtual and physical counters as viewed by the guest
6234using the KVM_ARM_SET_CNT_OFFSET ioctl and the following data structure:
6235
6236::
6237
6238	struct kvm_arm_counter_offset {
6239		__u64 counter_offset;
6240		__u64 reserved;
6241	};
6242
6243The offset describes a number of counter cycles that are subtracted from
6244both virtual and physical counter views (similar to the effects of the
6245CNTVOFF_EL2 and CNTPOFF_EL2 system registers, but only global). The offset
6246always applies to all vcpus (already created or created after this ioctl)
6247for this VM.
6248
6249It is userspace's responsibility to compute the offset based, for example,
6250on previous values of the guest counters.
6251
6252Any value other than 0 for the "reserved" field may result in an error
6253(-EINVAL) being returned. This ioctl can also return -EBUSY if any vcpu
6254ioctl is issued concurrently.
6255
6256Note that using this ioctl results in KVM ignoring subsequent userspace
6257writes to the CNTVCT_EL0 and CNTPCT_EL0 registers using the SET_ONE_REG
6258interface. No error will be returned, but the resulting offset will not be
6259applied.
6260
6261.. _KVM_ARM_GET_REG_WRITABLE_MASKS:
6262
62634.139 KVM_ARM_GET_REG_WRITABLE_MASKS
6264------------------------------------
6265
6266:Capability: KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES
6267:Architectures: arm64
6268:Type: vm ioctl
6269:Parameters: struct reg_mask_range (in/out)
6270:Returns: 0 on success, < 0 on error
6271
6272
6273::
6274
6275        #define KVM_ARM_FEATURE_ID_RANGE	0
6276        #define KVM_ARM_FEATURE_ID_RANGE_SIZE	(3 * 8 * 8)
6277
6278        struct reg_mask_range {
6279                __u64 addr;             /* Pointer to mask array */
6280                __u32 range;            /* Requested range */
6281                __u32 reserved[13];
6282        };
6283
6284This ioctl copies the writable masks for a selected range of registers to
6285userspace.
6286
6287The ``addr`` field is a pointer to the destination array where KVM copies
6288the writable masks.
6289
6290The ``range`` field indicates the requested range of registers.
6291``KVM_CHECK_EXTENSION`` for the ``KVM_CAP_ARM_SUPPORTED_REG_MASK_RANGES``
6292capability returns the supported ranges, expressed as a set of flags. Each
6293flag's bit index represents a possible value for the ``range`` field.
6294All other values are reserved for future use and KVM may return an error.
6295
6296The ``reserved[13]`` array is reserved for future use and should be 0, or
6297KVM may return an error.
6298
6299KVM_ARM_FEATURE_ID_RANGE (0)
6300^^^^^^^^^^^^^^^^^^^^^^^^^^^^
6301
6302The Feature ID range is defined as the AArch64 System register space with
6303op0==3, op1=={0, 1, 3}, CRn==0, CRm=={0-7}, op2=={0-7}.
6304
6305The mask returned array pointed to by ``addr`` is indexed by the macro
6306``ARM64_FEATURE_ID_RANGE_IDX(op0, op1, crn, crm, op2)``, allowing userspace
6307to know what fields can be changed for the system register described by
6308``op0, op1, crn, crm, op2``. KVM rejects ID register values that describe a
6309superset of the features supported by the system.
6310
63114.140 KVM_SET_USER_MEMORY_REGION2
6312---------------------------------
6313
6314:Capability: KVM_CAP_USER_MEMORY2
6315:Architectures: all
6316:Type: vm ioctl
6317:Parameters: struct kvm_userspace_memory_region2 (in)
6318:Returns: 0 on success, -1 on error
6319
6320KVM_SET_USER_MEMORY_REGION2 is an extension to KVM_SET_USER_MEMORY_REGION that
6321allows mapping guest_memfd memory into a guest.  All fields shared with
6322KVM_SET_USER_MEMORY_REGION identically.  Userspace can set KVM_MEM_GUEST_MEMFD
6323in flags to have KVM bind the memory region to a given guest_memfd range of
6324[guest_memfd_offset, guest_memfd_offset + memory_size].  The target guest_memfd
6325must point at a file created via KVM_CREATE_GUEST_MEMFD on the current VM, and
6326the target range must not be bound to any other memory region.  All standard
6327bounds checks apply (use common sense).
6328
6329::
6330
6331  struct kvm_userspace_memory_region2 {
6332	__u32 slot;
6333	__u32 flags;
6334	__u64 guest_phys_addr;
6335	__u64 memory_size; /* bytes */
6336	__u64 userspace_addr; /* start of the userspace allocated memory */
6337	__u64 guest_memfd_offset;
6338	__u32 guest_memfd;
6339	__u32 pad1;
6340	__u64 pad2[14];
6341  };
6342
6343A KVM_MEM_GUEST_MEMFD region _must_ have a valid guest_memfd (private memory) and
6344userspace_addr (shared memory).  However, "valid" for userspace_addr simply
6345means that the address itself must be a legal userspace address.  The backing
6346mapping for userspace_addr is not required to be valid/populated at the time of
6347KVM_SET_USER_MEMORY_REGION2, e.g. shared memory can be lazily mapped/allocated
6348on-demand.
6349
6350When mapping a gfn into the guest, KVM selects shared vs. private, i.e consumes
6351userspace_addr vs. guest_memfd, based on the gfn's KVM_MEMORY_ATTRIBUTE_PRIVATE
6352state.  At VM creation time, all memory is shared, i.e. the PRIVATE attribute
6353is '0' for all gfns.  Userspace can control whether memory is shared/private by
6354toggling KVM_MEMORY_ATTRIBUTE_PRIVATE via KVM_SET_MEMORY_ATTRIBUTES as needed.
6355
6356S390:
6357^^^^^
6358
6359Returns -EINVAL if the VM has the KVM_VM_S390_UCONTROL flag set.
6360Returns -EINVAL if called on a protected VM.
6361
63624.141 KVM_SET_MEMORY_ATTRIBUTES
6363-------------------------------
6364
6365:Capability: KVM_CAP_MEMORY_ATTRIBUTES
6366:Architectures: x86
6367:Type: vm ioctl
6368:Parameters: struct kvm_memory_attributes (in)
6369:Returns: 0 on success, <0 on error
6370
6371KVM_SET_MEMORY_ATTRIBUTES allows userspace to set memory attributes for a range
6372of guest physical memory.
6373
6374::
6375
6376  struct kvm_memory_attributes {
6377	__u64 address;
6378	__u64 size;
6379	__u64 attributes;
6380	__u64 flags;
6381  };
6382
6383  #define KVM_MEMORY_ATTRIBUTE_PRIVATE           (1ULL << 3)
6384
6385The address and size must be page aligned.  The supported attributes can be
6386retrieved via ioctl(KVM_CHECK_EXTENSION) on KVM_CAP_MEMORY_ATTRIBUTES.  If
6387executed on a VM, KVM_CAP_MEMORY_ATTRIBUTES precisely returns the attributes
6388supported by that VM.  If executed at system scope, KVM_CAP_MEMORY_ATTRIBUTES
6389returns all attributes supported by KVM.  The only attribute defined at this
6390time is KVM_MEMORY_ATTRIBUTE_PRIVATE, which marks the associated gfn as being
6391guest private memory.
6392
6393Note, there is no "get" API.  Userspace is responsible for explicitly tracking
6394the state of a gfn/page as needed.
6395
6396The "flags" field is reserved for future extensions and must be '0'.
6397
63984.142 KVM_CREATE_GUEST_MEMFD
6399----------------------------
6400
6401:Capability: KVM_CAP_GUEST_MEMFD
6402:Architectures: none
6403:Type: vm ioctl
6404:Parameters: struct kvm_create_guest_memfd(in)
6405:Returns: A file descriptor on success, <0 on error
6406
6407KVM_CREATE_GUEST_MEMFD creates an anonymous file and returns a file descriptor
6408that refers to it.  guest_memfd files are roughly analogous to files created
6409via memfd_create(), e.g. guest_memfd files live in RAM, have volatile storage,
6410and are automatically released when the last reference is dropped.  Unlike
6411"regular" memfd_create() files, guest_memfd files are bound to their owning
6412virtual machine (see below), cannot be mapped, read, or written by userspace,
6413and cannot be resized  (guest_memfd files do however support PUNCH_HOLE).
6414
6415::
6416
6417  struct kvm_create_guest_memfd {
6418	__u64 size;
6419	__u64 flags;
6420	__u64 reserved[6];
6421  };
6422
6423Conceptually, the inode backing a guest_memfd file represents physical memory,
6424i.e. is coupled to the virtual machine as a thing, not to a "struct kvm".  The
6425file itself, which is bound to a "struct kvm", is that instance's view of the
6426underlying memory, e.g. effectively provides the translation of guest addresses
6427to host memory.  This allows for use cases where multiple KVM structures are
6428used to manage a single virtual machine, e.g. when performing intrahost
6429migration of a virtual machine.
6430
6431KVM currently only supports mapping guest_memfd via KVM_SET_USER_MEMORY_REGION2,
6432and more specifically via the guest_memfd and guest_memfd_offset fields in
6433"struct kvm_userspace_memory_region2", where guest_memfd_offset is the offset
6434into the guest_memfd instance.  For a given guest_memfd file, there can be at
6435most one mapping per page, i.e. binding multiple memory regions to a single
6436guest_memfd range is not allowed (any number of memory regions can be bound to
6437a single guest_memfd file, but the bound ranges must not overlap).
6438
6439The capability KVM_CAP_GUEST_MEMFD_FLAGS enumerates the `flags` that can be
6440specified via KVM_CREATE_GUEST_MEMFD.  Currently defined flags:
6441
6442  ============================ ================================================
6443  GUEST_MEMFD_FLAG_MMAP        Enable using mmap() on the guest_memfd file
6444                               descriptor.
6445  GUEST_MEMFD_FLAG_INIT_SHARED Make all memory in the file shared during
6446                               KVM_CREATE_GUEST_MEMFD (memory files created
6447                               without INIT_SHARED will be marked private).
6448                               Shared memory can be faulted into host userspace
6449                               page tables. Private memory cannot.
6450  ============================ ================================================
6451
6452When the KVM MMU performs a PFN lookup to service a guest fault and the backing
6453guest_memfd has the GUEST_MEMFD_FLAG_MMAP set, then the fault will always be
6454consumed from guest_memfd, regardless of whether it is a shared or a private
6455fault.
6456
6457See KVM_SET_USER_MEMORY_REGION2 for additional details.
6458
64594.143 KVM_PRE_FAULT_MEMORY
6460---------------------------
6461
6462:Capability: KVM_CAP_PRE_FAULT_MEMORY
6463:Architectures: none
6464:Type: vcpu ioctl
6465:Parameters: struct kvm_pre_fault_memory (in/out)
6466:Returns: 0 if at least one page is processed, < 0 on error
6467
6468Errors:
6469
6470  ========== ===============================================================
6471  EINVAL     The specified `gpa` and `size` were invalid (e.g. not
6472             page aligned, causes an overflow, or size is zero).
6473  ENOENT     The specified `gpa` is outside defined memslots.
6474  EINTR      An unmasked signal is pending and no page was processed.
6475  EFAULT     The parameter address was invalid.
6476  EOPNOTSUPP Mapping memory for a GPA is unsupported by the
6477             hypervisor, and/or for the current vCPU state/mode.
6478  EIO        unexpected error conditions (also causes a WARN)
6479  ========== ===============================================================
6480
6481::
6482
6483  struct kvm_pre_fault_memory {
6484	/* in/out */
6485	__u64 gpa;
6486	__u64 size;
6487	/* in */
6488	__u64 flags;
6489	__u64 padding[5];
6490  };
6491
6492KVM_PRE_FAULT_MEMORY populates KVM's stage-2 page tables used to map memory
6493for the current vCPU state.  KVM maps memory as if the vCPU generated a
6494stage-2 read page fault, e.g. faults in memory as needed, but doesn't break
6495CoW.  However, KVM does not mark any newly created stage-2 PTE as Accessed.
6496
6497In the case of confidential VM types where there is an initial set up of
6498private guest memory before the guest is 'finalized'/measured, this ioctl
6499should only be issued after completing all the necessary setup to put the
6500guest into a 'finalized' state so that the above semantics can be reliably
6501ensured.
6502
6503In some cases, multiple vCPUs might share the page tables.  In this
6504case, the ioctl can be called in parallel.
6505
6506When the ioctl returns, the input values are updated to point to the
6507remaining range.  If `size` > 0 on return, the caller can just issue
6508the ioctl again with the same `struct kvm_map_memory` argument.
6509
6510Shadow page tables cannot support this ioctl because they
6511are indexed by virtual address or nested guest physical address.
6512Calling this ioctl when the guest is using shadow page tables (for
6513example because it is running a nested guest with nested page tables)
6514will fail with `EOPNOTSUPP` even if `KVM_CHECK_EXTENSION` reports
6515the capability to be present.
6516
6517`flags` must currently be zero.
6518
65194.144 KVM_S390_KEYOP
6520--------------------
6521
6522:Capability: KVM_CAP_S390_KEYOP
6523:Architectures: s390
6524:Type: vm ioctl
6525:Parameters: struct kvm_s390_keyop (in/out)
6526:Returns: 0 in case of success, < 0 on error
6527
6528The specified key operation is performed on the given guest address. The
6529previous storage key (or the relevant part thereof) will be returned in
6530`key`.
6531
6532::
6533
6534  struct kvm_s390_keyop {
6535	__u64 guest_addr;
6536	__u8  key;
6537	__u8  operation;
6538  };
6539
6540Currently supported values for ``operation``:
6541
6542KVM_S390_KEYOP_ISKE
6543  Returns the storage key for the guest address ``guest_addr`` in ``key``.
6544
6545KVM_S390_KEYOP_RRBE
6546  Resets the reference bit for the guest address ``guest_addr``, returning the
6547  R and C bits of the old storage key in ``key``; the remaining fields of
6548  the storage key will be set to 0.
6549
6550KVM_S390_KEYOP_SSKE
6551  Sets the storage key for the guest address ``guest_addr`` to the key
6552  specified in ``key``, returning the previous value in ``key``.
6553
6554.. _kvm_run:
6555
65565. The kvm_run structure
6557========================
6558
6559Application code obtains a pointer to the kvm_run structure by
6560mmap()ing a vcpu fd.  From that point, application code can control
6561execution by changing fields in kvm_run prior to calling the KVM_RUN
6562ioctl, and obtain information about the reason KVM_RUN returned by
6563looking up structure members.
6564
6565::
6566
6567  struct kvm_run {
6568	/* in */
6569	__u8 request_interrupt_window;
6570
6571Request that KVM_RUN return when it becomes possible to inject external
6572interrupts into the guest.  Useful in conjunction with KVM_INTERRUPT.
6573
6574::
6575
6576	__u8 immediate_exit;
6577
6578This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
6579exits immediately, returning -EINTR.  In the common scenario where a
6580signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
6581to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
6582Rather than blocking the signal outside KVM_RUN, userspace can set up
6583a signal handler that sets run->immediate_exit to a non-zero value.
6584
6585This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
6586
6587::
6588
6589	__u8 padding1[6];
6590
6591	/* out */
6592	__u32 exit_reason;
6593
6594When KVM_RUN has returned successfully (return value 0), this informs
6595application code why KVM_RUN has returned.  Allowable values for this
6596field are detailed below.
6597
6598::
6599
6600	__u8 ready_for_interrupt_injection;
6601
6602If request_interrupt_window has been specified, this field indicates
6603an interrupt can be injected now with KVM_INTERRUPT.
6604
6605::
6606
6607	__u8 if_flag;
6608
6609The value of the current interrupt flag.  Only valid if in-kernel
6610local APIC is not used.
6611
6612::
6613
6614	__u16 flags;
6615
6616More architecture-specific flags detailing state of the VCPU that may
6617affect the device's behavior. Current defined flags::
6618
6619  /* x86, set if the VCPU is in system management mode */
6620  #define KVM_RUN_X86_SMM          (1 << 0)
6621  /* x86, set if bus lock detected in VM */
6622  #define KVM_RUN_X86_BUS_LOCK     (1 << 1)
6623  /* x86, set if the VCPU is executing a nested (L2) guest */
6624  #define KVM_RUN_X86_GUEST_MODE   (1 << 2)
6625
6626  /* arm64, set for KVM_EXIT_DEBUG */
6627  #define KVM_DEBUG_ARCH_HSR_HIGH_VALID  (1 << 0)
6628
6629::
6630
6631	/* in (pre_kvm_run), out (post_kvm_run) */
6632	__u64 cr8;
6633
6634The value of the cr8 register.  Only valid if in-kernel local APIC is
6635not used.  Both input and output.
6636
6637::
6638
6639	__u64 apic_base;
6640
6641The value of the APIC BASE msr.  Only valid if in-kernel local
6642APIC is not used.  Both input and output.
6643
6644::
6645
6646	union {
6647		/* KVM_EXIT_UNKNOWN */
6648		struct {
6649			__u64 hardware_exit_reason;
6650		} hw;
6651
6652If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
6653reasons.  Further architecture-specific information is available in
6654hardware_exit_reason.
6655
6656::
6657
6658		/* KVM_EXIT_FAIL_ENTRY */
6659		struct {
6660			__u64 hardware_entry_failure_reason;
6661			__u32 cpu; /* if KVM_LAST_CPU */
6662		} fail_entry;
6663
6664If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
6665to unknown reasons.  Further architecture-specific information is
6666available in hardware_entry_failure_reason.
6667
6668::
6669
6670		/* KVM_EXIT_EXCEPTION */
6671		struct {
6672			__u32 exception;
6673			__u32 error_code;
6674		} ex;
6675
6676Unused.
6677
6678::
6679
6680		/* KVM_EXIT_IO */
6681		struct {
6682  #define KVM_EXIT_IO_IN  0
6683  #define KVM_EXIT_IO_OUT 1
6684			__u8 direction;
6685			__u8 size; /* bytes */
6686			__u16 port;
6687			__u32 count;
6688			__u64 data_offset; /* relative to kvm_run start */
6689		} io;
6690
6691If exit_reason is KVM_EXIT_IO, then the vcpu has
6692executed a port I/O instruction which could not be satisfied by kvm.
6693data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
6694where kvm expects application code to place the data for the next
6695KVM_RUN invocation (KVM_EXIT_IO_IN).  Data format is a packed array.
6696
6697::
6698
6699		/* KVM_EXIT_DEBUG */
6700		struct {
6701			struct kvm_debug_exit_arch arch;
6702		} debug;
6703
6704If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
6705for which architecture specific information is returned.
6706
6707::
6708
6709		/* KVM_EXIT_MMIO */
6710		struct {
6711			__u64 phys_addr;
6712			__u8  data[8];
6713			__u32 len;
6714			__u8  is_write;
6715		} mmio;
6716
6717If exit_reason is KVM_EXIT_MMIO, then the vcpu has
6718executed a memory-mapped I/O instruction which could not be satisfied
6719by kvm.  The 'data' member contains the written data if 'is_write' is
6720true, and should be filled by application code otherwise.
6721
6722The 'data' member contains, in its first 'len' bytes, the value as it would
6723appear if the VCPU performed a load or store of the appropriate width directly
6724to the byte array.
6725
6726.. note::
6727
6728      For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
6729      KVM_EXIT_EPR, KVM_EXIT_HYPERCALL, KVM_EXIT_TDX,
6730      KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
6731      operations are complete (and guest state is consistent) only after userspace
6732      has re-entered the kernel with KVM_RUN.  The kernel side will first finish
6733      incomplete operations and then check for pending signals.
6734
6735      The pending state of the operation is not preserved in state which is
6736      visible to userspace, thus userspace should ensure that the operation is
6737      completed before performing a live migration.  Userspace can re-enter the
6738      guest with an unmasked signal pending or with the immediate_exit field set
6739      to complete pending operations without allowing any further instructions
6740      to be executed.
6741
6742::
6743
6744		/* KVM_EXIT_HYPERCALL */
6745		struct {
6746			__u64 nr;
6747			__u64 args[6];
6748			__u64 ret;
6749			__u64 flags;
6750		} hypercall;
6751
6752
6753It is strongly recommended that userspace use ``KVM_EXIT_IO`` (x86) or
6754``KVM_EXIT_MMIO`` (all except s390) to implement functionality that
6755requires a guest to interact with host userspace.
6756
6757.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
6758
6759For arm64:
6760----------
6761
6762SMCCC exits can be enabled depending on the configuration of the SMCCC
6763filter. See the Documentation/virt/kvm/devices/vm.rst
6764``KVM_ARM_SMCCC_FILTER`` for more details.
6765
6766``nr`` contains the function ID of the guest's SMCCC call. Userspace is
6767expected to use the ``KVM_GET_ONE_REG`` ioctl to retrieve the call
6768parameters from the vCPU's GPRs.
6769
6770Definition of ``flags``:
6771 - ``KVM_HYPERCALL_EXIT_SMC``: Indicates that the guest used the SMC
6772   conduit to initiate the SMCCC call. If this bit is 0 then the guest
6773   used the HVC conduit for the SMCCC call.
6774
6775 - ``KVM_HYPERCALL_EXIT_16BIT``: Indicates that the guest used a 16bit
6776   instruction to initiate the SMCCC call. If this bit is 0 then the
6777   guest used a 32bit instruction. An AArch64 guest always has this
6778   bit set to 0.
6779
6780At the point of exit, PC points to the instruction immediately following
6781the trapping instruction.
6782
6783::
6784
6785		/* KVM_EXIT_TPR_ACCESS */
6786		struct {
6787			__u64 rip;
6788			__u32 is_write;
6789			__u32 pad;
6790		} tpr_access;
6791
6792To be documented (KVM_TPR_ACCESS_REPORTING).
6793
6794::
6795
6796		/* KVM_EXIT_S390_SIEIC */
6797		struct {
6798			__u8 icptcode;
6799			__u64 mask; /* psw upper half */
6800			__u64 addr; /* psw lower half */
6801			__u16 ipa;
6802			__u32 ipb;
6803		} s390_sieic;
6804
6805s390 specific.
6806
6807::
6808
6809		/* KVM_EXIT_S390_RESET */
6810  #define KVM_S390_RESET_POR       1
6811  #define KVM_S390_RESET_CLEAR     2
6812  #define KVM_S390_RESET_SUBSYSTEM 4
6813  #define KVM_S390_RESET_CPU_INIT  8
6814  #define KVM_S390_RESET_IPL       16
6815		__u64 s390_reset_flags;
6816
6817s390 specific.
6818
6819::
6820
6821		/* KVM_EXIT_S390_UCONTROL */
6822		struct {
6823			__u64 trans_exc_code;
6824			__u32 pgm_code;
6825		} s390_ucontrol;
6826
6827s390 specific. A page fault has occurred for a user controlled virtual
6828machine (KVM_VM_S390_UNCONTROL) on its host page table that cannot be
6829resolved by the kernel.
6830The program code and the translation exception code that were placed
6831in the cpu's lowcore are presented here as defined by the z Architecture
6832Principles of Operation Book in the Chapter for Dynamic Address Translation
6833(DAT)
6834
6835::
6836
6837		/* KVM_EXIT_DCR */
6838		struct {
6839			__u32 dcrn;
6840			__u32 data;
6841			__u8  is_write;
6842		} dcr;
6843
6844Deprecated - was used for 440 KVM.
6845
6846::
6847
6848		/* KVM_EXIT_OSI */
6849		struct {
6850			__u64 gprs[32];
6851		} osi;
6852
6853MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
6854hypercalls and exit with this exit struct that contains all the guest gprs.
6855
6856If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
6857Userspace can now handle the hypercall and when it's done modify the gprs as
6858necessary. Upon guest entry all guest GPRs will then be replaced by the values
6859in this struct.
6860
6861::
6862
6863		/* KVM_EXIT_PAPR_HCALL */
6864		struct {
6865			__u64 nr;
6866			__u64 ret;
6867			__u64 args[9];
6868		} papr_hcall;
6869
6870This is used on 64-bit PowerPC when emulating a pSeries partition,
6871e.g. with the 'pseries' machine type in qemu.  It occurs when the
6872guest does a hypercall using the 'sc 1' instruction.  The 'nr' field
6873contains the hypercall number (from the guest R3), and 'args' contains
6874the arguments (from the guest R4 - R12).  Userspace should put the
6875return code in 'ret' and any extra returned values in args[].
6876The possible hypercalls are defined in the Power Architecture Platform
6877Requirements (PAPR) document available from www.power.org (free
6878developer registration required to access it).
6879
6880::
6881
6882		/* KVM_EXIT_S390_TSCH */
6883		struct {
6884			__u16 subchannel_id;
6885			__u16 subchannel_nr;
6886			__u32 io_int_parm;
6887			__u32 io_int_word;
6888			__u32 ipb;
6889			__u8 dequeued;
6890		} s390_tsch;
6891
6892s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
6893and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
6894interrupt for the target subchannel has been dequeued and subchannel_id,
6895subchannel_nr, io_int_parm and io_int_word contain the parameters for that
6896interrupt. ipb is needed for instruction parameter decoding.
6897
6898::
6899
6900		/* KVM_EXIT_EPR */
6901		struct {
6902			__u32 epr;
6903		} epr;
6904
6905On FSL BookE PowerPC chips, the interrupt controller has a fast patch
6906interrupt acknowledge path to the core. When the core successfully
6907delivers an interrupt, it automatically populates the EPR register with
6908the interrupt vector number and acknowledges the interrupt inside
6909the interrupt controller.
6910
6911In case the interrupt controller lives in user space, we need to do
6912the interrupt acknowledge cycle through it to fetch the next to be
6913delivered interrupt vector using this exit.
6914
6915It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
6916external interrupt has just been delivered into the guest. User space
6917should put the acknowledged interrupt vector into the 'epr' field.
6918
6919::
6920
6921		/* KVM_EXIT_SYSTEM_EVENT */
6922		struct {
6923  #define KVM_SYSTEM_EVENT_SHUTDOWN       1
6924  #define KVM_SYSTEM_EVENT_RESET          2
6925  #define KVM_SYSTEM_EVENT_CRASH          3
6926  #define KVM_SYSTEM_EVENT_WAKEUP         4
6927  #define KVM_SYSTEM_EVENT_SUSPEND        5
6928  #define KVM_SYSTEM_EVENT_SEV_TERM       6
6929  #define KVM_SYSTEM_EVENT_TDX_FATAL      7
6930			__u32 type;
6931                        __u32 ndata;
6932                        __u64 data[16];
6933		} system_event;
6934
6935If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
6936a system-level event using some architecture specific mechanism (hypercall
6937or some special instruction). In case of ARM64, this is triggered using
6938HVC instruction based PSCI call from the vcpu.
6939
6940The 'type' field describes the system-level event type.
6941Valid values for 'type' are:
6942
6943 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
6944   VM. Userspace is not obliged to honour this, and if it does honour
6945   this does not need to destroy the VM synchronously (ie it may call
6946   KVM_RUN again before shutdown finally occurs).
6947 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
6948   As with SHUTDOWN, userspace can choose to ignore the request, or
6949   to schedule the reset to occur in the future and may call KVM_RUN again.
6950 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
6951   has requested a crash condition maintenance. Userspace can choose
6952   to ignore the request, or to gather VM memory core dump and/or
6953   reset/shutdown of the VM.
6954 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination.
6955   The guest physical address of the guest's GHCB is stored in `data[0]`.
6956 - KVM_SYSTEM_EVENT_TDX_FATAL -- a TDX guest reported a fatal error state.
6957   KVM doesn't do any parsing or conversion, it just dumps 16 general-purpose
6958   registers to userspace, in ascending order of the 4-bit indices for x86-64
6959   general-purpose registers in instruction encoding, as defined in the Intel
6960   SDM.
6961 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and
6962   KVM has recognized a wakeup event. Userspace may honor this event by
6963   marking the exiting vCPU as runnable, or deny it and call KVM_RUN again.
6964 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of
6965   the VM.
6966
6967If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain
6968architecture specific information for the system-level event.  Only
6969the first `ndata` items (possibly zero) of the data array are valid.
6970
6971 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if
6972   the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI
6973   specification.
6974
6975 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2
6976   if the guest issued a SYSTEM_OFF2 call according to v1.3 of the PSCI
6977   specification.
6978
6979 - for RISC-V, data[0] is set to the value of the second argument of the
6980   ``sbi_system_reset`` call.
6981
6982Previous versions of Linux defined a `flags` member in this struct.  The
6983field is now aliased to `data[0]`.  Userspace can assume that it is only
6984written if ndata is greater than 0.
6985
6986For arm/arm64:
6987--------------
6988
6989KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the
6990KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI
6991SYSTEM_SUSPEND function, KVM will exit to userspace with this event
6992type.
6993
6994It is the sole responsibility of userspace to implement the PSCI
6995SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND".
6996KVM does not change the vCPU's state before exiting to userspace, so
6997the call parameters are left in-place in the vCPU registers.
6998
6999Userspace is _required_ to take action for such an exit. It must
7000either:
7001
7002 - Honor the guest request to suspend the VM. Userspace can request
7003   in-kernel emulation of suspension by setting the calling vCPU's
7004   state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's
7005   state according to the parameters passed to the PSCI function when
7006   the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use"
7007   for details on the function parameters.
7008
7009 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2
7010   "Caller responsibilities" for possible return values.
7011
7012Hibernation using the PSCI SYSTEM_OFF2 call is enabled when PSCI v1.3
7013is enabled. If a guest invokes the PSCI SYSTEM_OFF2 function, KVM will
7014exit to userspace with the KVM_SYSTEM_EVENT_SHUTDOWN event type and with
7015data[0] set to KVM_SYSTEM_EVENT_SHUTDOWN_FLAG_PSCI_OFF2. The only
7016supported hibernate type for the SYSTEM_OFF2 function is HIBERNATE_OFF.
7017
7018::
7019
7020		/* KVM_EXIT_IOAPIC_EOI */
7021		struct {
7022			__u8 vector;
7023		} eoi;
7024
7025Indicates that the VCPU's in-kernel local APIC received an EOI for a
7026level-triggered IOAPIC interrupt.  This exit only triggers when the
7027IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
7028the userspace IOAPIC should process the EOI and retrigger the interrupt if
7029it is still asserted.  Vector is the LAPIC interrupt vector for which the
7030EOI was received.
7031
7032::
7033
7034		struct kvm_hyperv_exit {
7035  #define KVM_EXIT_HYPERV_SYNIC          1
7036  #define KVM_EXIT_HYPERV_HCALL          2
7037  #define KVM_EXIT_HYPERV_SYNDBG         3
7038			__u32 type;
7039			__u32 pad1;
7040			union {
7041				struct {
7042					__u32 msr;
7043					__u32 pad2;
7044					__u64 control;
7045					__u64 evt_page;
7046					__u64 msg_page;
7047				} synic;
7048				struct {
7049					__u64 input;
7050					__u64 result;
7051					__u64 params[2];
7052				} hcall;
7053				struct {
7054					__u32 msr;
7055					__u32 pad2;
7056					__u64 control;
7057					__u64 status;
7058					__u64 send_page;
7059					__u64 recv_page;
7060					__u64 pending_page;
7061				} syndbg;
7062			} u;
7063		};
7064		/* KVM_EXIT_HYPERV */
7065                struct kvm_hyperv_exit hyperv;
7066
7067Indicates that the VCPU exits into userspace to process some tasks
7068related to Hyper-V emulation.
7069
7070Valid values for 'type' are:
7071
7072	- KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
7073
7074Hyper-V SynIC state change. Notification is used to remap SynIC
7075event/message pages and to enable/disable SynIC messages/events processing
7076in userspace.
7077
7078	- KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
7079
7080Hyper-V Synthetic debugger state change. Notification is used to either update
7081the pending_page location or to send a control command (send the buffer located
7082in send_page or recv a buffer to recv_page).
7083
7084::
7085
7086		/* KVM_EXIT_ARM_NISV / KVM_EXIT_ARM_LDST64B */
7087		struct {
7088			__u64 esr_iss;
7089			__u64 fault_ipa;
7090		} arm_nisv;
7091
7092- KVM_EXIT_ARM_NISV:
7093
7094Used on arm64 systems. If a guest accesses memory not in a memslot,
7095KVM will typically return to userspace and ask it to do MMIO emulation on its
7096behalf. However, for certain classes of instructions, no instruction decode
7097(direction, length of memory access) is provided, and fetching and decoding
7098the instruction from the VM is overly complicated to live in the kernel.
7099
7100Historically, when this situation occurred, KVM would print a warning and kill
7101the VM. KVM assumed that if the guest accessed non-memslot memory, it was
7102trying to do I/O, which just couldn't be emulated, and the warning message was
7103phrased accordingly. However, what happened more often was that a guest bug
7104caused access outside the guest memory areas which should lead to a more
7105meaningful warning message and an external abort in the guest, if the access
7106did not fall within an I/O window.
7107
7108Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
7109this capability at VM creation. Once this is done, these types of errors will
7110instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
7111the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field.
7112Userspace can either fix up the access if it's actually an I/O access by
7113decoding the instruction from guest memory (if it's very brave) and continue
7114executing the guest, or it can decide to suspend, dump, or restart the guest.
7115
7116Note that KVM does not skip the faulting instruction as it does for
7117KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
7118if it decides to decode and emulate the instruction.
7119
7120This feature isn't available to protected VMs, as userspace does not
7121have access to the state that is required to perform the emulation.
7122Instead, a data abort exception is directly injected in the guest.
7123Note that although KVM_CAP_ARM_NISV_TO_USER will be reported if
7124queried outside of a protected VM context, the feature will not be
7125exposed if queried on a protected VM file descriptor.
7126
7127- KVM_EXIT_ARM_LDST64B:
7128
7129Used on arm64 systems. When a guest using a LD64B, ST64B, ST64BV, ST64BV0,
7130outside of a memslot, KVM will return to userspace with KVM_EXIT_ARM_LDST64B,
7131exposing the relevant ESR_EL2 information and faulting IPA, similarly to
7132KVM_EXIT_ARM_NISV.
7133
7134Userspace is supposed to fully emulate the instructions, which includes:
7135
7136	- fetch of the operands for a store, including ACCDATA_EL1 in the case
7137	  of a ST64BV0 instruction
7138	- deal with the endianness if the guest is big-endian
7139	- emulate the access, including the delivery of an exception if the
7140	  access didn't succeed
7141	- provide a return value in the case of ST64BV/ST64BV0
7142	- return the data in the case of a load
7143	- increment PC if the instruction was successfully executed
7144
7145Note that there is no expectation of performance for this emulation, as it
7146involves a large number of interaction with the guest state. It is, however,
7147expected that the instruction's semantics are preserved, specially the
7148single-copy atomicity property of the 64 byte access.
7149
7150This exit reason must be handled if userspace sets ID_AA64ISAR1_EL1.LS64 to a
7151non-zero value, indicating that FEAT_LS64* is enabled.
7152
7153::
7154
7155		/* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
7156		struct {
7157			__u8 error; /* user -> kernel */
7158			__u8 pad[7];
7159			__u32 reason; /* kernel -> user */
7160			__u32 index; /* kernel -> user */
7161			__u64 data; /* kernel <-> user */
7162		} msr;
7163
7164Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
7165enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
7166may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
7167exit for writes.
7168
7169The "reason" field specifies why the MSR interception occurred. Userspace will
7170only receive MSR exits when a particular reason was requested during through
7171ENABLE_CAP. Currently valid exit reasons are:
7172
7173============================ ========================================
7174 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM
7175 KVM_MSR_EXIT_REASON_INVAL   access to invalid MSRs or reserved bits
7176 KVM_MSR_EXIT_REASON_FILTER  access blocked by KVM_X86_SET_MSR_FILTER
7177============================ ========================================
7178
7179For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest
7180wants to read. To respond to this request with a successful read, userspace
7181writes the respective data into the "data" field and must continue guest
7182execution to ensure the read data is transferred into guest register state.
7183
7184If the RDMSR request was unsuccessful, userspace indicates that with a "1" in
7185the "error" field. This will inject a #GP into the guest when the VCPU is
7186executed again.
7187
7188For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest
7189wants to write. Once finished processing the event, userspace must continue
7190vCPU execution. If the MSR write was unsuccessful, userspace also sets the
7191"error" field to "1".
7192
7193See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering.
7194
7195::
7196
7197
7198		struct kvm_xen_exit {
7199  #define KVM_EXIT_XEN_HCALL          1
7200			__u32 type;
7201			union {
7202				struct {
7203					__u32 longmode;
7204					__u32 cpl;
7205					__u64 input;
7206					__u64 result;
7207					__u64 params[6];
7208				} hcall;
7209			} u;
7210		};
7211		/* KVM_EXIT_XEN */
7212                struct kvm_hyperv_exit xen;
7213
7214Indicates that the VCPU exits into userspace to process some tasks
7215related to Xen emulation.
7216
7217Valid values for 'type' are:
7218
7219  - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
7220    Userspace is expected to place the hypercall result into the appropriate
7221    field before invoking KVM_RUN again.
7222
7223::
7224
7225		/* KVM_EXIT_RISCV_SBI */
7226		struct {
7227			unsigned long extension_id;
7228			unsigned long function_id;
7229			unsigned long args[6];
7230			unsigned long ret[2];
7231		} riscv_sbi;
7232
7233If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has
7234done a SBI call which is not handled by KVM RISC-V kernel module. The details
7235of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The
7236'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the
7237'function_id' field represents function ID of given SBI extension. The 'args'
7238array field of 'riscv_sbi' represents parameters for the SBI call and 'ret'
7239array field represents return values. The userspace should update the return
7240values of SBI call before resuming the VCPU. For more details on RISC-V SBI
7241spec refer, https://github.com/riscv/riscv-sbi-doc.
7242
7243::
7244
7245		/* KVM_EXIT_MEMORY_FAULT */
7246		struct {
7247  #define KVM_MEMORY_EXIT_FLAG_PRIVATE	(1ULL << 3)
7248			__u64 flags;
7249			__u64 gpa;
7250			__u64 size;
7251		} memory_fault;
7252
7253KVM_EXIT_MEMORY_FAULT indicates the vCPU has encountered a memory fault that
7254could not be resolved by KVM.  The 'gpa' and 'size' (in bytes) describe the
7255guest physical address range [gpa, gpa + size) of the fault.  The 'flags' field
7256describes properties of the faulting access that are likely pertinent:
7257
7258 - KVM_MEMORY_EXIT_FLAG_PRIVATE - When set, indicates the memory fault occurred
7259   on a private memory access.  When clear, indicates the fault occurred on a
7260   shared access.
7261
7262Note!  KVM_EXIT_MEMORY_FAULT is unique among all KVM exit reasons in that it
7263accompanies a return code of '-1', not '0'!  errno will always be set to EFAULT
7264or EHWPOISON when KVM exits with KVM_EXIT_MEMORY_FAULT, userspace should assume
7265kvm_run.exit_reason is stale/undefined for all other error numbers.
7266
7267::
7268
7269    /* KVM_EXIT_NOTIFY */
7270    struct {
7271  #define KVM_NOTIFY_CONTEXT_INVALID	(1 << 0)
7272      __u32 flags;
7273    } notify;
7274
7275Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is
7276enabled, a VM exit generated if no event window occurs in VM non-root mode
7277for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when
7278enabling the cap, it would exit to userspace with the exit reason
7279KVM_EXIT_NOTIFY for further handling. The "flags" field contains more
7280detailed info.
7281
7282The valid value for 'flags' is:
7283
7284  - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid
7285    in VMCS. It would run into unknown result if resume the target VM.
7286
7287::
7288
7289		/* KVM_EXIT_TDX */
7290		struct {
7291			__u64 flags;
7292			__u64 nr;
7293			union {
7294				struct {
7295					u64 ret;
7296					u64 data[5];
7297				} unknown;
7298				struct {
7299					u64 ret;
7300					u64 gpa;
7301					u64 size;
7302				} get_quote;
7303				struct {
7304					u64 ret;
7305					u64 leaf;
7306					u64 r11, r12, r13, r14;
7307				} get_tdvmcall_info;
7308				struct {
7309					u64 ret;
7310					u64 vector;
7311				} setup_event_notify;
7312			};
7313		} tdx;
7314
7315Process a TDVMCALL from the guest.  KVM forwards select TDVMCALL based
7316on the Guest-Hypervisor Communication Interface (GHCI) specification;
7317KVM bridges these requests to the userspace VMM with minimal changes,
7318placing the inputs in the union and copying them back to the guest
7319on re-entry.
7320
7321Flags are currently always zero, whereas ``nr`` contains the TDVMCALL
7322number from register R11.  The remaining field of the union provide the
7323inputs and outputs of the TDVMCALL.  Currently the following values of
7324``nr`` are defined:
7325
7326 * ``TDVMCALL_GET_QUOTE``: the guest has requested to generate a TD-Quote
7327   signed by a service hosting TD-Quoting Enclave operating on the host.
7328   Parameters and return value are in the ``get_quote`` field of the union.
7329   The ``gpa`` field and ``size`` specify the guest physical address
7330   (without the shared bit set) and the size of a shared-memory buffer, in
7331   which the TDX guest passes a TD Report.  The ``ret`` field represents
7332   the return value of the GetQuote request.  When the request has been
7333   queued successfully, the TDX guest can poll the status field in the
7334   shared-memory area to check whether the Quote generation is completed or
7335   not. When completed, the generated Quote is returned via the same buffer.
7336
7337 * ``TDVMCALL_GET_TD_VM_CALL_INFO``: the guest has requested the support
7338   status of TDVMCALLs.  The output values for the given leaf should be
7339   placed in fields from ``r11`` to ``r14`` of the ``get_tdvmcall_info``
7340   field of the union.
7341
7342 * ``TDVMCALL_SETUP_EVENT_NOTIFY_INTERRUPT``: the guest has requested to
7343   set up a notification interrupt for vector ``vector``.
7344
7345KVM may add support for more values in the future that may cause a userspace
7346exit, even without calls to ``KVM_ENABLE_CAP`` or similar.  In this case,
7347it will enter with output fields already valid; in the common case, the
7348``unknown.ret`` field of the union will be ``TDVMCALL_STATUS_SUBFUNC_UNSUPPORTED``.
7349Userspace need not do anything if it does not wish to support a TDVMCALL.
7350
7351::
7352
7353		/* KVM_EXIT_ARM_SEA */
7354		struct {
7355  #define KVM_EXIT_ARM_SEA_FLAG_GPA_VALID   (1ULL << 0)
7356			__u64 flags;
7357			__u64 esr;
7358			__u64 gva;
7359			__u64 gpa;
7360		} arm_sea;
7361
7362Used on arm64 systems. When the VM capability ``KVM_CAP_ARM_SEA_TO_USER`` is
7363enabled, a KVM exits to userspace if a guest access causes a synchronous
7364external abort (SEA) and the host APEI fails to handle the SEA.
7365
7366``esr`` is set to a sanitized value of ESR_EL2 from the exception taken to KVM,
7367consisting of the following fields:
7368
7369 - ``ESR_EL2.EC``
7370 - ``ESR_EL2.IL``
7371 - ``ESR_EL2.FnV``
7372 - ``ESR_EL2.EA``
7373 - ``ESR_EL2.CM``
7374 - ``ESR_EL2.WNR``
7375 - ``ESR_EL2.FSC``
7376 - ``ESR_EL2.SET`` (when FEAT_RAS is implemented for the VM)
7377
7378``gva`` is set to the value of FAR_EL2 from the exception taken to KVM when
7379``ESR_EL2.FnV == 0``. Otherwise, the value of ``gva`` is unknown.
7380
7381``gpa`` is set to the faulting IPA from the exception taken to KVM when
7382the ``KVM_EXIT_ARM_SEA_FLAG_GPA_VALID`` flag is set. Otherwise, the value of
7383``gpa`` is unknown.
7384
7385::
7386
7387		/* Fix the size of the union. */
7388		char padding[256];
7389	};
7390
7391	/*
7392	 * shared registers between kvm and userspace.
7393	 * kvm_valid_regs specifies the register classes set by the host
7394	 * kvm_dirty_regs specified the register classes dirtied by userspace
7395	 * struct kvm_sync_regs is architecture specific, as well as the
7396	 * bits for kvm_valid_regs and kvm_dirty_regs
7397	 */
7398	__u64 kvm_valid_regs;
7399	__u64 kvm_dirty_regs;
7400	union {
7401		struct kvm_sync_regs regs;
7402		char padding[SYNC_REGS_SIZE_BYTES];
7403	} s;
7404
7405If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
7406certain guest registers without having to call SET/GET_*REGS. Thus we can
7407avoid some system call overhead if userspace has to handle the exit.
7408Userspace can query the validity of the structure by checking
7409kvm_valid_regs for specific bits. These bits are architecture specific
7410and usually define the validity of a groups of registers. (e.g. one bit
7411for general purpose registers)
7412
7413Please note that the kernel is allowed to use the kvm_run structure as the
7414primary storage for certain register types. Therefore, the kernel may use the
7415values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
7416
7417::
7418
7419		/* KVM_EXIT_SNP_REQ_CERTS */
7420		struct kvm_exit_snp_req_certs {
7421			__u64 gpa;
7422			__u64 npages;
7423			__u64 ret;
7424		};
7425
7426KVM_EXIT_SNP_REQ_CERTS indicates an SEV-SNP guest with certificate-fetching
7427enabled (see KVM_SEV_SNP_ENABLE_REQ_CERTS) has generated an Extended Guest
7428Request NAE #VMGEXIT (SNP_GUEST_REQUEST) with message type MSG_REPORT_REQ,
7429i.e. has requested an attestation report from firmware, and would like the
7430certificate data corresponding to the attestation report signature to be
7431provided by the hypervisor as part of the request.
7432
7433To allow for userspace to provide the certificate, the 'gpa' and 'npages'
7434are forwarded verbatim from the guest request (the RAX and RBX GHCB fields
7435respectively).  'ret' is not an "output" from KVM, and is always '0' on
7436exit.  KVM verifies the 'gpa' is 4KiB aligned prior to exiting to userspace,
7437but otherwise the information from the guest isn't validated.
7438
7439Upon the next KVM_RUN, e.g. after userspace has serviced the request (or not),
7440KVM will complete the #VMGEXIT, using the 'ret' field to determine whether to
7441signal success or failure to the guest, and on failure, what reason code will
7442be communicated via SW_EXITINFO2.  If 'ret' is set to an unsupported value (see
7443the table below), KVM_RUN will fail with -EINVAL.  For a 'ret' of 'ENOSPC', KVM
7444also consumes the 'npages' field, i.e. userspace can use the field to inform
7445the guest of the number of pages needed to hold all the certificate data.
7446
7447The supported 'ret' values and their respective SW_EXITINFO2 encodings:
7448
7449  ======     =============================================================
7450  0          0x0, i.e. success.  KVM will emit an SNP_GUEST_REQUEST command
7451             to SNP firmware.
7452  ENOSPC     0x0000000100000000, i.e. not enough guest pages to hold the
7453             certificate table and certificate data.  KVM will also set the
7454             RBX field in the GHBC to 'npages'.
7455  EAGAIN     0x0000000200000000, i.e. the host is busy and the guest should
7456             retry the request.
7457  EIO        0xffffffff00000000, for all other errors (this return code is
7458             a KVM-defined hypervisor value, as allowed by the GHCB)
7459  ======     =============================================================
7460
7461
7462.. _cap_enable:
7463
74646. Capabilities that can be enabled on vCPUs
7465============================================
7466
7467There are certain capabilities that change the behavior of the virtual CPU or
7468the virtual machine when enabled. To enable them, please see
7469:ref:`KVM_ENABLE_CAP`.
7470
7471Below you can find a list of capabilities and what their effect on the vCPU or
7472the virtual machine is when enabling them.
7473
7474The following information is provided along with the description:
7475
7476  Architectures:
7477      which instruction set architectures provide this ioctl.
7478      x86 includes both i386 and x86_64.
7479
7480  Target:
7481      whether this is a per-vcpu or per-vm capability.
7482
7483  Parameters:
7484      what parameters are accepted by the capability.
7485
7486  Returns:
7487      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
7488      are not detailed, but errors with specific meanings are.
7489
7490
74916.1 KVM_CAP_PPC_OSI
7492-------------------
7493
7494:Architectures: ppc
7495:Target: vcpu
7496:Parameters: none
7497:Returns: 0 on success; -1 on error
7498
7499This capability enables interception of OSI hypercalls that otherwise would
7500be treated as normal system calls to be injected into the guest. OSI hypercalls
7501were invented by Mac-on-Linux to have a standardized communication mechanism
7502between the guest and the host.
7503
7504When this capability is enabled, KVM_EXIT_OSI can occur.
7505
7506
75076.2 KVM_CAP_PPC_PAPR
7508--------------------
7509
7510:Architectures: ppc
7511:Target: vcpu
7512:Parameters: none
7513:Returns: 0 on success; -1 on error
7514
7515This capability enables interception of PAPR hypercalls. PAPR hypercalls are
7516done using the hypercall instruction "sc 1".
7517
7518It also sets the guest privilege level to "supervisor" mode. Usually the guest
7519runs in "hypervisor" privilege mode with a few missing features.
7520
7521In addition to the above, it changes the semantics of SDR1. In this mode, the
7522HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
7523HTAB invisible to the guest.
7524
7525When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
7526
7527
75286.3 KVM_CAP_SW_TLB
7529------------------
7530
7531:Architectures: ppc
7532:Target: vcpu
7533:Parameters: args[0] is the address of a struct kvm_config_tlb
7534:Returns: 0 on success; -1 on error
7535
7536::
7537
7538  struct kvm_config_tlb {
7539	__u64 params;
7540	__u64 array;
7541	__u32 mmu_type;
7542	__u32 array_len;
7543  };
7544
7545Configures the virtual CPU's TLB array, establishing a shared memory area
7546between userspace and KVM.  The "params" and "array" fields are userspace
7547addresses of mmu-type-specific data structures.  The "array_len" field is an
7548safety mechanism, and should be set to the size in bytes of the memory that
7549userspace has reserved for the array.  It must be at least the size dictated
7550by "mmu_type" and "params".
7551
7552While KVM_RUN is active, the shared region is under control of KVM.  Its
7553contents are undefined, and any modification by userspace results in
7554boundedly undefined behavior.
7555
7556On return from KVM_RUN, the shared region will reflect the current state of
7557the guest's TLB.  If userspace makes any changes, it must call KVM_DIRTY_TLB
7558to tell KVM which entries have been changed, prior to calling KVM_RUN again
7559on this vcpu.
7560
7561For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
7562
7563 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
7564 - The "array" field points to an array of type "struct
7565   kvm_book3e_206_tlb_entry".
7566 - The array consists of all entries in the first TLB, followed by all
7567   entries in the second TLB.
7568 - Within a TLB, entries are ordered first by increasing set number.  Within a
7569   set, entries are ordered by way (increasing ESEL).
7570 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
7571   where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
7572 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
7573   hardware ignores this value for TLB0.
7574
75756.4 KVM_CAP_S390_CSS_SUPPORT
7576----------------------------
7577
7578:Architectures: s390
7579:Target: vcpu
7580:Parameters: none
7581:Returns: 0 on success; -1 on error
7582
7583This capability enables support for handling of channel I/O instructions.
7584
7585TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
7586handled in-kernel, while the other I/O instructions are passed to userspace.
7587
7588When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
7589SUBCHANNEL intercepts.
7590
7591Note that even though this capability is enabled per-vcpu, the complete
7592virtual machine is affected.
7593
75946.5 KVM_CAP_PPC_EPR
7595-------------------
7596
7597:Architectures: ppc
7598:Target: vcpu
7599:Parameters: args[0] defines whether the proxy facility is active
7600:Returns: 0 on success; -1 on error
7601
7602This capability enables or disables the delivery of interrupts through the
7603external proxy facility.
7604
7605When enabled (args[0] != 0), every time the guest gets an external interrupt
7606delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
7607to receive the topmost interrupt vector.
7608
7609When disabled (args[0] == 0), behavior is as if this facility is unsupported.
7610
7611When this capability is enabled, KVM_EXIT_EPR can occur.
7612
76136.6 KVM_CAP_IRQ_MPIC
7614--------------------
7615
7616:Architectures: ppc
7617:Parameters: args[0] is the MPIC device fd;
7618             args[1] is the MPIC CPU number for this vcpu
7619
7620This capability connects the vcpu to an in-kernel MPIC device.
7621
76226.7 KVM_CAP_IRQ_XICS
7623--------------------
7624
7625:Architectures: ppc
7626:Target: vcpu
7627:Parameters: args[0] is the XICS device fd;
7628             args[1] is the XICS CPU number (server ID) for this vcpu
7629
7630This capability connects the vcpu to an in-kernel XICS device.
7631
76326.8 KVM_CAP_S390_IRQCHIP
7633------------------------
7634
7635:Architectures: s390
7636:Target: vm
7637:Parameters: none
7638
7639This capability enables the in-kernel irqchip for s390. Please refer to
7640"4.24 KVM_CREATE_IRQCHIP" for details.
7641
76426.9 KVM_CAP_MIPS_FPU
7643--------------------
7644
7645:Architectures: mips
7646:Target: vcpu
7647:Parameters: args[0] is reserved for future use (should be 0).
7648
7649This capability allows the use of the host Floating Point Unit by the guest. It
7650allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
7651done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
7652accessed (depending on the current guest FPU register mode), and the Status.FR,
7653Config5.FRE bits are accessible via the KVM API and also from the guest,
7654depending on them being supported by the FPU.
7655
76566.10 KVM_CAP_MIPS_MSA
7657---------------------
7658
7659:Architectures: mips
7660:Target: vcpu
7661:Parameters: args[0] is reserved for future use (should be 0).
7662
7663This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
7664It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
7665Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
7666registers can be accessed, and the Config5.MSAEn bit is accessible via the
7667KVM API and also from the guest.
7668
76696.74 KVM_CAP_SYNC_REGS
7670----------------------
7671
7672:Architectures: s390, x86
7673:Target: s390: always enabled, x86: vcpu
7674:Parameters: none
7675:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
7676          sets are supported
7677          (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
7678
7679As described above in the kvm_sync_regs struct info in section :ref:`kvm_run`,
7680KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
7681without having to call SET/GET_*REGS". This reduces overhead by eliminating
7682repeated ioctl calls for setting and/or getting register values. This is
7683particularly important when userspace is making synchronous guest state
7684modifications, e.g. when emulating and/or intercepting instructions in
7685userspace.
7686
7687For s390 specifics, please refer to the source code.
7688
7689For x86:
7690
7691- the register sets to be copied out to kvm_run are selectable
7692  by userspace (rather that all sets being copied out for every exit).
7693- vcpu_events are available in addition to regs and sregs.
7694
7695For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
7696function as an input bit-array field set by userspace to indicate the
7697specific register sets to be copied out on the next exit.
7698
7699To indicate when userspace has modified values that should be copied into
7700the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
7701This is done using the same bitflags as for the 'kvm_valid_regs' field.
7702If the dirty bit is not set, then the register set values will not be copied
7703into the vCPU even if they've been modified.
7704
7705Unused bitfields in the bitarrays must be set to zero.
7706
7707::
7708
7709  struct kvm_sync_regs {
7710        struct kvm_regs regs;
7711        struct kvm_sregs sregs;
7712        struct kvm_vcpu_events events;
7713  };
7714
77156.75 KVM_CAP_PPC_IRQ_XIVE
7716-------------------------
7717
7718:Architectures: ppc
7719:Target: vcpu
7720:Parameters: args[0] is the XIVE device fd;
7721             args[1] is the XIVE CPU number (server ID) for this vcpu
7722
7723This capability connects the vcpu to an in-kernel XIVE device.
7724
77256.76 KVM_CAP_HYPERV_SYNIC
7726-------------------------
7727
7728:Architectures: x86
7729:Target: vcpu
7730
7731This capability, if KVM_CHECK_EXTENSION indicates that it is
7732available, means that the kernel has an implementation of the
7733Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
7734used to support Windows Hyper-V based guest paravirt drivers(VMBus).
7735
7736In order to use SynIC, it has to be activated by setting this
7737capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
7738will disable the use of APIC hardware virtualization even if supported
7739by the CPU, as it's incompatible with SynIC auto-EOI behavior.
7740
77416.77 KVM_CAP_HYPERV_SYNIC2
7742--------------------------
7743
7744:Architectures: x86
7745:Target: vcpu
7746
7747This capability enables a newer version of Hyper-V Synthetic interrupt
7748controller (SynIC).  The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
7749doesn't clear SynIC message and event flags pages when they are enabled by
7750writing to the respective MSRs.
7751
77526.78 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
7753-----------------------------------
7754
7755:Architectures: x86
7756:Target: vcpu
7757
7758This capability indicates that KVM running on top of Hyper-V hypervisor
7759enables Direct TLB flush for its guests meaning that TLB flush
7760hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
7761Due to the different ABI for hypercall parameters between Hyper-V and
7762KVM, enabling this capability effectively disables all hypercall
7763handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
7764flush hypercalls by Hyper-V) so userspace should disable KVM identification
7765in CPUID and only exposes Hyper-V identification. In this case, guest
7766thinks it's running on Hyper-V and only use Hyper-V hypercalls.
7767
77686.79 KVM_CAP_HYPERV_ENFORCE_CPUID
7769---------------------------------
7770
7771:Architectures: x86
7772:Target: vcpu
7773
7774When enabled, KVM will disable emulated Hyper-V features provided to the
7775guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
7776currently implemented Hyper-V features are provided unconditionally when
7777Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
7778leaf.
7779
77806.80 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
7781-------------------------------------
7782
7783:Architectures: x86
7784:Target: vcpu
7785
7786When enabled, KVM will disable paravirtual features provided to the
7787guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
7788(0x40000001). Otherwise, a guest may use the paravirtual features
7789regardless of what has actually been exposed through the CPUID leaf.
7790
7791.. _KVM_CAP_DIRTY_LOG_RING:
7792
7793
7794.. _cap_enable_vm:
7795
77967. Capabilities that can be enabled on VMs
7797==========================================
7798
7799There are certain capabilities that change the behavior of the virtual
7800machine when enabled. To enable them, please see section
7801:ref:`KVM_ENABLE_CAP`. Below you can find a list of capabilities and
7802what their effect on the VM is when enabling them.
7803
7804The following information is provided along with the description:
7805
7806  Architectures:
7807      which instruction set architectures provide this ioctl.
7808      x86 includes both i386 and x86_64.
7809
7810  Parameters:
7811      what parameters are accepted by the capability.
7812
7813  Returns:
7814      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
7815      are not detailed, but errors with specific meanings are.
7816
7817
78187.1 KVM_CAP_PPC_ENABLE_HCALL
7819----------------------------
7820
7821:Architectures: ppc
7822:Parameters: args[0] is the sPAPR hcall number;
7823	     args[1] is 0 to disable, 1 to enable in-kernel handling
7824
7825This capability controls whether individual sPAPR hypercalls (hcalls)
7826get handled by the kernel or not.  Enabling or disabling in-kernel
7827handling of an hcall is effective across the VM.  On creation, an
7828initial set of hcalls are enabled for in-kernel handling, which
7829consists of those hcalls for which in-kernel handlers were implemented
7830before this capability was implemented.  If disabled, the kernel will
7831not to attempt to handle the hcall, but will always exit to userspace
7832to handle it.  Note that it may not make sense to enable some and
7833disable others of a group of related hcalls, but KVM does not prevent
7834userspace from doing that.
7835
7836If the hcall number specified is not one that has an in-kernel
7837implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
7838error.
7839
78407.2 KVM_CAP_S390_USER_SIGP
7841--------------------------
7842
7843:Architectures: s390
7844:Parameters: none
7845
7846This capability controls which SIGP orders will be handled completely in user
7847space. With this capability enabled, all fast orders will be handled completely
7848in the kernel:
7849
7850- SENSE
7851- SENSE RUNNING
7852- EXTERNAL CALL
7853- EMERGENCY SIGNAL
7854- CONDITIONAL EMERGENCY SIGNAL
7855
7856All other orders will be handled completely in user space.
7857
7858Only privileged operation exceptions will be checked for in the kernel (or even
7859in the hardware prior to interception). If this capability is not enabled, the
7860old way of handling SIGP orders is used (partially in kernel and user space).
7861
78627.3 KVM_CAP_S390_VECTOR_REGISTERS
7863---------------------------------
7864
7865:Architectures: s390
7866:Parameters: none
7867:Returns: 0 on success, negative value on error
7868
7869Allows use of the vector registers introduced with z13 processor, and
7870provides for the synchronization between host and user space.  Will
7871return -EINVAL if the machine does not support vectors.
7872
78737.4 KVM_CAP_S390_USER_STSI
7874--------------------------
7875
7876:Architectures: s390
7877:Parameters: none
7878
7879This capability allows post-handlers for the STSI instruction. After
7880initial handling in the kernel, KVM exits to user space with
7881KVM_EXIT_S390_STSI to allow user space to insert further data.
7882
7883Before exiting to userspace, kvm handlers should fill in s390_stsi field of
7884vcpu->run::
7885
7886  struct {
7887	__u64 addr;
7888	__u8 ar;
7889	__u8 reserved;
7890	__u8 fc;
7891	__u8 sel1;
7892	__u16 sel2;
7893  } s390_stsi;
7894
7895  @addr - guest address of STSI SYSIB
7896  @fc   - function code
7897  @sel1 - selector 1
7898  @sel2 - selector 2
7899  @ar   - access register number
7900
7901KVM handlers should exit to userspace with rc = -EREMOTE.
7902
79037.5 KVM_CAP_SPLIT_IRQCHIP
7904-------------------------
7905
7906:Architectures: x86
7907:Parameters: args[0] - number of routes reserved for userspace IOAPICs
7908:Returns: 0 on success, -1 on error
7909
7910Create a local apic for each processor in the kernel. This can be used
7911instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
7912IOAPIC and PIC (and also the PIT, even though this has to be enabled
7913separately).
7914
7915This capability also enables in kernel routing of interrupt requests;
7916when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
7917used in the IRQ routing table.  The first args[0] MSI routes are reserved
7918for the IOAPIC pins.  Whenever the LAPIC receives an EOI for these routes,
7919a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
7920
7921Fails if VCPU has already been created, or if the irqchip is already in the
7922kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
7923
79247.6 KVM_CAP_S390_RI
7925-------------------
7926
7927:Architectures: s390
7928:Parameters: none
7929
7930Allows use of runtime-instrumentation introduced with zEC12 processor.
7931Will return -EINVAL if the machine does not support runtime-instrumentation.
7932Will return -EBUSY if a VCPU has already been created.
7933
79347.7 KVM_CAP_X2APIC_API
7935----------------------
7936
7937:Architectures: x86
7938:Parameters: args[0] - features that should be enabled
7939:Returns: 0 on success, -EINVAL when args[0] contains invalid features
7940
7941Valid feature flags in args[0] are::
7942
7943  #define KVM_X2APIC_API_USE_32BIT_IDS                          (1ULL << 0)
7944  #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK                (1ULL << 1)
7945  #define KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST              (1ULL << 2)
7946  #define KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST             (1ULL << 3)
7947
7948Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
7949KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
7950allowing the use of 32-bit APIC IDs.  See KVM_CAP_X2APIC_API in their
7951respective sections.
7952
7953KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
7954in logical mode or with more than 255 VCPUs.  Otherwise, KVM treats 0xff
7955as a broadcast even in x2APIC mode in order to support physical x2APIC
7956without interrupt remapping.  This is undesirable in logical mode,
7957where 0xff represents CPUs 0-7 in cluster 0.
7958
7959Setting KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST instructs KVM to enable
7960Suppress EOI Broadcasts.  KVM will advertise support for Suppress EOI
7961Broadcast to the guest and suppress LAPIC EOI broadcasts when the guest
7962sets the Suppress EOI Broadcast bit in the SPIV register.  This flag is
7963supported only when using a split IRQCHIP.
7964
7965Setting KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST disables support for
7966Suppress EOI Broadcasts entirely, i.e. instructs KVM to NOT advertise
7967support to the guest.
7968
7969Modern VMMs should either enable KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST
7970or KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST.  If not, legacy quirky
7971behavior will be used by KVM: in split IRQCHIP mode, KVM will advertise
7972support for Suppress EOI Broadcasts but not actually suppress EOI
7973broadcasts; for in-kernel IRQCHIP mode, KVM will not advertise support for
7974Suppress EOI Broadcasts.
7975
7976Setting both KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST and
7977KVM_X2APIC_DISABLE_SUPPRESS_EOI_BROADCAST will fail with an EINVAL error,
7978as will setting KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST without a split
7979IRCHIP.
7980
79817.8 KVM_CAP_S390_USER_INSTR0
7982----------------------------
7983
7984:Architectures: s390
7985:Parameters: none
7986
7987With this capability enabled, the illegal instruction 0x0000 (2 bytes) will
7988be intercepted and forwarded to user space. User space can use this
7989mechanism e.g. to realize 2-byte software breakpoints. The kernel will
7990not inject an operating exception for these instructions, user space has
7991to take care of that.
7992
7993This capability can be enabled dynamically even if VCPUs were already
7994created and are running.
7995
79967.9 KVM_CAP_S390_GS
7997-------------------
7998
7999:Architectures: s390
8000:Parameters: none
8001:Returns: 0 on success; -EINVAL if the machine does not support
8002          guarded storage; -EBUSY if a VCPU has already been created.
8003
8004Allows use of guarded storage for the KVM guest.
8005
80067.10 KVM_CAP_S390_AIS
8007---------------------
8008
8009:Architectures: s390
8010:Parameters: none
8011
8012Allow use of adapter-interruption suppression.
8013:Returns: 0 on success; -EBUSY if a VCPU has already been created.
8014
80157.11 KVM_CAP_PPC_SMT
8016--------------------
8017
8018:Architectures: ppc
8019:Parameters: vsmt_mode, flags
8020
8021Enabling this capability on a VM provides userspace with a way to set
8022the desired virtual SMT mode (i.e. the number of virtual CPUs per
8023virtual core).  The virtual SMT mode, vsmt_mode, must be a power of 2
8024between 1 and 8.  On POWER8, vsmt_mode must also be no greater than
8025the number of threads per subcore for the host.  Currently flags must
8026be 0.  A successful call to enable this capability will result in
8027vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
8028subsequently queried for the VM.  This capability is only supported by
8029HV KVM, and can only be set before any VCPUs have been created.
8030The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
8031modes are available.
8032
80337.12 KVM_CAP_PPC_FWNMI
8034----------------------
8035
8036:Architectures: ppc
8037:Parameters: none
8038
8039With this capability a machine check exception in the guest address
8040space will cause KVM to exit the guest with NMI exit reason. This
8041enables QEMU to build error log and branch to guest kernel registered
8042machine check handling routine. Without this capability KVM will
8043branch to guests' 0x200 interrupt vector.
8044
80457.13 KVM_CAP_X86_DISABLE_EXITS
8046------------------------------
8047
8048:Architectures: x86
8049:Parameters: args[0] defines which exits are disabled
8050:Returns: 0 on success, -EINVAL when args[0] contains invalid exits
8051          or if any vCPUs have already been created
8052
8053Valid bits in args[0] are::
8054
8055  #define KVM_X86_DISABLE_EXITS_MWAIT            (1 << 0)
8056  #define KVM_X86_DISABLE_EXITS_HLT              (1 << 1)
8057  #define KVM_X86_DISABLE_EXITS_PAUSE            (1 << 2)
8058  #define KVM_X86_DISABLE_EXITS_CSTATE           (1 << 3)
8059  #define KVM_X86_DISABLE_EXITS_APERFMPERF       (1 << 4)
8060
8061Enabling this capability on a VM provides userspace with a way to no
8062longer intercept some instructions for improved latency in some
8063workloads, and is suggested when vCPUs are associated to dedicated
8064physical CPUs.  More bits can be added in the future; userspace can
8065just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
8066all such vmexits.
8067
8068Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
8069
8070Virtualizing the ``IA32_APERF`` and ``IA32_MPERF`` MSRs requires more
8071than just disabling APERF/MPERF exits. While both Intel and AMD
8072document strict usage conditions for these MSRs--emphasizing that only
8073the ratio of their deltas over a time interval (T0 to T1) is
8074architecturally defined--simply passing through the MSRs can still
8075produce an incorrect ratio.
8076
8077This erroneous ratio can occur if, between T0 and T1:
8078
80791. The vCPU thread migrates between logical processors.
80802. Live migration or suspend/resume operations take place.
80813. Another task shares the vCPU's logical processor.
80824. C-states lower than C0 are emulated (e.g., via HLT interception).
80835. The guest TSC frequency doesn't match the host TSC frequency.
8084
8085Due to these complexities, KVM does not automatically associate this
8086passthrough capability with the guest CPUID bit,
8087``CPUID.6:ECX.APERFMPERF[bit 0]``. Userspace VMMs that deem this
8088mechanism adequate for virtualizing the ``IA32_APERF`` and
8089``IA32_MPERF`` MSRs must set the guest CPUID bit explicitly.
8090
8091
80927.14 KVM_CAP_S390_HPAGE_1M
8093--------------------------
8094
8095:Architectures: s390
8096:Parameters: none
8097:Returns: 0 on success, -EINVAL if hpage module parameter was not set
8098	  or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
8099	  flag set
8100
8101With this capability the KVM support for memory backing with 1m pages
8102through hugetlbfs can be enabled for a VM. After the capability is
8103enabled, cmma can't be enabled anymore and pfmfi and the storage key
8104interpretation are disabled. If cmma has already been enabled or the
8105hpage module parameter is not set to 1, -EINVAL is returned.
8106
8107While it is generally possible to create a huge page backed VM without
8108this capability, the VM will not be able to run.
8109
81107.15 KVM_CAP_MSR_PLATFORM_INFO
8111------------------------------
8112
8113:Architectures: x86
8114:Parameters: args[0] whether feature should be enabled or not
8115
8116With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
8117a #GP would be raised when the guest tries to access. Currently, this
8118capability does not enable write permissions of this MSR for the guest.
8119
81207.16 KVM_CAP_PPC_NESTED_HV
8121--------------------------
8122
8123:Architectures: ppc
8124:Parameters: none
8125:Returns: 0 on success, -EINVAL when the implementation doesn't support
8126	  nested-HV virtualization.
8127
8128HV-KVM on POWER9 and later systems allows for "nested-HV"
8129virtualization, which provides a way for a guest VM to run guests that
8130can run using the CPU's supervisor mode (privileged non-hypervisor
8131state).  Enabling this capability on a VM depends on the CPU having
8132the necessary functionality and on the facility being enabled with a
8133kvm-hv module parameter.
8134
81357.17 KVM_CAP_EXCEPTION_PAYLOAD
8136------------------------------
8137
8138:Architectures: x86
8139:Parameters: args[0] whether feature should be enabled or not
8140
8141With this capability enabled, CR2 will not be modified prior to the
8142emulated VM-exit when L1 intercepts a #PF exception that occurs in
8143L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
8144the emulated VM-exit when L1 intercepts a #DB exception that occurs in
8145L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
8146#DB) exception for L2, exception.has_payload will be set and the
8147faulting address (or the new DR6 bits*) will be reported in the
8148exception_payload field. Similarly, when userspace injects a #PF (or
8149#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
8150exception.has_payload and to put the faulting address - or the new DR6
8151bits\ [#]_ - in the exception_payload field.
8152
8153This capability also enables exception.pending in struct
8154kvm_vcpu_events, which allows userspace to distinguish between pending
8155and injected exceptions.
8156
8157
8158.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
8159       will clear DR6.RTM.
8160
81617.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
8162--------------------------------------
8163
8164:Architectures: x86, arm64, mips
8165:Parameters: args[0] whether feature should be enabled or not
8166
8167Valid flags are::
8168
8169  #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE   (1 << 0)
8170  #define KVM_DIRTY_LOG_INITIALLY_SET           (1 << 1)
8171
8172With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
8173automatically clear and write-protect all pages that are returned as dirty.
8174Rather, userspace will have to do this operation separately using
8175KVM_CLEAR_DIRTY_LOG.
8176
8177At the cost of a slightly more complicated operation, this provides better
8178scalability and responsiveness for two reasons.  First,
8179KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
8180than requiring to sync a full memslot; this ensures that KVM does not
8181take spinlocks for an extended period of time.  Second, in some cases a
8182large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
8183userspace actually using the data in the page.  Pages can be modified
8184during this time, which is inefficient for both the guest and userspace:
8185the guest will incur a higher penalty due to write protection faults,
8186while userspace can see false reports of dirty pages.  Manual reprotection
8187helps reducing this time, improving guest performance and reducing the
8188number of dirty log false positives.
8189
8190With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
8191will be initialized to 1 when created.  This also improves performance because
8192dirty logging can be enabled gradually in small chunks on the first call
8193to KVM_CLEAR_DIRTY_LOG.  KVM_DIRTY_LOG_INITIALLY_SET depends on
8194KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
8195x86, arm64 and riscv for now).
8196
8197KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
8198KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
8199it hard or impossible to use it correctly.  The availability of
8200KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
8201Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
8202
82037.19 KVM_CAP_PPC_SECURE_GUEST
8204------------------------------
8205
8206:Architectures: ppc
8207
8208This capability indicates that KVM is running on a host that has
8209ultravisor firmware and thus can support a secure guest.  On such a
8210system, a guest can ask the ultravisor to make it a secure guest,
8211one whose memory is inaccessible to the host except for pages which
8212are explicitly requested to be shared with the host.  The ultravisor
8213notifies KVM when a guest requests to become a secure guest, and KVM
8214has the opportunity to veto the transition.
8215
8216If present, this capability can be enabled for a VM, meaning that KVM
8217will allow the transition to secure guest mode.  Otherwise KVM will
8218veto the transition.
8219
82207.20 KVM_CAP_HALT_POLL
8221----------------------
8222
8223:Architectures: all
8224:Target: VM
8225:Parameters: args[0] is the maximum poll time in nanoseconds
8226:Returns: 0 on success; -1 on error
8227
8228KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the
8229maximum halt-polling time for all vCPUs in the target VM. This capability can
8230be invoked at any time and any number of times to dynamically change the
8231maximum halt-polling time.
8232
8233See Documentation/virt/kvm/halt-polling.rst for more information on halt
8234polling.
8235
82367.21 KVM_CAP_X86_USER_SPACE_MSR
8237-------------------------------
8238
8239:Architectures: x86
8240:Target: VM
8241:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
8242:Returns: 0 on success; -1 on error
8243
8244This capability allows userspace to intercept RDMSR and WRMSR instructions if
8245access to an MSR is denied.  By default, KVM injects #GP on denied accesses.
8246
8247When a guest requests to read or write an MSR, KVM may not implement all MSRs
8248that are relevant to a respective system. It also does not differentiate by
8249CPU type.
8250
8251To allow more fine grained control over MSR handling, userspace may enable
8252this capability. With it enabled, MSR accesses that match the mask specified in
8253args[0] and would trigger a #GP inside the guest will instead trigger
8254KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications.  Userspace
8255can then implement model specific MSR handling and/or user notifications
8256to inform a user that an MSR was not emulated/virtualized by KVM.
8257
8258The valid mask flags are:
8259
8260============================ ===============================================
8261 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs
8262 KVM_MSR_EXIT_REASON_INVAL   intercept accesses that are architecturally
8263                             invalid according to the vCPU model and/or mode
8264 KVM_MSR_EXIT_REASON_FILTER  intercept accesses that are denied by userspace
8265                             via KVM_X86_SET_MSR_FILTER
8266============================ ===============================================
8267
82687.22 KVM_CAP_X86_BUS_LOCK_EXIT
8269-------------------------------
8270
8271:Architectures: x86
8272:Target: VM
8273:Parameters: args[0] defines the policy used when bus locks detected in guest
8274:Returns: 0 on success, -EINVAL when args[0] contains invalid bits
8275
8276Valid bits in args[0] are::
8277
8278  #define KVM_BUS_LOCK_DETECTION_OFF      (1 << 0)
8279  #define KVM_BUS_LOCK_DETECTION_EXIT     (1 << 1)
8280
8281Enabling this capability on a VM provides userspace with a way to select a
8282policy to handle the bus locks detected in guest. Userspace can obtain the
8283supported modes from the result of KVM_CHECK_EXTENSION and define it through
8284the KVM_ENABLE_CAP. The supported modes are mutually-exclusive.
8285
8286This capability allows userspace to force VM exits on bus locks detected in the
8287guest, irrespective whether or not the host has enabled split-lock detection
8288(which triggers an #AC exception that KVM intercepts). This capability is
8289intended to mitigate attacks where a malicious/buggy guest can exploit bus
8290locks to degrade the performance of the whole system.
8291
8292If KVM_BUS_LOCK_DETECTION_OFF is set, KVM doesn't force guest bus locks to VM
8293exit, although the host kernel's split-lock #AC detection still applies, if
8294enabled.
8295
8296If KVM_BUS_LOCK_DETECTION_EXIT is set, KVM enables a CPU feature that ensures
8297bus locks in the guest trigger a VM exit, and KVM exits to userspace for all
8298such VM exits, e.g. to allow userspace to throttle the offending guest and/or
8299apply some other policy-based mitigation. When exiting to userspace, KVM sets
8300KVM_RUN_X86_BUS_LOCK in vcpu-run->flags, and conditionally sets the exit_reason
8301to KVM_EXIT_X86_BUS_LOCK.
8302
8303Due to differences in the underlying hardware implementation, the vCPU's RIP at
8304the time of exit diverges between Intel and AMD.  On Intel hosts, RIP points at
8305the next instruction, i.e. the exit is trap-like.  On AMD hosts, RIP points at
8306the offending instruction, i.e. the exit is fault-like.
8307
8308Note! Detected bus locks may be coincident with other exits to userspace, i.e.
8309KVM_RUN_X86_BUS_LOCK should be checked regardless of the primary exit reason if
8310userspace wants to take action on all detected bus locks.
8311
83127.23 KVM_CAP_PPC_DAWR1
8313----------------------
8314
8315:Architectures: ppc
8316:Parameters: none
8317:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
8318
8319This capability can be used to check / enable 2nd DAWR feature provided
8320by POWER10 processor.
8321
8322
83237.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
8324-------------------------------------
8325
8326:Architectures: x86 SEV enabled
8327:Type: vm
8328:Parameters: args[0] is the fd of the source vm
8329:Returns: 0 on success; ENOTTY on error
8330
8331This capability enables userspace to copy encryption context from the vm
8332indicated by the fd to the vm this is called on.
8333
8334This is intended to support in-guest workloads scheduled by the host. This
8335allows the in-guest workload to maintain its own NPTs and keeps the two vms
8336from accidentally clobbering each other with interrupts and the like (separate
8337APIC/MSRs/etc).
8338
83397.25 KVM_CAP_SGX_ATTRIBUTE
8340--------------------------
8341
8342:Architectures: x86
8343:Target: VM
8344:Parameters: args[0] is a file handle of a SGX attribute file in securityfs
8345:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
8346          attribute is not supported by KVM.
8347
8348KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
8349more privileged enclave attributes.  args[0] must hold a file handle to a valid
8350SGX attribute file corresponding to an attribute that is supported/restricted
8351by KVM (currently only PROVISIONKEY).
8352
8353The SGX subsystem restricts access to a subset of enclave attributes to provide
8354additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
8355is restricted to deter malware from using the PROVISIONKEY to obtain a stable
8356system fingerprint.  To prevent userspace from circumventing such restrictions
8357by running an enclave in a VM, KVM prevents access to privileged attributes by
8358default.
8359
8360See Documentation/arch/x86/sgx.rst for more details.
8361
83627.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
8363--------------------------------------
8364
8365:Architectures: x86
8366:Parameters: args[0] whether the feature should be enabled or not
8367
8368When this capability is enabled, an emulation failure will result in an exit
8369to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
8370to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
8371to 15 instruction bytes for any exit to userspace resulting from an emulation
8372failure.  When these exits to userspace occur use the emulation_failure struct
8373instead of the internal struct.  They both have the same layout, but the
8374emulation_failure struct matches the content better.  It also explicitly
8375defines the 'flags' field which is used to describe the fields in the struct
8376that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
8377set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
8378in them.)
8379
83807.28 KVM_CAP_ARM_MTE
8381--------------------
8382
8383:Architectures: arm64
8384:Parameters: none
8385
8386This capability indicates that KVM (and the hardware) supports exposing the
8387Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
8388VMM before creating any VCPUs to allow the guest access. Note that MTE is only
8389available to a guest running in AArch64 mode and enabling this capability will
8390cause attempts to create AArch32 VCPUs to fail.
8391
8392When enabled the guest is able to access tags associated with any memory given
8393to the guest. KVM will ensure that the tags are maintained during swap or
8394hibernation of the host; however the VMM needs to manually save/restore the
8395tags as appropriate if the VM is migrated.
8396
8397When this capability is enabled all memory in memslots must be mapped as
8398``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``),
8399attempts to create a memslot with an invalid mmap will result in an
8400-EINVAL return.
8401
8402When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
8403perform a bulk copy of tags to/from the guest.
8404
84057.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM
8406-------------------------------------
8407
8408:Architectures: x86 SEV enabled
8409:Type: vm
8410:Parameters: args[0] is the fd of the source vm
8411:Returns: 0 on success
8412
8413This capability enables userspace to migrate the encryption context from the VM
8414indicated by the fd to the VM this is called on.
8415
8416This is intended to support intra-host migration of VMs between userspace VMMs,
8417upgrading the VMM process without interrupting the guest.
8418
84197.31 KVM_CAP_DISABLE_QUIRKS2
8420----------------------------
8421
8422:Parameters: args[0] - set of KVM quirks to disable
8423:Architectures: x86
8424:Type: vm
8425
8426This capability, if enabled, will cause KVM to disable some behavior
8427quirks.
8428
8429Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
8430quirks that can be disabled in KVM.
8431
8432The argument to KVM_ENABLE_CAP for this capability is a bitmask of
8433quirks to disable, and must be a subset of the bitmask returned by
8434KVM_CHECK_EXTENSION.
8435
8436The valid bits in cap.args[0] are:
8437
8438=================================== ============================================
8439 KVM_X86_QUIRK_LINT0_REENABLED      By default, the reset value for the LVT
8440                                    LINT0 register is 0x700 (APIC_MODE_EXTINT).
8441                                    When this quirk is disabled, the reset value
8442                                    is 0x10000 (APIC_LVT_MASKED).
8443
8444 KVM_X86_QUIRK_CD_NW_CLEARED        By default, KVM clears CR0.CD and CR0.NW on
8445                                    AMD CPUs to workaround buggy guest firmware
8446                                    that runs in perpetuity with CR0.CD, i.e.
8447                                    with caches in "no fill" mode.
8448
8449                                    When this quirk is disabled, KVM does not
8450                                    change the value of CR0.CD and CR0.NW.
8451
8452 KVM_X86_QUIRK_LAPIC_MMIO_HOLE      By default, the MMIO LAPIC interface is
8453                                    available even when configured for x2APIC
8454                                    mode. When this quirk is disabled, KVM
8455                                    disables the MMIO LAPIC interface if the
8456                                    LAPIC is in x2APIC mode.
8457
8458 KVM_X86_QUIRK_OUT_7E_INC_RIP       By default, KVM pre-increments %rip before
8459                                    exiting to userspace for an OUT instruction
8460                                    to port 0x7e. When this quirk is disabled,
8461                                    KVM does not pre-increment %rip before
8462                                    exiting to userspace.
8463
8464 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets
8465                                    CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if
8466                                    IA32_MISC_ENABLE[bit 18] (MWAIT) is set.
8467                                    Additionally, when this quirk is disabled,
8468                                    KVM clears CPUID.01H:ECX[bit 3] if
8469                                    IA32_MISC_ENABLE[bit 18] is cleared.
8470
8471 KVM_X86_QUIRK_FIX_HYPERCALL_INSN   By default, KVM rewrites guest
8472                                    VMMCALL/VMCALL instructions to match the
8473                                    vendor's hypercall instruction for the
8474                                    system. When this quirk is disabled, KVM
8475                                    will no longer rewrite invalid guest
8476                                    hypercall instructions. Executing the
8477                                    incorrect hypercall instruction will
8478                                    generate a #UD within the guest.
8479
8480KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if
8481                                    they are intercepted) as NOPs regardless of
8482                                    whether or not MONITOR/MWAIT are supported
8483                                    according to guest CPUID.  When this quirk
8484                                    is disabled and KVM_X86_DISABLE_EXITS_MWAIT
8485                                    is not set (MONITOR/MWAIT are intercepted),
8486                                    KVM will inject a #UD on MONITOR/MWAIT if
8487                                    they're unsupported per guest CPUID.  Note,
8488                                    KVM will modify MONITOR/MWAIT support in
8489                                    guest CPUID on writes to MISC_ENABLE if
8490                                    KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is
8491                                    disabled.
8492
8493KVM_X86_QUIRK_SLOT_ZAP_ALL          By default, for KVM_X86_DEFAULT_VM VMs, KVM
8494                                    invalidates all SPTEs in all memslots and
8495                                    address spaces when a memslot is deleted or
8496                                    moved.  When this quirk is disabled (or the
8497                                    VM type isn't KVM_X86_DEFAULT_VM), KVM only
8498                                    ensures the backing memory of the deleted
8499                                    or moved memslot isn't reachable, i.e KVM
8500                                    _may_ invalidate only SPTEs related to the
8501                                    memslot.
8502
8503KVM_X86_QUIRK_STUFF_FEATURE_MSRS    By default, at vCPU creation, KVM sets the
8504                                    vCPU's MSR_IA32_PERF_CAPABILITIES (0x345),
8505                                    MSR_IA32_ARCH_CAPABILITIES (0x10a),
8506                                    MSR_PLATFORM_INFO (0xce), and all VMX MSRs
8507                                    (0x480..0x492) to the maximal capabilities
8508                                    supported by KVM.  KVM also sets
8509                                    MSR_IA32_UCODE_REV (0x8b) to an arbitrary
8510                                    value (which is different for Intel vs.
8511                                    AMD).  Lastly, when guest CPUID is set (by
8512                                    userspace), KVM modifies select VMX MSR
8513                                    fields to force consistency between guest
8514                                    CPUID and L2's effective ISA.  When this
8515                                    quirk is disabled, KVM zeroes the vCPU's MSR
8516                                    values (with two exceptions, see below),
8517                                    i.e. treats the feature MSRs like CPUID
8518                                    leaves and gives userspace full control of
8519                                    the vCPU model definition.  This quirk does
8520                                    not affect VMX MSRs CR0/CR4_FIXED1 (0x487
8521                                    and 0x489), as KVM does now allow them to
8522                                    be set by userspace (KVM sets them based on
8523                                    guest CPUID, for safety purposes).
8524
8525KVM_X86_QUIRK_IGNORE_GUEST_PAT      By default, on Intel platforms, KVM ignores
8526                                    guest PAT and forces the effective memory
8527                                    type to WB in EPT.  The quirk is not available
8528                                    on Intel platforms which are incapable of
8529                                    safely honoring guest PAT (i.e., without CPU
8530                                    self-snoop, KVM always ignores guest PAT and
8531                                    forces effective memory type to WB).  It is
8532                                    also ignored on AMD platforms or, on Intel,
8533                                    when a VM has non-coherent DMA devices
8534                                    assigned; KVM always honors guest PAT in
8535                                    such case. The quirk is needed to avoid
8536                                    slowdowns on certain Intel Xeon platforms
8537                                    (e.g. ICX, SPR) where self-snoop feature is
8538                                    supported but UC is slow enough to cause
8539                                    issues with some older guests that use
8540                                    UC instead of WC to map the video RAM.
8541                                    Userspace can disable the quirk to honor
8542                                    guest PAT if it knows that there is no such
8543                                    guest software, for example if it does not
8544                                    expose a bochs graphics device (which is
8545                                    known to have had a buggy driver).
8546=================================== ============================================
8547
85487.32 KVM_CAP_MAX_VCPU_ID
8549------------------------
8550
8551:Architectures: x86
8552:Target: VM
8553:Parameters: args[0] - maximum APIC ID value set for current VM
8554:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS
8555          supported in KVM or if it has been set.
8556
8557This capability allows userspace to specify maximum possible APIC ID
8558assigned for current VM session prior to the creation of vCPUs, saving
8559memory for data structures indexed by the APIC ID.  Userspace is able
8560to calculate the limit to APIC ID values from designated
8561CPU topology.
8562
8563The value can be changed only until KVM_ENABLE_CAP is set to a nonzero
8564value or until a vCPU is created.  Upon creation of the first vCPU,
8565if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM
8566uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as
8567the maximum APIC ID.
8568
85697.33 KVM_CAP_X86_NOTIFY_VMEXIT
8570------------------------------
8571
8572:Architectures: x86
8573:Target: VM
8574:Parameters: args[0] is the value of notify window as well as some flags
8575:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify
8576          VM exit is unsupported.
8577
8578Bits 63:32 of args[0] are used for notify window.
8579Bits 31:0 of args[0] are for some flags. Valid bits are::
8580
8581  #define KVM_X86_NOTIFY_VMEXIT_ENABLED    (1 << 0)
8582  #define KVM_X86_NOTIFY_VMEXIT_USER       (1 << 1)
8583
8584This capability allows userspace to configure the notify VM exit on/off
8585in per-VM scope during VM creation. Notify VM exit is disabled by default.
8586When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will
8587enable this feature with the notify window provided, which will generate
8588a VM exit if no event window occurs in VM non-root mode for a specified of
8589time (notify window).
8590
8591If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen,
8592KVM would exit to userspace for handling.
8593
8594This capability is aimed to mitigate the threat that malicious VMs can
8595cause CPU stuck (due to event windows don't open up) and make the CPU
8596unavailable to host or other VMs.
8597
85987.35 KVM_CAP_X86_APIC_BUS_CYCLES_NS
8599-----------------------------------
8600
8601:Architectures: x86
8602:Target: VM
8603:Parameters: args[0] is the desired APIC bus clock rate, in nanoseconds
8604:Returns: 0 on success, -EINVAL if args[0] contains an invalid value for the
8605          frequency or if any vCPUs have been created, -ENXIO if a virtual
8606          local APIC has not been created using KVM_CREATE_IRQCHIP.
8607
8608This capability sets the VM's APIC bus clock frequency, used by KVM's in-kernel
8609virtual APIC when emulating APIC timers.  KVM's default value can be retrieved
8610by KVM_CHECK_EXTENSION.
8611
8612Note: Userspace is responsible for correctly configuring CPUID 0x15, a.k.a. the
8613core crystal clock frequency, if a non-zero CPUID 0x15 is exposed to the guest.
8614
86157.36 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8616----------------------------------------------------------
8617
8618:Architectures: x86, arm64, riscv
8619:Type: vm
8620:Parameters: args[0] - size of the dirty log ring
8621
8622KVM is capable of tracking dirty memory using ring buffers that are
8623mmapped into userspace; there is one dirty ring per vcpu.
8624
8625The dirty ring is available to userspace as an array of
8626``struct kvm_dirty_gfn``.  Each dirty entry is defined as::
8627
8628  struct kvm_dirty_gfn {
8629          __u32 flags;
8630          __u32 slot; /* as_id | slot_id */
8631          __u64 offset;
8632  };
8633
8634The following values are defined for the flags field to define the
8635current state of the entry::
8636
8637  #define KVM_DIRTY_GFN_F_DIRTY           BIT(0)
8638  #define KVM_DIRTY_GFN_F_RESET           BIT(1)
8639  #define KVM_DIRTY_GFN_F_MASK            0x3
8640
8641Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
8642ioctl to enable this capability for the new guest and set the size of
8643the rings.  Enabling the capability is only allowed before creating any
8644vCPU, and the size of the ring must be a power of two.  The larger the
8645ring buffer, the less likely the ring is full and the VM is forced to
8646exit to userspace. The optimal size depends on the workload, but it is
8647recommended that it be at least 64 KiB (4096 entries).
8648
8649Just like for dirty page bitmaps, the buffer tracks writes to
8650all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
8651set in KVM_SET_USER_MEMORY_REGION.  Once a memory region is registered
8652with the flag set, userspace can start harvesting dirty pages from the
8653ring buffer.
8654
8655An entry in the ring buffer can be unused (flag bits ``00``),
8656dirty (flag bits ``01``) or harvested (flag bits ``1X``).  The
8657state machine for the entry is as follows::
8658
8659          dirtied         harvested        reset
8660     00 -----------> 01 -------------> 1X -------+
8661      ^                                          |
8662      |                                          |
8663      +------------------------------------------+
8664
8665To harvest the dirty pages, userspace accesses the mmapped ring buffer
8666to read the dirty GFNs.  If the flags has the DIRTY bit set (at this stage
8667the RESET bit must be cleared), then it means this GFN is a dirty GFN.
8668The userspace should harvest this GFN and mark the flags from state
8669``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
8670to show that this GFN is harvested and waiting for a reset), and move
8671on to the next GFN.  The userspace should continue to do this until the
8672flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
8673all the dirty GFNs that were available.
8674
8675Note that on weakly ordered architectures, userspace accesses to the
8676ring buffer (and more specifically the 'flags' field) must be ordered,
8677using load-acquire/store-release accessors when available, or any
8678other memory barrier that will ensure this ordering.
8679
8680It's not necessary for userspace to harvest the all dirty GFNs at once.
8681However it must collect the dirty GFNs in sequence, i.e., the userspace
8682program cannot skip one dirty GFN to collect the one next to it.
8683
8684After processing one or more entries in the ring buffer, userspace
8685calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
8686it, so that the kernel will reprotect those collected GFNs.
8687Therefore, the ioctl must be called *before* reading the content of
8688the dirty pages.
8689
8690The dirty ring can get full.  When it happens, the KVM_RUN of the
8691vcpu will return with exit reason KVM_EXIT_DIRTY_RING_FULL.
8692
8693The dirty ring interface has a major difference comparing to the
8694KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
8695userspace, it's still possible that the kernel has not yet flushed the
8696processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
8697flushing is done by the KVM_GET_DIRTY_LOG ioctl).  To achieve that, one
8698needs to kick the vcpu out of KVM_RUN using a signal.  The resulting
8699vmexit ensures that all dirty GFNs are flushed to the dirty rings.
8700
8701NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that
8702should be exposed by weakly ordered architecture, in order to indicate
8703the additional memory ordering requirements imposed on userspace when
8704reading the state of an entry and mutating it from DIRTY to HARVESTED.
8705Architecture with TSO-like ordering (such as x86) are allowed to
8706expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8707to userspace.
8708
8709After enabling the dirty rings, the userspace needs to detect the
8710capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the
8711ring structures can be backed by per-slot bitmaps. With this capability
8712advertised, it means the architecture can dirty guest pages without
8713vcpu/ring context, so that some of the dirty information will still be
8714maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP
8715can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8716hasn't been enabled, or any memslot has been existing.
8717
8718Note that the bitmap here is only a backup of the ring structure. The
8719use of the ring and bitmap combination is only beneficial if there is
8720only a very small amount of memory that is dirtied out of vcpu/ring
8721context. Otherwise, the stand-alone per-slot bitmap mechanism needs to
8722be considered.
8723
8724To collect dirty bits in the backup bitmap, userspace can use the same
8725KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all
8726the generation of the dirty bits is done in a single pass. Collecting
8727the dirty bitmap should be the very last thing that the VMM does before
8728considering the state as complete. VMM needs to ensure that the dirty
8729state is final and avoid missing dirty pages from another ioctl ordered
8730after the bitmap collection.
8731
8732NOTE: Multiple examples of using the backup bitmap: (1) save vgic/its
8733tables through command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} on
8734KVM device "kvm-arm-vgic-its". (2) restore vgic/its tables through
8735command KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_RESTORE_TABLES} on KVM device
8736"kvm-arm-vgic-its". VGICv3 LPI pending status is restored. (3) save
8737vgic3 pending table through KVM_DEV_ARM_VGIC_{GRP_CTRL, SAVE_PENDING_TABLES}
8738command on KVM device "kvm-arm-vgic-v3".
8739
87407.37 KVM_CAP_PMU_CAPABILITY
8741---------------------------
8742
8743:Architectures: x86
8744:Type: vm
8745:Parameters: arg[0] is bitmask of PMU virtualization capabilities.
8746:Returns: 0 on success, -EINVAL when arg[0] contains invalid bits
8747
8748This capability alters PMU virtualization in KVM.
8749
8750Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
8751PMU virtualization capabilities that can be adjusted on a VM.
8752
8753The argument to KVM_ENABLE_CAP is also a bitmask and selects specific
8754PMU virtualization capabilities to be applied to the VM.  This can
8755only be invoked on a VM prior to the creation of VCPUs.
8756
8757At this time, KVM_PMU_CAP_DISABLE is the only capability.  Setting
8758this capability will disable PMU virtualization for that VM.  Usermode
8759should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
8760
87617.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
8762-------------------------------------
8763
8764:Architectures: x86
8765:Type: vm
8766:Parameters: arg[0] must be 0.
8767:Returns: 0 on success, -EPERM if the userspace process does not
8768          have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been
8769          created.
8770
8771This capability disables the NX huge pages mitigation for iTLB MULTIHIT.
8772
8773The capability has no effect if the nx_huge_pages module parameter is not set.
8774
8775This capability may only be set before any vCPUs are created.
8776
87777.39 KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE
8778---------------------------------------
8779
8780:Architectures: arm64
8781:Type: vm
8782:Parameters: arg[0] is the new split chunk size.
8783:Returns: 0 on success, -EINVAL if any memslot was already created.
8784
8785This capability sets the chunk size used in Eager Page Splitting.
8786
8787Eager Page Splitting improves the performance of dirty-logging (used
8788in live migrations) when guest memory is backed by huge-pages.  It
8789avoids splitting huge-pages (into PAGE_SIZE pages) on fault, by doing
8790it eagerly when enabling dirty logging (with the
8791KVM_MEM_LOG_DIRTY_PAGES flag for a memory region), or when using
8792KVM_CLEAR_DIRTY_LOG.
8793
8794The chunk size specifies how many pages to break at a time, using a
8795single allocation for each chunk. Bigger the chunk size, more pages
8796need to be allocated ahead of time.
8797
8798The chunk size needs to be a valid block size. The list of acceptable
8799block sizes is exposed in KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES as a
880064-bit bitmap (each bit describing a block size). The default value is
88010, to disable the eager page splitting.
8802
88037.40 KVM_CAP_EXIT_HYPERCALL
8804---------------------------
8805
8806:Architectures: x86
8807:Type: vm
8808
8809This capability, if enabled, will cause KVM to exit to userspace
8810with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
8811
8812Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
8813of hypercalls that can be configured to exit to userspace.
8814Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
8815
8816The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
8817of the result of KVM_CHECK_EXTENSION.  KVM will forward to userspace
8818the hypercalls whose corresponding bit is in the argument, and return
8819ENOSYS for the others.
8820
88217.41 KVM_CAP_ARM_SYSTEM_SUSPEND
8822-------------------------------
8823
8824:Architectures: arm64
8825:Type: vm
8826
8827When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of
8828type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request.
8829
88307.42 KVM_CAP_ARM_WRITABLE_IMP_ID_REGS
8831-------------------------------------
8832
8833:Architectures: arm64
8834:Target: VM
8835:Parameters: None
8836:Returns: 0 on success, -EINVAL if vCPUs have been created before enabling this
8837          capability.
8838
8839This capability changes the behavior of the registers that identify a PE
8840implementation of the Arm architecture: MIDR_EL1, REVIDR_EL1, and AIDR_EL1.
8841By default, these registers are visible to userspace but treated as invariant.
8842
8843When this capability is enabled, KVM allows userspace to change the
8844aforementioned registers before the first KVM_RUN. These registers are VM
8845scoped, meaning that the same set of values are presented on all vCPUs in a
8846given VM.
8847
88487.43 KVM_CAP_RISCV_MP_STATE_RESET
8849---------------------------------
8850
8851:Architectures: riscv
8852:Type: VM
8853:Parameters: None
8854:Returns: 0 on success, -EINVAL if arg[0] is not zero
8855
8856When this capability is enabled, KVM resets the VCPU when setting
8857MP_STATE_INIT_RECEIVED through IOCTL.  The original MP_STATE is preserved.
8858
88597.44 KVM_CAP_ARM_CACHEABLE_PFNMAP_SUPPORTED
8860-------------------------------------------
8861
8862:Architectures: arm64
8863:Target: VM
8864:Parameters: None
8865
8866This capability indicate to the userspace whether a PFNMAP memory region
8867can be safely mapped as cacheable. This relies on the presence of
8868force write back (FWB) feature support on the hardware.
8869
88707.45 KVM_CAP_ARM_SEA_TO_USER
8871----------------------------
8872
8873:Architecture: arm64
8874:Target: VM
8875:Parameters: none
8876:Returns: 0 on success, -EINVAL if unsupported.
8877
8878When this capability is enabled, KVM may exit to userspace for SEAs taken to
8879EL2 resulting from a guest access. See ``KVM_EXIT_ARM_SEA`` for more
8880information.
8881
88827.46 KVM_CAP_S390_USER_OPEREXEC
8883-------------------------------
8884
8885:Architectures: s390
8886:Parameters: none
8887
8888When this capability is enabled KVM forwards all operation exceptions
8889that it doesn't handle itself to user space. This also includes the
88900x0000 instructions managed by KVM_CAP_S390_USER_INSTR0. This is
8891helpful if user space wants to emulate instructions which are not
8892(yet) implemented in hardware.
8893
8894This capability can be enabled dynamically even if VCPUs were already
8895created and are running.
8896
88978. Other capabilities.
8898======================
8899
8900This section lists capabilities that give information about other
8901features of the KVM implementation.
8902
89038.1 KVM_CAP_PPC_HWRNG
8904---------------------
8905
8906:Architectures: ppc
8907
8908This capability, if KVM_CHECK_EXTENSION indicates that it is
8909available, means that the kernel has an implementation of the
8910H_RANDOM hypercall backed by a hardware random-number generator.
8911If present, the kernel H_RANDOM handler can be enabled for guest use
8912with the KVM_CAP_PPC_ENABLE_HCALL capability.
8913
89148.3 KVM_CAP_PPC_MMU_RADIX
8915-------------------------
8916
8917:Architectures: ppc
8918
8919This capability, if KVM_CHECK_EXTENSION indicates that it is
8920available, means that the kernel can support guests using the
8921radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
8922processor).
8923
89248.4 KVM_CAP_PPC_MMU_HASH_V3
8925---------------------------
8926
8927:Architectures: ppc
8928
8929This capability, if KVM_CHECK_EXTENSION indicates that it is
8930available, means that the kernel can support guests using the
8931hashed page table MMU defined in Power ISA V3.00 (as implemented in
8932the POWER9 processor), including in-memory segment tables.
8933
89348.5 KVM_CAP_MIPS_VZ
8935-------------------
8936
8937:Architectures: mips
8938
8939This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
8940it is available, means that full hardware assisted virtualization capabilities
8941of the hardware are available for use through KVM. An appropriate
8942KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
8943utilises it.
8944
8945If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
8946available, it means that the VM is using full hardware assisted virtualization
8947capabilities of the hardware. This is useful to check after creating a VM with
8948KVM_VM_MIPS_DEFAULT.
8949
8950The value returned by KVM_CHECK_EXTENSION should be compared against known
8951values (see below). All other values are reserved. This is to allow for the
8952possibility of other hardware assisted virtualization implementations which
8953may be incompatible with the MIPS VZ ASE.
8954
8955==  ==========================================================================
8956 0  The trap & emulate implementation is in use to run guest code in user
8957    mode. Guest virtual memory segments are rearranged to fit the guest in the
8958    user mode address space.
8959
8960 1  The MIPS VZ ASE is in use, providing full hardware assisted
8961    virtualization, including standard guest virtual memory segments.
8962==  ==========================================================================
8963
89648.7 KVM_CAP_MIPS_64BIT
8965----------------------
8966
8967:Architectures: mips
8968
8969This capability indicates the supported architecture type of the guest, i.e. the
8970supported register and address width.
8971
8972The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
8973kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
8974be checked specifically against known values (see below). All other values are
8975reserved.
8976
8977==  ========================================================================
8978 0  MIPS32 or microMIPS32.
8979    Both registers and addresses are 32-bits wide.
8980    It will only be possible to run 32-bit guest code.
8981
8982 1  MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
8983    Registers are 64-bits wide, but addresses are 32-bits wide.
8984    64-bit guest code may run but cannot access MIPS64 memory segments.
8985    It will also be possible to run 32-bit guest code.
8986
8987 2  MIPS64 or microMIPS64 with access to all address segments.
8988    Both registers and addresses are 64-bits wide.
8989    It will be possible to run 64-bit or 32-bit guest code.
8990==  ========================================================================
8991
89928.9 KVM_CAP_ARM_USER_IRQ
8993------------------------
8994
8995:Architectures: arm64
8996
8997This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
8998that if userspace creates a VM without an in-kernel interrupt controller, it
8999will be notified of changes to the output level of in-kernel emulated devices,
9000which can generate virtual interrupts, presented to the VM.
9001For such VMs, on every return to userspace, the kernel
9002updates the vcpu's run->s.regs.device_irq_level field to represent the actual
9003output level of the device.
9004
9005Whenever kvm detects a change in the device output level, kvm guarantees at
9006least one return to userspace before running the VM.  This exit could either
9007be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
9008userspace can always sample the device output level and re-compute the state of
9009the userspace interrupt controller.  Userspace should always check the state
9010of run->s.regs.device_irq_level on every kvm exit.
9011The value in run->s.regs.device_irq_level can represent both level and edge
9012triggered interrupt signals, depending on the device.  Edge triggered interrupt
9013signals will exit to userspace with the bit in run->s.regs.device_irq_level
9014set exactly once per edge signal.
9015
9016The field run->s.regs.device_irq_level is available independent of
9017run->kvm_valid_regs or run->kvm_dirty_regs bits.
9018
9019If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
9020number larger than 0 indicating the version of this capability is implemented
9021and thereby which bits in run->s.regs.device_irq_level can signal values.
9022
9023Currently the following bits are defined for the device_irq_level bitmap::
9024
9025  KVM_CAP_ARM_USER_IRQ >= 1:
9026
9027    KVM_ARM_DEV_EL1_VTIMER -  EL1 virtual timer
9028    KVM_ARM_DEV_EL1_PTIMER -  EL1 physical timer
9029    KVM_ARM_DEV_PMU        -  ARM PMU overflow interrupt signal
9030
9031Future versions of kvm may implement additional events. These will get
9032indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
9033listed above.
9034
90358.10 KVM_CAP_PPC_SMT_POSSIBLE
9036-----------------------------
9037
9038:Architectures: ppc
9039
9040Querying this capability returns a bitmap indicating the possible
9041virtual SMT modes that can be set using KVM_CAP_PPC_SMT.  If bit N
9042(counting from the right) is set, then a virtual SMT mode of 2^N is
9043available.
9044
90458.12 KVM_CAP_HYPERV_VP_INDEX
9046----------------------------
9047
9048:Architectures: x86
9049
9050This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr.  Its
9051value is used to denote the target vcpu for a SynIC interrupt.  For
9052compatibility, KVM initializes this msr to KVM's internal vcpu index.  When this
9053capability is absent, userspace can still query this msr's value.
9054
90558.13 KVM_CAP_S390_AIS_MIGRATION
9056-------------------------------
9057
9058:Architectures: s390
9059
9060This capability indicates if the flic device will be able to get/set the
9061AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
9062to discover this without having to create a flic device.
9063
90648.14 KVM_CAP_S390_PSW
9065---------------------
9066
9067:Architectures: s390
9068
9069This capability indicates that the PSW is exposed via the kvm_run structure.
9070
90718.15 KVM_CAP_S390_GMAP
9072----------------------
9073
9074:Architectures: s390
9075
9076This capability indicates that the user space memory used as guest mapping can
9077be anywhere in the user memory address space, as long as the memory slots are
9078aligned and sized to a segment (1MB) boundary.
9079
90808.16 KVM_CAP_S390_COW
9081---------------------
9082
9083:Architectures: s390
9084
9085This capability indicates that the user space memory used as guest mapping can
9086use copy-on-write semantics as well as dirty pages tracking via read-only page
9087tables.
9088
90898.17 KVM_CAP_S390_BPB
9090---------------------
9091
9092:Architectures: s390
9093
9094This capability indicates that kvm will implement the interfaces to handle
9095reset, migration and nested KVM for branch prediction blocking. The stfle
9096facility 82 should not be provided to the guest without this capability.
9097
90988.18 KVM_CAP_HYPERV_TLBFLUSH
9099----------------------------
9100
9101:Architectures: x86
9102
9103This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
9104hypercalls:
9105HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
9106HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
9107
91088.19 KVM_CAP_ARM_INJECT_SERROR_ESR
9109----------------------------------
9110
9111:Architectures: arm64
9112
9113This capability indicates that userspace can specify (via the
9114KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
9115takes a virtual SError interrupt exception.
9116If KVM advertises this capability, userspace can only specify the ISS field for
9117the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
9118CPU when the exception is taken. If this virtual SError is taken to EL1 using
9119AArch64, this value will be reported in the ISS field of ESR_ELx.
9120
9121See KVM_CAP_VCPU_EVENTS for more details.
9122
91238.20 KVM_CAP_HYPERV_SEND_IPI
9124----------------------------
9125
9126:Architectures: x86
9127
9128This capability indicates that KVM supports paravirtualized Hyper-V IPI send
9129hypercalls:
9130HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
9131
91328.22 KVM_CAP_S390_VCPU_RESETS
9133-----------------------------
9134
9135:Architectures: s390
9136
9137This capability indicates that the KVM_S390_NORMAL_RESET and
9138KVM_S390_CLEAR_RESET ioctls are available.
9139
91408.23 KVM_CAP_S390_PROTECTED
9141---------------------------
9142
9143:Architectures: s390
9144
9145This capability indicates that the Ultravisor has been initialized and
9146KVM can therefore start protected VMs.
9147This capability governs the KVM_S390_PV_COMMAND ioctl and the
9148KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
9149guests when the state change is invalid.
9150
91518.24 KVM_CAP_STEAL_TIME
9152-----------------------
9153
9154:Architectures: arm64, x86
9155
9156This capability indicates that KVM supports steal time accounting.
9157When steal time accounting is supported it may be enabled with
9158architecture-specific interfaces.  This capability and the architecture-
9159specific interfaces must be consistent, i.e. if one says the feature
9160is supported, than the other should as well and vice versa.  For arm64
9161see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
9162For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME".
9163
91648.25 KVM_CAP_S390_DIAG318
9165-------------------------
9166
9167:Architectures: s390
9168
9169This capability enables a guest to set information about its control program
9170(i.e. guest kernel type and version). The information is helpful during
9171system/firmware service events, providing additional data about the guest
9172environments running on the machine.
9173
9174The information is associated with the DIAGNOSE 0x318 instruction, which sets
9175an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
9176a 7-byte Control Program Version Code (CPVC). The CPNC determines what
9177environment the control program is running in (e.g. Linux, z/VM...), and the
9178CPVC is used for information specific to OS (e.g. Linux version, Linux
9179distribution...)
9180
9181If this capability is available, then the CPNC and CPVC can be synchronized
9182between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
9183
91848.26 KVM_CAP_X86_USER_SPACE_MSR
9185-------------------------------
9186
9187:Architectures: x86
9188
9189This capability indicates that KVM supports deflection of MSR reads and
9190writes to user space. It can be enabled on a VM level. If enabled, MSR
9191accesses that would usually trigger a #GP by KVM into the guest will
9192instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
9193KVM_EXIT_X86_WRMSR exit notifications.
9194
91958.27 KVM_CAP_X86_MSR_FILTER
9196---------------------------
9197
9198:Architectures: x86
9199
9200This capability indicates that KVM supports that accesses to user defined MSRs
9201may be rejected. With this capability exposed, KVM exports new VM ioctl
9202KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
9203ranges that KVM should deny access to.
9204
9205In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
9206trap and emulate MSRs that are outside of the scope of KVM as well as
9207limit the attack surface on KVM's MSR emulation code.
9208
92098.30 KVM_CAP_XEN_HVM
9210--------------------
9211
9212:Architectures: x86
9213
9214This capability indicates the features that Xen supports for hosting Xen
9215PVHVM guests. Valid flags are::
9216
9217  #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR		(1 << 0)
9218  #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL		(1 << 1)
9219  #define KVM_XEN_HVM_CONFIG_SHARED_INFO		(1 << 2)
9220  #define KVM_XEN_HVM_CONFIG_RUNSTATE			(1 << 3)
9221  #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL		(1 << 4)
9222  #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND		(1 << 5)
9223  #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG	(1 << 6)
9224  #define KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE	(1 << 7)
9225
9226The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
9227ioctl is available, for the guest to set its hypercall page.
9228
9229If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
9230provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
9231contents, to request that KVM generate hypercall page content automatically
9232and also enable interception of guest hypercalls with KVM_EXIT_XEN.
9233
9234The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
9235KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
9236KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
9237for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
9238vcpu_info is set.
9239
9240The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
9241features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
9242supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
9243
9244The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries
9245of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority
9246field set to indicate 2 level event channel delivery.
9247
9248The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports
9249injecting event channel events directly into the guest with the
9250KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the
9251KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the
9252KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes.
9253related to event channel delivery, timers, and the XENVER_version
9254interception.
9255
9256The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports
9257the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR
9258and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the
9259XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during
9260updates of the runstate information. Note that versions of KVM which support
9261the RUNSTATE feature above, but not the RUNSTATE_UPDATE_FLAG feature, will
9262always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure,
9263which is perhaps counterintuitive. When this flag is advertised, KVM will
9264behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless
9265specifically enabled (by the guest making the hypercall, causing the VMM
9266to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute).
9267
9268The KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag indicates that KVM supports
9269clearing the PVCLOCK_TSC_STABLE_BIT flag in Xen pvclock sources. This will be
9270done when the KVM_CAP_XEN_HVM ioctl sets the
9271KVM_XEN_HVM_CONFIG_PVCLOCK_TSC_UNSTABLE flag.
9272
92738.31 KVM_CAP_SPAPR_MULTITCE
9274---------------------------
9275
9276:Architectures: ppc
9277:Type: vm
9278
9279This capability means the kernel is capable of handling hypercalls
9280H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
9281space. This significantly accelerates DMA operations for PPC KVM guests.
9282User space should expect that its handlers for these hypercalls
9283are not going to be called if user space previously registered LIOBN
9284in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
9285
9286In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
9287user space might have to advertise it for the guest. For example,
9288IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
9289present in the "ibm,hypertas-functions" device-tree property.
9290
9291The hypercalls mentioned above may or may not be processed successfully
9292in the kernel based fast path. If they can not be handled by the kernel,
9293they will get passed on to user space. So user space still has to have
9294an implementation for these despite the in kernel acceleration.
9295
9296This capability is always enabled.
9297
92988.32 KVM_CAP_PTP_KVM
9299--------------------
9300
9301:Architectures: arm64
9302
9303This capability indicates that the KVM virtual PTP service is
9304supported in the host. A VMM can check whether the service is
9305available to the guest on migration.
9306
93078.37 KVM_CAP_S390_PROTECTED_DUMP
9308--------------------------------
9309
9310:Architectures: s390
9311:Type: vm
9312
9313This capability indicates that KVM and the Ultravisor support dumping
9314PV guests. The `KVM_PV_DUMP` command is available for the
9315`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides
9316dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is
9317available and supports the `KVM_PV_DUMP_CPU` subcommand.
9318
93198.39 KVM_CAP_S390_CPU_TOPOLOGY
9320------------------------------
9321
9322:Architectures: s390
9323:Type: vm
9324
9325This capability indicates that KVM will provide the S390 CPU Topology
9326facility which consist of the interpretation of the PTF instruction for
9327the function code 2 along with interception and forwarding of both the
9328PTF instruction with function codes 0 or 1 and the STSI(15,1,x)
9329instruction to the userland hypervisor.
9330
9331The stfle facility 11, CPU Topology facility, should not be indicated
9332to the guest without this capability.
9333
9334When this capability is present, KVM provides a new attribute group
9335on vm fd, KVM_S390_VM_CPU_TOPOLOGY.
9336This new attribute allows to get, set or clear the Modified Change
9337Topology Report (MTCR) bit of the SCA through the kvm_device_attr
9338structure.
9339
9340When getting the Modified Change Topology Report value, the attr->addr
9341must point to a byte where the value will be stored or retrieved from.
9342
93438.41 KVM_CAP_VM_TYPES
9344---------------------
9345
9346:Architectures: x86
9347:Type: system ioctl
9348
9349This capability returns a bitmap of support VM types.  The 1-setting of bit @n
9350means the VM type with value @n is supported.  Possible values of @n are::
9351
9352  #define KVM_X86_DEFAULT_VM	0
9353  #define KVM_X86_SW_PROTECTED_VM	1
9354  #define KVM_X86_SEV_VM	2
9355  #define KVM_X86_SEV_ES_VM	3
9356
9357Note, KVM_X86_SW_PROTECTED_VM is currently only for development and testing.
9358Do not use KVM_X86_SW_PROTECTED_VM for "real" VMs, and especially not in
9359production.  The behavior and effective ABI for software-protected VMs is
9360unstable.
9361
93628.42 KVM_CAP_PPC_RPT_INVALIDATE
9363-------------------------------
9364
9365:Architectures: ppc
9366
9367This capability indicates that the kernel is capable of handling
9368H_RPT_INVALIDATE hcall.
9369
9370In order to enable the use of H_RPT_INVALIDATE in the guest,
9371user space might have to advertise it for the guest. For example,
9372IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
9373present in the "ibm,hypertas-functions" device-tree property.
9374
9375This capability is enabled for hypervisors on platforms like POWER9
9376that support radix MMU.
9377
93788.43 KVM_CAP_PPC_AIL_MODE_3
9379---------------------------
9380
9381:Architectures: ppc
9382
9383This capability indicates that the kernel supports the mode 3 setting for the
9384"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location"
9385resource that is controlled with the H_SET_MODE hypercall.
9386
9387This capability allows a guest kernel to use a better-performance mode for
9388handling interrupts and system calls.
9389
93908.44 KVM_CAP_MEMORY_FAULT_INFO
9391------------------------------
9392
9393:Architectures: x86
9394
9395The presence of this capability indicates that KVM_RUN will fill
9396kvm_run.memory_fault if KVM cannot resolve a guest page fault VM-Exit, e.g. if
9397there is a valid memslot but no backing VMA for the corresponding host virtual
9398address.
9399
9400The information in kvm_run.memory_fault is valid if and only if KVM_RUN returns
9401an error with errno=EFAULT or errno=EHWPOISON *and* kvm_run.exit_reason is set
9402to KVM_EXIT_MEMORY_FAULT.
9403
9404Note: Userspaces which attempt to resolve memory faults so that they can retry
9405KVM_RUN are encouraged to guard against repeatedly receiving the same
9406error/annotated fault.
9407
9408See KVM_EXIT_MEMORY_FAULT for more information.
9409
94108.45 KVM_CAP_X86_GUEST_MODE
9411---------------------------
9412
9413:Architectures: x86
9414
9415The presence of this capability indicates that KVM_RUN will update the
9416KVM_RUN_X86_GUEST_MODE bit in kvm_run.flags to indicate whether the
9417vCPU was executing nested guest code when it exited.
9418
94198.46 KVM_CAP_S390_KEYOP
9420-----------------------
9421
9422:Architectures: s390
9423
9424The presence of this capability indicates that the KVM_S390_KEYOP ioctl is
9425available.
9426
9427KVM exits with the register state of either the L1 or L2 guest
9428depending on which executed at the time of an exit. Userspace must
9429take care to differentiate between these cases.
9430
94319. Known KVM API problems
9432=========================
9433
9434In some cases, KVM's API has some inconsistencies or common pitfalls
9435that userspace need to be aware of.  This section details some of
9436these issues.
9437
9438Most of them are architecture specific, so the section is split by
9439architecture.
9440
94419.1. x86
9442--------
9443
9444``KVM_GET_SUPPORTED_CPUID`` issues
9445^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9446
9447In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible
9448to take its result and pass it directly to ``KVM_SET_CPUID2``.  This section
9449documents some cases in which that requires some care.
9450
9451Local APIC features
9452~~~~~~~~~~~~~~~~~~~
9453
9454CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``,
9455but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or
9456``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of
9457the local APIC.
9458
9459The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature.
9460
9461On older versions of Linux, CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by
9462``KVM_GET_SUPPORTED_CPUID``, but it can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER``
9463is present and the kernel has enabled in-kernel emulation of the local APIC.
9464On newer versions, ``KVM_GET_SUPPORTED_CPUID`` does report the bit as available.
9465
9466CPU topology
9467~~~~~~~~~~~~
9468
9469Several CPUID values include topology information for the host CPU:
94700x0b and 0x1f for Intel systems, 0x8000001e for AMD systems.  Different
9471versions of KVM return different values for this information and userspace
9472should not rely on it.  Currently they return all zeroes.
9473
9474If userspace wishes to set up a guest topology, it should be careful that
9475the values of these three leaves differ for each CPU.  In particular,
9476the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX
9477for 0x8000001e; the latter also encodes the core id and node id in bits
94787:0 of EBX and ECX respectively.
9479
9480Obsolete ioctls and capabilities
9481^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9482
9483KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually
9484available.  Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if
9485available.
9486
9487Ordering of KVM_GET_*/KVM_SET_* ioctls
9488^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9489
9490TBD
9491