xref: /openbmc/linux/Documentation/virt/kvm/api.rst (revision 27ab1c1c)
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 a set of ioctls that are issued to control various aspects
11of a virtual machine.  The ioctls belong to the following classes:
12
13 - System ioctls: These query and set global attributes which affect the
14   whole kvm subsystem.  In addition a system ioctl is used to create
15   virtual machines.
16
17 - VM ioctls: These query and set attributes that affect an entire virtual
18   machine, for example memory layout.  In addition a VM ioctl is used to
19   create virtual cpus (vcpus) and devices.
20
21   VM ioctls must be issued from the same process (address space) that was
22   used to create the VM.
23
24 - vcpu ioctls: These query and set attributes that control the operation
25   of a single virtual cpu.
26
27   vcpu ioctls should be issued from the same thread that was used to create
28   the vcpu, except for asynchronous vcpu ioctl that are marked as such in
29   the documentation.  Otherwise, the first ioctl after switching threads
30   could see a performance impact.
31
32 - device ioctls: These query and set attributes that control the operation
33   of a single device.
34
35   device ioctls must be issued from the same process (address space) that
36   was used to create the VM.
37
382. File descriptors
39===================
40
41The kvm API is centered around file descriptors.  An initial
42open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
43can be used to issue system ioctls.  A KVM_CREATE_VM ioctl on this
44handle will create a VM file descriptor which can be used to issue VM
45ioctls.  A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
46create a virtual cpu or device and return a file descriptor pointing to
47the new resource.  Finally, ioctls on a vcpu or device fd can be used
48to control the vcpu or device.  For vcpus, this includes the important
49task of actually running guest code.
50
51In general file descriptors can be migrated among processes by means
52of fork() and the SCM_RIGHTS facility of unix domain socket.  These
53kinds of tricks are explicitly not supported by kvm.  While they will
54not cause harm to the host, their actual behavior is not guaranteed by
55the API.  See "General description" for details on the ioctl usage
56model that is supported by KVM.
57
58It is important to note that althought VM ioctls may only be issued from
59the process that created the VM, a VM's lifecycle is associated with its
60file descriptor, not its creator (process).  In other words, the VM and
61its resources, *including the associated address space*, are not freed
62until the last reference to the VM's file descriptor has been released.
63For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will
64not be freed until both the parent (original) process and its child have
65put their references to the VM's file descriptor.
66
67Because a VM's resources are not freed until the last reference to its
68file descriptor is released, creating additional references to a VM
69via fork(), dup(), etc... without careful consideration is strongly
70discouraged and may have unwanted side effects, e.g. memory allocated
71by and on behalf of the VM's process may not be freed/unaccounted when
72the VM is shut down.
73
74
753. Extensions
76=============
77
78As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
79incompatible change are allowed.  However, there is an extension
80facility that allows backward-compatible extensions to the API to be
81queried and used.
82
83The extension mechanism is not based on the Linux version number.
84Instead, kvm defines extension identifiers and a facility to query
85whether a particular extension identifier is available.  If it is, a
86set of ioctls is available for application use.
87
88
894. API description
90==================
91
92This section describes ioctls that can be used to control kvm guests.
93For each ioctl, the following information is provided along with a
94description:
95
96  Capability:
97      which KVM extension provides this ioctl.  Can be 'basic',
98      which means that is will be provided by any kernel that supports
99      API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
100      means availability needs to be checked with KVM_CHECK_EXTENSION
101      (see section 4.4), or 'none' which means that while not all kernels
102      support this ioctl, there's no capability bit to check its
103      availability: for kernels that don't support the ioctl,
104      the ioctl returns -ENOTTY.
105
106  Architectures:
107      which instruction set architectures provide this ioctl.
108      x86 includes both i386 and x86_64.
109
110  Type:
111      system, vm, or vcpu.
112
113  Parameters:
114      what parameters are accepted by the ioctl.
115
116  Returns:
117      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
118      are not detailed, but errors with specific meanings are.
119
120
1214.1 KVM_GET_API_VERSION
122-----------------------
123
124:Capability: basic
125:Architectures: all
126:Type: system ioctl
127:Parameters: none
128:Returns: the constant KVM_API_VERSION (=12)
129
130This identifies the API version as the stable kvm API. It is not
131expected that this number will change.  However, Linux 2.6.20 and
1322.6.21 report earlier versions; these are not documented and not
133supported.  Applications should refuse to run if KVM_GET_API_VERSION
134returns a value other than 12.  If this check passes, all ioctls
135described as 'basic' will be available.
136
137
1384.2 KVM_CREATE_VM
139-----------------
140
141:Capability: basic
142:Architectures: all
143:Type: system ioctl
144:Parameters: machine type identifier (KVM_VM_*)
145:Returns: a VM fd that can be used to control the new virtual machine.
146
147The new VM has no virtual cpus and no memory.
148You probably want to use 0 as machine type.
149
150In order to create user controlled virtual machines on S390, check
151KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
152privileged user (CAP_SYS_ADMIN).
153
154To use hardware assisted virtualization on MIPS (VZ ASE) rather than
155the default trap & emulate implementation (which changes the virtual
156memory layout to fit in user mode), check KVM_CAP_MIPS_VZ and use the
157flag KVM_VM_MIPS_VZ.
158
159
160On arm64, the physical address size for a VM (IPA Size limit) is limited
161to 40bits by default. The limit can be configured if the host supports the
162extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
163KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
164identifier, where IPA_Bits is the maximum width of any physical
165address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
166machine type identifier.
167
168e.g, to configure a guest to use 48bit physical address size::
169
170    vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
171
172The requested size (IPA_Bits) must be:
173
174 ==   =========================================================
175  0   Implies default size, 40bits (for backward compatibility)
176  N   Implies N bits, where N is a positive integer such that,
177      32 <= N <= Host_IPA_Limit
178 ==   =========================================================
179
180Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
181is dependent on the CPU capability and the kernel configuration. The limit can
182be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
183ioctl() at run-time.
184
185Please note that configuring the IPA size does not affect the capability
186exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
187size of the address translated by the stage2 level (guest physical to
188host physical address translations).
189
190
1914.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
192----------------------------------------------------------
193
194:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
195:Architectures: x86
196:Type: system ioctl
197:Parameters: struct kvm_msr_list (in/out)
198:Returns: 0 on success; -1 on error
199
200Errors:
201
202  ======     ============================================================
203  EFAULT     the msr index list cannot be read from or written to
204  E2BIG      the msr index list is to be to fit in the array specified by
205             the user.
206  ======     ============================================================
207
208::
209
210  struct kvm_msr_list {
211	__u32 nmsrs; /* number of msrs in entries */
212	__u32 indices[0];
213  };
214
215The user fills in the size of the indices array in nmsrs, and in return
216kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
217indices array with their numbers.
218
219KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported.  The list
220varies by kvm version and host processor, but does not change otherwise.
221
222Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
223not returned in the MSR list, as different vcpus can have a different number
224of banks, as set via the KVM_X86_SETUP_MCE ioctl.
225
226KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
227to the KVM_GET_MSRS system ioctl.  This lets userspace probe host capabilities
228and processor features that are exposed via MSRs (e.g., VMX capabilities).
229This list also varies by kvm version and host processor, but does not change
230otherwise.
231
232
2334.4 KVM_CHECK_EXTENSION
234-----------------------
235
236:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
237:Architectures: all
238:Type: system ioctl, vm ioctl
239:Parameters: extension identifier (KVM_CAP_*)
240:Returns: 0 if unsupported; 1 (or some other positive integer) if supported
241
242The API allows the application to query about extensions to the core
243kvm API.  Userspace passes an extension identifier (an integer) and
244receives an integer that describes the extension availability.
245Generally 0 means no and 1 means yes, but some extensions may report
246additional information in the integer return value.
247
248Based on their initialization different VMs may have different capabilities.
249It is thus encouraged to use the vm ioctl to query for capabilities (available
250with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
251
2524.5 KVM_GET_VCPU_MMAP_SIZE
253--------------------------
254
255:Capability: basic
256:Architectures: all
257:Type: system ioctl
258:Parameters: none
259:Returns: size of vcpu mmap area, in bytes
260
261The KVM_RUN ioctl (cf.) communicates with userspace via a shared
262memory region.  This ioctl returns the size of that region.  See the
263KVM_RUN documentation for details.
264
265Besides the size of the KVM_RUN communication region, other areas of
266the VCPU file descriptor can be mmap-ed, including:
267
268- if KVM_CAP_COALESCED_MMIO is available, a page at
269  KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
270  this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
271  KVM_CAP_COALESCED_MMIO is not documented yet.
272
273- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
274  KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE.  For more information on
275  KVM_CAP_DIRTY_LOG_RING, see section 8.3.
276
277
2784.6 KVM_SET_MEMORY_REGION
279-------------------------
280
281:Capability: basic
282:Architectures: all
283:Type: vm ioctl
284:Parameters: struct kvm_memory_region (in)
285:Returns: 0 on success, -1 on error
286
287This ioctl is obsolete and has been removed.
288
289
2904.7 KVM_CREATE_VCPU
291-------------------
292
293:Capability: basic
294:Architectures: all
295:Type: vm ioctl
296:Parameters: vcpu id (apic id on x86)
297:Returns: vcpu fd on success, -1 on error
298
299This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
300The vcpu id is an integer in the range [0, max_vcpu_id).
301
302The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
303the KVM_CHECK_EXTENSION ioctl() at run-time.
304The maximum possible value for max_vcpus can be retrieved using the
305KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
306
307If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
308cpus max.
309If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
310same as the value returned from KVM_CAP_NR_VCPUS.
311
312The maximum possible value for max_vcpu_id can be retrieved using the
313KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
314
315If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
316is the same as the value returned from KVM_CAP_MAX_VCPUS.
317
318On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
319threads in one or more virtual CPU cores.  (This is because the
320hardware requires all the hardware threads in a CPU core to be in the
321same partition.)  The KVM_CAP_PPC_SMT capability indicates the number
322of vcpus per virtual core (vcore).  The vcore id is obtained by
323dividing the vcpu id by the number of vcpus per vcore.  The vcpus in a
324given vcore will always be in the same physical core as each other
325(though that might be a different physical core from time to time).
326Userspace can control the threading (SMT) mode of the guest by its
327allocation of vcpu ids.  For example, if userspace wants
328single-threaded guest vcpus, it should make all vcpu ids be a multiple
329of the number of vcpus per vcore.
330
331For virtual cpus that have been created with S390 user controlled virtual
332machines, the resulting vcpu fd can be memory mapped at page offset
333KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
334cpu's hardware control block.
335
336
3374.8 KVM_GET_DIRTY_LOG (vm ioctl)
338--------------------------------
339
340:Capability: basic
341:Architectures: all
342:Type: vm ioctl
343:Parameters: struct kvm_dirty_log (in/out)
344:Returns: 0 on success, -1 on error
345
346::
347
348  /* for KVM_GET_DIRTY_LOG */
349  struct kvm_dirty_log {
350	__u32 slot;
351	__u32 padding;
352	union {
353		void __user *dirty_bitmap; /* one bit per page */
354		__u64 padding;
355	};
356  };
357
358Given a memory slot, return a bitmap containing any pages dirtied
359since the last call to this ioctl.  Bit 0 is the first page in the
360memory slot.  Ensure the entire structure is cleared to avoid padding
361issues.
362
363If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
364the address space for which you want to return the dirty bitmap.  See
365KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
366
367The bits in the dirty bitmap are cleared before the ioctl returns, unless
368KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled.  For more information,
369see the description of the capability.
370
3714.9 KVM_SET_MEMORY_ALIAS
372------------------------
373
374:Capability: basic
375:Architectures: x86
376:Type: vm ioctl
377:Parameters: struct kvm_memory_alias (in)
378:Returns: 0 (success), -1 (error)
379
380This ioctl is obsolete and has been removed.
381
382
3834.10 KVM_RUN
384------------
385
386:Capability: basic
387:Architectures: all
388:Type: vcpu ioctl
389:Parameters: none
390:Returns: 0 on success, -1 on error
391
392Errors:
393
394  =======    ==============================================================
395  EINTR      an unmasked signal is pending
396  ENOEXEC    the vcpu hasn't been initialized or the guest tried to execute
397             instructions from device memory (arm64)
398  ENOSYS     data abort outside memslots with no syndrome info and
399             KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
400  EPERM      SVE feature set but not finalized (arm64)
401  =======    ==============================================================
402
403This ioctl is used to run a guest virtual cpu.  While there are no
404explicit parameters, there is an implicit parameter block that can be
405obtained by mmap()ing the vcpu fd at offset 0, with the size given by
406KVM_GET_VCPU_MMAP_SIZE.  The parameter block is formatted as a 'struct
407kvm_run' (see below).
408
409
4104.11 KVM_GET_REGS
411-----------------
412
413:Capability: basic
414:Architectures: all except ARM, arm64
415:Type: vcpu ioctl
416:Parameters: struct kvm_regs (out)
417:Returns: 0 on success, -1 on error
418
419Reads the general purpose registers from the vcpu.
420
421::
422
423  /* x86 */
424  struct kvm_regs {
425	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
426	__u64 rax, rbx, rcx, rdx;
427	__u64 rsi, rdi, rsp, rbp;
428	__u64 r8,  r9,  r10, r11;
429	__u64 r12, r13, r14, r15;
430	__u64 rip, rflags;
431  };
432
433  /* mips */
434  struct kvm_regs {
435	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
436	__u64 gpr[32];
437	__u64 hi;
438	__u64 lo;
439	__u64 pc;
440  };
441
442
4434.12 KVM_SET_REGS
444-----------------
445
446:Capability: basic
447:Architectures: all except ARM, arm64
448:Type: vcpu ioctl
449:Parameters: struct kvm_regs (in)
450:Returns: 0 on success, -1 on error
451
452Writes the general purpose registers into the vcpu.
453
454See KVM_GET_REGS for the data structure.
455
456
4574.13 KVM_GET_SREGS
458------------------
459
460:Capability: basic
461:Architectures: x86, ppc
462:Type: vcpu ioctl
463:Parameters: struct kvm_sregs (out)
464:Returns: 0 on success, -1 on error
465
466Reads special registers from the vcpu.
467
468::
469
470  /* x86 */
471  struct kvm_sregs {
472	struct kvm_segment cs, ds, es, fs, gs, ss;
473	struct kvm_segment tr, ldt;
474	struct kvm_dtable gdt, idt;
475	__u64 cr0, cr2, cr3, cr4, cr8;
476	__u64 efer;
477	__u64 apic_base;
478	__u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
479  };
480
481  /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
482
483interrupt_bitmap is a bitmap of pending external interrupts.  At most
484one bit may be set.  This interrupt has been acknowledged by the APIC
485but not yet injected into the cpu core.
486
487
4884.14 KVM_SET_SREGS
489------------------
490
491:Capability: basic
492:Architectures: x86, ppc
493:Type: vcpu ioctl
494:Parameters: struct kvm_sregs (in)
495:Returns: 0 on success, -1 on error
496
497Writes special registers into the vcpu.  See KVM_GET_SREGS for the
498data structures.
499
500
5014.15 KVM_TRANSLATE
502------------------
503
504:Capability: basic
505:Architectures: x86
506:Type: vcpu ioctl
507:Parameters: struct kvm_translation (in/out)
508:Returns: 0 on success, -1 on error
509
510Translates a virtual address according to the vcpu's current address
511translation mode.
512
513::
514
515  struct kvm_translation {
516	/* in */
517	__u64 linear_address;
518
519	/* out */
520	__u64 physical_address;
521	__u8  valid;
522	__u8  writeable;
523	__u8  usermode;
524	__u8  pad[5];
525  };
526
527
5284.16 KVM_INTERRUPT
529------------------
530
531:Capability: basic
532:Architectures: x86, ppc, mips
533:Type: vcpu ioctl
534:Parameters: struct kvm_interrupt (in)
535:Returns: 0 on success, negative on failure.
536
537Queues a hardware interrupt vector to be injected.
538
539::
540
541  /* for KVM_INTERRUPT */
542  struct kvm_interrupt {
543	/* in */
544	__u32 irq;
545  };
546
547X86:
548^^^^
549
550:Returns:
551
552	========= ===================================
553	  0       on success,
554	 -EEXIST  if an interrupt is already enqueued
555	 -EINVAL  the irq number is invalid
556	 -ENXIO   if the PIC is in the kernel
557	 -EFAULT  if the pointer is invalid
558	========= ===================================
559
560Note 'irq' is an interrupt vector, not an interrupt pin or line. This
561ioctl is useful if the in-kernel PIC is not used.
562
563PPC:
564^^^^
565
566Queues an external interrupt to be injected. This ioctl is overleaded
567with 3 different irq values:
568
569a) KVM_INTERRUPT_SET
570
571   This injects an edge type external interrupt into the guest once it's ready
572   to receive interrupts. When injected, the interrupt is done.
573
574b) KVM_INTERRUPT_UNSET
575
576   This unsets any pending interrupt.
577
578   Only available with KVM_CAP_PPC_UNSET_IRQ.
579
580c) KVM_INTERRUPT_SET_LEVEL
581
582   This injects a level type external interrupt into the guest context. The
583   interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
584   is triggered.
585
586   Only available with KVM_CAP_PPC_IRQ_LEVEL.
587
588Note that any value for 'irq' other than the ones stated above is invalid
589and incurs unexpected behavior.
590
591This is an asynchronous vcpu ioctl and can be invoked from any thread.
592
593MIPS:
594^^^^^
595
596Queues an external interrupt to be injected into the virtual CPU. A negative
597interrupt number dequeues the interrupt.
598
599This is an asynchronous vcpu ioctl and can be invoked from any thread.
600
601
6024.17 KVM_DEBUG_GUEST
603--------------------
604
605:Capability: basic
606:Architectures: none
607:Type: vcpu ioctl
608:Parameters: none)
609:Returns: -1 on error
610
611Support for this has been removed.  Use KVM_SET_GUEST_DEBUG instead.
612
613
6144.18 KVM_GET_MSRS
615-----------------
616
617:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
618:Architectures: x86
619:Type: system ioctl, vcpu ioctl
620:Parameters: struct kvm_msrs (in/out)
621:Returns: number of msrs successfully returned;
622          -1 on error
623
624When used as a system ioctl:
625Reads the values of MSR-based features that are available for the VM.  This
626is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
627The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
628in a system ioctl.
629
630When used as a vcpu ioctl:
631Reads model-specific registers from the vcpu.  Supported msr indices can
632be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
633
634::
635
636  struct kvm_msrs {
637	__u32 nmsrs; /* number of msrs in entries */
638	__u32 pad;
639
640	struct kvm_msr_entry entries[0];
641  };
642
643  struct kvm_msr_entry {
644	__u32 index;
645	__u32 reserved;
646	__u64 data;
647  };
648
649Application code should set the 'nmsrs' member (which indicates the
650size of the entries array) and the 'index' member of each array entry.
651kvm will fill in the 'data' member.
652
653
6544.19 KVM_SET_MSRS
655-----------------
656
657:Capability: basic
658:Architectures: x86
659:Type: vcpu ioctl
660:Parameters: struct kvm_msrs (in)
661:Returns: number of msrs successfully set (see below), -1 on error
662
663Writes model-specific registers to the vcpu.  See KVM_GET_MSRS for the
664data structures.
665
666Application code should set the 'nmsrs' member (which indicates the
667size of the entries array), and the 'index' and 'data' members of each
668array entry.
669
670It tries to set the MSRs in array entries[] one by one. If setting an MSR
671fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
672by KVM, etc..., it stops processing the MSR list and returns the number of
673MSRs that have been set successfully.
674
675
6764.20 KVM_SET_CPUID
677------------------
678
679:Capability: basic
680:Architectures: x86
681:Type: vcpu ioctl
682:Parameters: struct kvm_cpuid (in)
683:Returns: 0 on success, -1 on error
684
685Defines the vcpu responses to the cpuid instruction.  Applications
686should use the KVM_SET_CPUID2 ioctl if available.
687
688Note, when this IOCTL fails, KVM gives no guarantees that previous valid CPUID
689configuration (if there is) is not corrupted. Userspace can get a copy of the
690resulting CPUID configuration through KVM_GET_CPUID2 in case.
691
692::
693
694  struct kvm_cpuid_entry {
695	__u32 function;
696	__u32 eax;
697	__u32 ebx;
698	__u32 ecx;
699	__u32 edx;
700	__u32 padding;
701  };
702
703  /* for KVM_SET_CPUID */
704  struct kvm_cpuid {
705	__u32 nent;
706	__u32 padding;
707	struct kvm_cpuid_entry entries[0];
708  };
709
710
7114.21 KVM_SET_SIGNAL_MASK
712------------------------
713
714:Capability: basic
715:Architectures: all
716:Type: vcpu ioctl
717:Parameters: struct kvm_signal_mask (in)
718:Returns: 0 on success, -1 on error
719
720Defines which signals are blocked during execution of KVM_RUN.  This
721signal mask temporarily overrides the threads signal mask.  Any
722unblocked signal received (except SIGKILL and SIGSTOP, which retain
723their traditional behaviour) will cause KVM_RUN to return with -EINTR.
724
725Note the signal will only be delivered if not blocked by the original
726signal mask.
727
728::
729
730  /* for KVM_SET_SIGNAL_MASK */
731  struct kvm_signal_mask {
732	__u32 len;
733	__u8  sigset[0];
734  };
735
736
7374.22 KVM_GET_FPU
738----------------
739
740:Capability: basic
741:Architectures: x86
742:Type: vcpu ioctl
743:Parameters: struct kvm_fpu (out)
744:Returns: 0 on success, -1 on error
745
746Reads the floating point state from the vcpu.
747
748::
749
750  /* for KVM_GET_FPU and KVM_SET_FPU */
751  struct kvm_fpu {
752	__u8  fpr[8][16];
753	__u16 fcw;
754	__u16 fsw;
755	__u8  ftwx;  /* in fxsave format */
756	__u8  pad1;
757	__u16 last_opcode;
758	__u64 last_ip;
759	__u64 last_dp;
760	__u8  xmm[16][16];
761	__u32 mxcsr;
762	__u32 pad2;
763  };
764
765
7664.23 KVM_SET_FPU
767----------------
768
769:Capability: basic
770:Architectures: x86
771:Type: vcpu ioctl
772:Parameters: struct kvm_fpu (in)
773:Returns: 0 on success, -1 on error
774
775Writes the floating point state to the vcpu.
776
777::
778
779  /* for KVM_GET_FPU and KVM_SET_FPU */
780  struct kvm_fpu {
781	__u8  fpr[8][16];
782	__u16 fcw;
783	__u16 fsw;
784	__u8  ftwx;  /* in fxsave format */
785	__u8  pad1;
786	__u16 last_opcode;
787	__u64 last_ip;
788	__u64 last_dp;
789	__u8  xmm[16][16];
790	__u32 mxcsr;
791	__u32 pad2;
792  };
793
794
7954.24 KVM_CREATE_IRQCHIP
796-----------------------
797
798:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
799:Architectures: x86, ARM, arm64, s390
800:Type: vm ioctl
801:Parameters: none
802:Returns: 0 on success, -1 on error
803
804Creates an interrupt controller model in the kernel.
805On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
806future vcpus to have a local APIC.  IRQ routing for GSIs 0-15 is set to both
807PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
808On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
809KVM_CREATE_DEVICE, which also supports creating a GICv2.  Using
810KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
811On s390, a dummy irq routing table is created.
812
813Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
814before KVM_CREATE_IRQCHIP can be used.
815
816
8174.25 KVM_IRQ_LINE
818-----------------
819
820:Capability: KVM_CAP_IRQCHIP
821:Architectures: x86, arm, arm64
822:Type: vm ioctl
823:Parameters: struct kvm_irq_level
824:Returns: 0 on success, -1 on error
825
826Sets the level of a GSI input to the interrupt controller model in the kernel.
827On some architectures it is required that an interrupt controller model has
828been previously created with KVM_CREATE_IRQCHIP.  Note that edge-triggered
829interrupts require the level to be set to 1 and then back to 0.
830
831On real hardware, interrupt pins can be active-low or active-high.  This
832does not matter for the level field of struct kvm_irq_level: 1 always
833means active (asserted), 0 means inactive (deasserted).
834
835x86 allows the operating system to program the interrupt polarity
836(active-low/active-high) for level-triggered interrupts, and KVM used
837to consider the polarity.  However, due to bitrot in the handling of
838active-low interrupts, the above convention is now valid on x86 too.
839This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED.  Userspace
840should not present interrupts to the guest as active-low unless this
841capability is present (or unless it is not using the in-kernel irqchip,
842of course).
843
844
845ARM/arm64 can signal an interrupt either at the CPU level, or at the
846in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
847use PPIs designated for specific cpus.  The irq field is interpreted
848like this::
849
850  bits:  |  31 ... 28  | 27 ... 24 | 23  ... 16 | 15 ... 0 |
851  field: | vcpu2_index | irq_type  | vcpu_index |  irq_id  |
852
853The irq_type field has the following values:
854
855- irq_type[0]:
856	       out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
857- irq_type[1]:
858	       in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
859               (the vcpu_index field is ignored)
860- irq_type[2]:
861	       in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
862
863(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
864
865In both cases, level is used to assert/deassert the line.
866
867When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
868identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
869must be zero.
870
871Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
872injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
873be used for a userspace interrupt controller.
874
875::
876
877  struct kvm_irq_level {
878	union {
879		__u32 irq;     /* GSI */
880		__s32 status;  /* not used for KVM_IRQ_LEVEL */
881	};
882	__u32 level;           /* 0 or 1 */
883  };
884
885
8864.26 KVM_GET_IRQCHIP
887--------------------
888
889:Capability: KVM_CAP_IRQCHIP
890:Architectures: x86
891:Type: vm ioctl
892:Parameters: struct kvm_irqchip (in/out)
893:Returns: 0 on success, -1 on error
894
895Reads the state of a kernel interrupt controller created with
896KVM_CREATE_IRQCHIP into a buffer provided by the caller.
897
898::
899
900  struct kvm_irqchip {
901	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
902	__u32 pad;
903        union {
904		char dummy[512];  /* reserving space */
905		struct kvm_pic_state pic;
906		struct kvm_ioapic_state ioapic;
907	} chip;
908  };
909
910
9114.27 KVM_SET_IRQCHIP
912--------------------
913
914:Capability: KVM_CAP_IRQCHIP
915:Architectures: x86
916:Type: vm ioctl
917:Parameters: struct kvm_irqchip (in)
918:Returns: 0 on success, -1 on error
919
920Sets the state of a kernel interrupt controller created with
921KVM_CREATE_IRQCHIP from a buffer provided by the caller.
922
923::
924
925  struct kvm_irqchip {
926	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
927	__u32 pad;
928        union {
929		char dummy[512];  /* reserving space */
930		struct kvm_pic_state pic;
931		struct kvm_ioapic_state ioapic;
932	} chip;
933  };
934
935
9364.28 KVM_XEN_HVM_CONFIG
937-----------------------
938
939:Capability: KVM_CAP_XEN_HVM
940:Architectures: x86
941:Type: vm ioctl
942:Parameters: struct kvm_xen_hvm_config (in)
943:Returns: 0 on success, -1 on error
944
945Sets the MSR that the Xen HVM guest uses to initialize its hypercall
946page, and provides the starting address and size of the hypercall
947blobs in userspace.  When the guest writes the MSR, kvm copies one
948page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
949memory.
950
951::
952
953  struct kvm_xen_hvm_config {
954	__u32 flags;
955	__u32 msr;
956	__u64 blob_addr_32;
957	__u64 blob_addr_64;
958	__u8 blob_size_32;
959	__u8 blob_size_64;
960	__u8 pad2[30];
961  };
962
963
9644.29 KVM_GET_CLOCK
965------------------
966
967:Capability: KVM_CAP_ADJUST_CLOCK
968:Architectures: x86
969:Type: vm ioctl
970:Parameters: struct kvm_clock_data (out)
971:Returns: 0 on success, -1 on error
972
973Gets the current timestamp of kvmclock as seen by the current guest. In
974conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
975such as migration.
976
977When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
978set of bits that KVM can return in struct kvm_clock_data's flag member.
979
980The only flag defined now is KVM_CLOCK_TSC_STABLE.  If set, the returned
981value is the exact kvmclock value seen by all VCPUs at the instant
982when KVM_GET_CLOCK was called.  If clear, the returned value is simply
983CLOCK_MONOTONIC plus a constant offset; the offset can be modified
984with KVM_SET_CLOCK.  KVM will try to make all VCPUs follow this clock,
985but the exact value read by each VCPU could differ, because the host
986TSC is not stable.
987
988::
989
990  struct kvm_clock_data {
991	__u64 clock;  /* kvmclock current value */
992	__u32 flags;
993	__u32 pad[9];
994  };
995
996
9974.30 KVM_SET_CLOCK
998------------------
999
1000:Capability: KVM_CAP_ADJUST_CLOCK
1001:Architectures: x86
1002:Type: vm ioctl
1003:Parameters: struct kvm_clock_data (in)
1004:Returns: 0 on success, -1 on error
1005
1006Sets the current timestamp of kvmclock to the value specified in its parameter.
1007In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1008such as migration.
1009
1010::
1011
1012  struct kvm_clock_data {
1013	__u64 clock;  /* kvmclock current value */
1014	__u32 flags;
1015	__u32 pad[9];
1016  };
1017
1018
10194.31 KVM_GET_VCPU_EVENTS
1020------------------------
1021
1022:Capability: KVM_CAP_VCPU_EVENTS
1023:Extended by: KVM_CAP_INTR_SHADOW
1024:Architectures: x86, arm, arm64
1025:Type: vcpu ioctl
1026:Parameters: struct kvm_vcpu_event (out)
1027:Returns: 0 on success, -1 on error
1028
1029X86:
1030^^^^
1031
1032Gets currently pending exceptions, interrupts, and NMIs as well as related
1033states of the vcpu.
1034
1035::
1036
1037  struct kvm_vcpu_events {
1038	struct {
1039		__u8 injected;
1040		__u8 nr;
1041		__u8 has_error_code;
1042		__u8 pending;
1043		__u32 error_code;
1044	} exception;
1045	struct {
1046		__u8 injected;
1047		__u8 nr;
1048		__u8 soft;
1049		__u8 shadow;
1050	} interrupt;
1051	struct {
1052		__u8 injected;
1053		__u8 pending;
1054		__u8 masked;
1055		__u8 pad;
1056	} nmi;
1057	__u32 sipi_vector;
1058	__u32 flags;
1059	struct {
1060		__u8 smm;
1061		__u8 pending;
1062		__u8 smm_inside_nmi;
1063		__u8 latched_init;
1064	} smi;
1065	__u8 reserved[27];
1066	__u8 exception_has_payload;
1067	__u64 exception_payload;
1068  };
1069
1070The following bits are defined in the flags field:
1071
1072- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1073  interrupt.shadow contains a valid state.
1074
1075- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1076  valid state.
1077
1078- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1079  exception_has_payload, exception_payload, and exception.pending
1080  fields contain a valid state. This bit will be set whenever
1081  KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1082
1083ARM/ARM64:
1084^^^^^^^^^^
1085
1086If the guest accesses a device that is being emulated by the host kernel in
1087such a way that a real device would generate a physical SError, KVM may make
1088a virtual SError pending for that VCPU. This system error interrupt remains
1089pending until the guest takes the exception by unmasking PSTATE.A.
1090
1091Running the VCPU may cause it to take a pending SError, or make an access that
1092causes an SError to become pending. The event's description is only valid while
1093the VPCU is not running.
1094
1095This API provides a way to read and write the pending 'event' state that is not
1096visible to the guest. To save, restore or migrate a VCPU the struct representing
1097the state can be read then written using this GET/SET API, along with the other
1098guest-visible registers. It is not possible to 'cancel' an SError that has been
1099made pending.
1100
1101A device being emulated in user-space may also wish to generate an SError. To do
1102this the events structure can be populated by user-space. The current state
1103should be read first, to ensure no existing SError is pending. If an existing
1104SError is pending, the architecture's 'Multiple SError interrupts' rules should
1105be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1106Serviceability (RAS) Specification").
1107
1108SError exceptions always have an ESR value. Some CPUs have the ability to
1109specify what the virtual SError's ESR value should be. These systems will
1110advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1111always have a non-zero value when read, and the agent making an SError pending
1112should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1113the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1114with exception.has_esr as zero, KVM will choose an ESR.
1115
1116Specifying exception.has_esr on a system that does not support it will return
1117-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1118will return -EINVAL.
1119
1120It is not possible to read back a pending external abort (injected via
1121KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1122directly to the virtual CPU).
1123
1124::
1125
1126  struct kvm_vcpu_events {
1127	struct {
1128		__u8 serror_pending;
1129		__u8 serror_has_esr;
1130		__u8 ext_dabt_pending;
1131		/* Align it to 8 bytes */
1132		__u8 pad[5];
1133		__u64 serror_esr;
1134	} exception;
1135	__u32 reserved[12];
1136  };
1137
11384.32 KVM_SET_VCPU_EVENTS
1139------------------------
1140
1141:Capability: KVM_CAP_VCPU_EVENTS
1142:Extended by: KVM_CAP_INTR_SHADOW
1143:Architectures: x86, arm, arm64
1144:Type: vcpu ioctl
1145:Parameters: struct kvm_vcpu_event (in)
1146:Returns: 0 on success, -1 on error
1147
1148X86:
1149^^^^
1150
1151Set pending exceptions, interrupts, and NMIs as well as related states of the
1152vcpu.
1153
1154See KVM_GET_VCPU_EVENTS for the data structure.
1155
1156Fields that may be modified asynchronously by running VCPUs can be excluded
1157from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1158smi.pending. Keep the corresponding bits in the flags field cleared to
1159suppress overwriting the current in-kernel state. The bits are:
1160
1161===============================  ==================================
1162KVM_VCPUEVENT_VALID_NMI_PENDING  transfer nmi.pending to the kernel
1163KVM_VCPUEVENT_VALID_SIPI_VECTOR  transfer sipi_vector
1164KVM_VCPUEVENT_VALID_SMM          transfer the smi sub-struct.
1165===============================  ==================================
1166
1167If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1168the flags field to signal that interrupt.shadow contains a valid state and
1169shall be written into the VCPU.
1170
1171KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1172
1173If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1174can be set in the flags field to signal that the
1175exception_has_payload, exception_payload, and exception.pending fields
1176contain a valid state and shall be written into the VCPU.
1177
1178ARM/ARM64:
1179^^^^^^^^^^
1180
1181User space may need to inject several types of events to the guest.
1182
1183Set the pending SError exception state for this VCPU. It is not possible to
1184'cancel' an Serror that has been made pending.
1185
1186If the guest performed an access to I/O memory which could not be handled by
1187userspace, for example because of missing instruction syndrome decode
1188information or because there is no device mapped at the accessed IPA, then
1189userspace can ask the kernel to inject an external abort using the address
1190from the exiting fault on the VCPU. It is a programming error to set
1191ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1192KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1193KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1194how userspace reports accesses for the above cases to guests, across different
1195userspace implementations. Nevertheless, userspace can still emulate all Arm
1196exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1197
1198See KVM_GET_VCPU_EVENTS for the data structure.
1199
1200
12014.33 KVM_GET_DEBUGREGS
1202----------------------
1203
1204:Capability: KVM_CAP_DEBUGREGS
1205:Architectures: x86
1206:Type: vm ioctl
1207:Parameters: struct kvm_debugregs (out)
1208:Returns: 0 on success, -1 on error
1209
1210Reads debug registers from the vcpu.
1211
1212::
1213
1214  struct kvm_debugregs {
1215	__u64 db[4];
1216	__u64 dr6;
1217	__u64 dr7;
1218	__u64 flags;
1219	__u64 reserved[9];
1220  };
1221
1222
12234.34 KVM_SET_DEBUGREGS
1224----------------------
1225
1226:Capability: KVM_CAP_DEBUGREGS
1227:Architectures: x86
1228:Type: vm ioctl
1229:Parameters: struct kvm_debugregs (in)
1230:Returns: 0 on success, -1 on error
1231
1232Writes debug registers into the vcpu.
1233
1234See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1235yet and must be cleared on entry.
1236
1237
12384.35 KVM_SET_USER_MEMORY_REGION
1239-------------------------------
1240
1241:Capability: KVM_CAP_USER_MEMORY
1242:Architectures: all
1243:Type: vm ioctl
1244:Parameters: struct kvm_userspace_memory_region (in)
1245:Returns: 0 on success, -1 on error
1246
1247::
1248
1249  struct kvm_userspace_memory_region {
1250	__u32 slot;
1251	__u32 flags;
1252	__u64 guest_phys_addr;
1253	__u64 memory_size; /* bytes */
1254	__u64 userspace_addr; /* start of the userspace allocated memory */
1255  };
1256
1257  /* for kvm_memory_region::flags */
1258  #define KVM_MEM_LOG_DIRTY_PAGES	(1UL << 0)
1259  #define KVM_MEM_READONLY	(1UL << 1)
1260
1261This ioctl allows the user to create, modify or delete a guest physical
1262memory slot.  Bits 0-15 of "slot" specify the slot id and this value
1263should be less than the maximum number of user memory slots supported per
1264VM.  The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1265Slots may not overlap in guest physical address space.
1266
1267If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1268specifies the address space which is being modified.  They must be
1269less than the value that KVM_CHECK_EXTENSION returns for the
1270KVM_CAP_MULTI_ADDRESS_SPACE capability.  Slots in separate address spaces
1271are unrelated; the restriction on overlapping slots only applies within
1272each address space.
1273
1274Deleting a slot is done by passing zero for memory_size.  When changing
1275an existing slot, it may be moved in the guest physical memory space,
1276or its flags may be modified, but it may not be resized.
1277
1278Memory for the region is taken starting at the address denoted by the
1279field userspace_addr, which must point at user addressable memory for
1280the entire memory slot size.  Any object may back this memory, including
1281anonymous memory, ordinary files, and hugetlbfs.
1282
1283On architectures that support a form of address tagging, userspace_addr must
1284be an untagged address.
1285
1286It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1287be identical.  This allows large pages in the guest to be backed by large
1288pages in the host.
1289
1290The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1291KVM_MEM_READONLY.  The former can be set to instruct KVM to keep track of
1292writes to memory within the slot.  See KVM_GET_DIRTY_LOG ioctl to know how to
1293use it.  The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1294to make a new slot read-only.  In this case, writes to this memory will be
1295posted to userspace as KVM_EXIT_MMIO exits.
1296
1297When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1298the memory region are automatically reflected into the guest.  For example, an
1299mmap() that affects the region will be made visible immediately.  Another
1300example is madvise(MADV_DROP).
1301
1302It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
1303The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
1304allocation and is deprecated.
1305
1306
13074.36 KVM_SET_TSS_ADDR
1308---------------------
1309
1310:Capability: KVM_CAP_SET_TSS_ADDR
1311:Architectures: x86
1312:Type: vm ioctl
1313:Parameters: unsigned long tss_address (in)
1314:Returns: 0 on success, -1 on error
1315
1316This ioctl defines the physical address of a three-page region in the guest
1317physical address space.  The region must be within the first 4GB of the
1318guest physical address space and must not conflict with any memory slot
1319or any mmio address.  The guest may malfunction if it accesses this memory
1320region.
1321
1322This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1323because of a quirk in the virtualization implementation (see the internals
1324documentation when it pops into existence).
1325
1326
13274.37 KVM_ENABLE_CAP
1328-------------------
1329
1330:Capability: KVM_CAP_ENABLE_CAP
1331:Architectures: mips, ppc, s390
1332:Type: vcpu ioctl
1333:Parameters: struct kvm_enable_cap (in)
1334:Returns: 0 on success; -1 on error
1335
1336:Capability: KVM_CAP_ENABLE_CAP_VM
1337:Architectures: all
1338:Type: vm ioctl
1339:Parameters: struct kvm_enable_cap (in)
1340:Returns: 0 on success; -1 on error
1341
1342.. note::
1343
1344   Not all extensions are enabled by default. Using this ioctl the application
1345   can enable an extension, making it available to the guest.
1346
1347On systems that do not support this ioctl, it always fails. On systems that
1348do support it, it only works for extensions that are supported for enablement.
1349
1350To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1351be used.
1352
1353::
1354
1355  struct kvm_enable_cap {
1356       /* in */
1357       __u32 cap;
1358
1359The capability that is supposed to get enabled.
1360
1361::
1362
1363       __u32 flags;
1364
1365A bitfield indicating future enhancements. Has to be 0 for now.
1366
1367::
1368
1369       __u64 args[4];
1370
1371Arguments for enabling a feature. If a feature needs initial values to
1372function properly, this is the place to put them.
1373
1374::
1375
1376       __u8  pad[64];
1377  };
1378
1379The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1380for vm-wide capabilities.
1381
13824.38 KVM_GET_MP_STATE
1383---------------------
1384
1385:Capability: KVM_CAP_MP_STATE
1386:Architectures: x86, s390, arm, arm64
1387:Type: vcpu ioctl
1388:Parameters: struct kvm_mp_state (out)
1389:Returns: 0 on success; -1 on error
1390
1391::
1392
1393  struct kvm_mp_state {
1394	__u32 mp_state;
1395  };
1396
1397Returns the vcpu's current "multiprocessing state" (though also valid on
1398uniprocessor guests).
1399
1400Possible values are:
1401
1402   ==========================    ===============================================
1403   KVM_MP_STATE_RUNNABLE         the vcpu is currently running [x86,arm/arm64]
1404   KVM_MP_STATE_UNINITIALIZED    the vcpu is an application processor (AP)
1405                                 which has not yet received an INIT signal [x86]
1406   KVM_MP_STATE_INIT_RECEIVED    the vcpu has received an INIT signal, and is
1407                                 now ready for a SIPI [x86]
1408   KVM_MP_STATE_HALTED           the vcpu has executed a HLT instruction and
1409                                 is waiting for an interrupt [x86]
1410   KVM_MP_STATE_SIPI_RECEIVED    the vcpu has just received a SIPI (vector
1411                                 accessible via KVM_GET_VCPU_EVENTS) [x86]
1412   KVM_MP_STATE_STOPPED          the vcpu is stopped [s390,arm/arm64]
1413   KVM_MP_STATE_CHECK_STOP       the vcpu is in a special error state [s390]
1414   KVM_MP_STATE_OPERATING        the vcpu is operating (running or halted)
1415                                 [s390]
1416   KVM_MP_STATE_LOAD             the vcpu is in a special load/startup state
1417                                 [s390]
1418   ==========================    ===============================================
1419
1420On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1421in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1422these architectures.
1423
1424For arm/arm64:
1425^^^^^^^^^^^^^^
1426
1427The only states that are valid are KVM_MP_STATE_STOPPED and
1428KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1429
14304.39 KVM_SET_MP_STATE
1431---------------------
1432
1433:Capability: KVM_CAP_MP_STATE
1434:Architectures: x86, s390, arm, arm64
1435:Type: vcpu ioctl
1436:Parameters: struct kvm_mp_state (in)
1437:Returns: 0 on success; -1 on error
1438
1439Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1440arguments.
1441
1442On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1443in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1444these architectures.
1445
1446For arm/arm64:
1447^^^^^^^^^^^^^^
1448
1449The only states that are valid are KVM_MP_STATE_STOPPED and
1450KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1451
14524.40 KVM_SET_IDENTITY_MAP_ADDR
1453------------------------------
1454
1455:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1456:Architectures: x86
1457:Type: vm ioctl
1458:Parameters: unsigned long identity (in)
1459:Returns: 0 on success, -1 on error
1460
1461This ioctl defines the physical address of a one-page region in the guest
1462physical address space.  The region must be within the first 4GB of the
1463guest physical address space and must not conflict with any memory slot
1464or any mmio address.  The guest may malfunction if it accesses this memory
1465region.
1466
1467Setting the address to 0 will result in resetting the address to its default
1468(0xfffbc000).
1469
1470This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1471because of a quirk in the virtualization implementation (see the internals
1472documentation when it pops into existence).
1473
1474Fails if any VCPU has already been created.
1475
14764.41 KVM_SET_BOOT_CPU_ID
1477------------------------
1478
1479:Capability: KVM_CAP_SET_BOOT_CPU_ID
1480:Architectures: x86
1481:Type: vm ioctl
1482:Parameters: unsigned long vcpu_id
1483:Returns: 0 on success, -1 on error
1484
1485Define which vcpu is the Bootstrap Processor (BSP).  Values are the same
1486as the vcpu id in KVM_CREATE_VCPU.  If this ioctl is not called, the default
1487is vcpu 0.
1488
1489
14904.42 KVM_GET_XSAVE
1491------------------
1492
1493:Capability: KVM_CAP_XSAVE
1494:Architectures: x86
1495:Type: vcpu ioctl
1496:Parameters: struct kvm_xsave (out)
1497:Returns: 0 on success, -1 on error
1498
1499
1500::
1501
1502  struct kvm_xsave {
1503	__u32 region[1024];
1504  };
1505
1506This ioctl would copy current vcpu's xsave struct to the userspace.
1507
1508
15094.43 KVM_SET_XSAVE
1510------------------
1511
1512:Capability: KVM_CAP_XSAVE
1513:Architectures: x86
1514:Type: vcpu ioctl
1515:Parameters: struct kvm_xsave (in)
1516:Returns: 0 on success, -1 on error
1517
1518::
1519
1520
1521  struct kvm_xsave {
1522	__u32 region[1024];
1523  };
1524
1525This ioctl would copy userspace's xsave struct to the kernel.
1526
1527
15284.44 KVM_GET_XCRS
1529-----------------
1530
1531:Capability: KVM_CAP_XCRS
1532:Architectures: x86
1533:Type: vcpu ioctl
1534:Parameters: struct kvm_xcrs (out)
1535:Returns: 0 on success, -1 on error
1536
1537::
1538
1539  struct kvm_xcr {
1540	__u32 xcr;
1541	__u32 reserved;
1542	__u64 value;
1543  };
1544
1545  struct kvm_xcrs {
1546	__u32 nr_xcrs;
1547	__u32 flags;
1548	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1549	__u64 padding[16];
1550  };
1551
1552This ioctl would copy current vcpu's xcrs to the userspace.
1553
1554
15554.45 KVM_SET_XCRS
1556-----------------
1557
1558:Capability: KVM_CAP_XCRS
1559:Architectures: x86
1560:Type: vcpu ioctl
1561:Parameters: struct kvm_xcrs (in)
1562:Returns: 0 on success, -1 on error
1563
1564::
1565
1566  struct kvm_xcr {
1567	__u32 xcr;
1568	__u32 reserved;
1569	__u64 value;
1570  };
1571
1572  struct kvm_xcrs {
1573	__u32 nr_xcrs;
1574	__u32 flags;
1575	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1576	__u64 padding[16];
1577  };
1578
1579This ioctl would set vcpu's xcr to the value userspace specified.
1580
1581
15824.46 KVM_GET_SUPPORTED_CPUID
1583----------------------------
1584
1585:Capability: KVM_CAP_EXT_CPUID
1586:Architectures: x86
1587:Type: system ioctl
1588:Parameters: struct kvm_cpuid2 (in/out)
1589:Returns: 0 on success, -1 on error
1590
1591::
1592
1593  struct kvm_cpuid2 {
1594	__u32 nent;
1595	__u32 padding;
1596	struct kvm_cpuid_entry2 entries[0];
1597  };
1598
1599  #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
1600  #define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1) /* deprecated */
1601  #define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2) /* deprecated */
1602
1603  struct kvm_cpuid_entry2 {
1604	__u32 function;
1605	__u32 index;
1606	__u32 flags;
1607	__u32 eax;
1608	__u32 ebx;
1609	__u32 ecx;
1610	__u32 edx;
1611	__u32 padding[3];
1612  };
1613
1614This ioctl returns x86 cpuid features which are supported by both the
1615hardware and kvm in its default configuration.  Userspace can use the
1616information returned by this ioctl to construct cpuid information (for
1617KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1618userspace capabilities, and with user requirements (for example, the
1619user may wish to constrain cpuid to emulate older hardware, or for
1620feature consistency across a cluster).
1621
1622Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1623expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1624its default configuration. If userspace enables such capabilities, it
1625is responsible for modifying the results of this ioctl appropriately.
1626
1627Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1628with the 'nent' field indicating the number of entries in the variable-size
1629array 'entries'.  If the number of entries is too low to describe the cpu
1630capabilities, an error (E2BIG) is returned.  If the number is too high,
1631the 'nent' field is adjusted and an error (ENOMEM) is returned.  If the
1632number is just right, the 'nent' field is adjusted to the number of valid
1633entries in the 'entries' array, which is then filled.
1634
1635The entries returned are the host cpuid as returned by the cpuid instruction,
1636with unknown or unsupported features masked out.  Some features (for example,
1637x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1638emulate them efficiently. The fields in each entry are defined as follows:
1639
1640  function:
1641         the eax value used to obtain the entry
1642
1643  index:
1644         the ecx value used to obtain the entry (for entries that are
1645         affected by ecx)
1646
1647  flags:
1648     an OR of zero or more of the following:
1649
1650        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1651           if the index field is valid
1652
1653   eax, ebx, ecx, edx:
1654         the values returned by the cpuid instruction for
1655         this function/index combination
1656
1657The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1658as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1659support.  Instead it is reported via::
1660
1661  ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1662
1663if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1664feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1665
1666
16674.47 KVM_PPC_GET_PVINFO
1668-----------------------
1669
1670:Capability: KVM_CAP_PPC_GET_PVINFO
1671:Architectures: ppc
1672:Type: vm ioctl
1673:Parameters: struct kvm_ppc_pvinfo (out)
1674:Returns: 0 on success, !0 on error
1675
1676::
1677
1678  struct kvm_ppc_pvinfo {
1679	__u32 flags;
1680	__u32 hcall[4];
1681	__u8  pad[108];
1682  };
1683
1684This ioctl fetches PV specific information that need to be passed to the guest
1685using the device tree or other means from vm context.
1686
1687The hcall array defines 4 instructions that make up a hypercall.
1688
1689If any additional field gets added to this structure later on, a bit for that
1690additional piece of information will be set in the flags bitmap.
1691
1692The flags bitmap is defined as::
1693
1694   /* the host supports the ePAPR idle hcall
1695   #define KVM_PPC_PVINFO_FLAGS_EV_IDLE   (1<<0)
1696
16974.52 KVM_SET_GSI_ROUTING
1698------------------------
1699
1700:Capability: KVM_CAP_IRQ_ROUTING
1701:Architectures: x86 s390 arm arm64
1702:Type: vm ioctl
1703:Parameters: struct kvm_irq_routing (in)
1704:Returns: 0 on success, -1 on error
1705
1706Sets the GSI routing table entries, overwriting any previously set entries.
1707
1708On arm/arm64, GSI routing has the following limitation:
1709
1710- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1711
1712::
1713
1714  struct kvm_irq_routing {
1715	__u32 nr;
1716	__u32 flags;
1717	struct kvm_irq_routing_entry entries[0];
1718  };
1719
1720No flags are specified so far, the corresponding field must be set to zero.
1721
1722::
1723
1724  struct kvm_irq_routing_entry {
1725	__u32 gsi;
1726	__u32 type;
1727	__u32 flags;
1728	__u32 pad;
1729	union {
1730		struct kvm_irq_routing_irqchip irqchip;
1731		struct kvm_irq_routing_msi msi;
1732		struct kvm_irq_routing_s390_adapter adapter;
1733		struct kvm_irq_routing_hv_sint hv_sint;
1734		__u32 pad[8];
1735	} u;
1736  };
1737
1738  /* gsi routing entry types */
1739  #define KVM_IRQ_ROUTING_IRQCHIP 1
1740  #define KVM_IRQ_ROUTING_MSI 2
1741  #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1742  #define KVM_IRQ_ROUTING_HV_SINT 4
1743
1744flags:
1745
1746- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1747  type, specifies that the devid field contains a valid value.  The per-VM
1748  KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1749  the device ID.  If this capability is not available, userspace should
1750  never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1751- zero otherwise
1752
1753::
1754
1755  struct kvm_irq_routing_irqchip {
1756	__u32 irqchip;
1757	__u32 pin;
1758  };
1759
1760  struct kvm_irq_routing_msi {
1761	__u32 address_lo;
1762	__u32 address_hi;
1763	__u32 data;
1764	union {
1765		__u32 pad;
1766		__u32 devid;
1767	};
1768  };
1769
1770If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1771for the device that wrote the MSI message.  For PCI, this is usually a
1772BFD identifier in the lower 16 bits.
1773
1774On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1775feature of KVM_CAP_X2APIC_API capability is enabled.  If it is enabled,
1776address_hi bits 31-8 provide bits 31-8 of the destination id.  Bits 7-0 of
1777address_hi must be zero.
1778
1779::
1780
1781  struct kvm_irq_routing_s390_adapter {
1782	__u64 ind_addr;
1783	__u64 summary_addr;
1784	__u64 ind_offset;
1785	__u32 summary_offset;
1786	__u32 adapter_id;
1787  };
1788
1789  struct kvm_irq_routing_hv_sint {
1790	__u32 vcpu;
1791	__u32 sint;
1792  };
1793
1794
17954.55 KVM_SET_TSC_KHZ
1796--------------------
1797
1798:Capability: KVM_CAP_TSC_CONTROL
1799:Architectures: x86
1800:Type: vcpu ioctl
1801:Parameters: virtual tsc_khz
1802:Returns: 0 on success, -1 on error
1803
1804Specifies the tsc frequency for the virtual machine. The unit of the
1805frequency is KHz.
1806
1807
18084.56 KVM_GET_TSC_KHZ
1809--------------------
1810
1811:Capability: KVM_CAP_GET_TSC_KHZ
1812:Architectures: x86
1813:Type: vcpu ioctl
1814:Parameters: none
1815:Returns: virtual tsc-khz on success, negative value on error
1816
1817Returns the tsc frequency of the guest. The unit of the return value is
1818KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1819error.
1820
1821
18224.57 KVM_GET_LAPIC
1823------------------
1824
1825:Capability: KVM_CAP_IRQCHIP
1826:Architectures: x86
1827:Type: vcpu ioctl
1828:Parameters: struct kvm_lapic_state (out)
1829:Returns: 0 on success, -1 on error
1830
1831::
1832
1833  #define KVM_APIC_REG_SIZE 0x400
1834  struct kvm_lapic_state {
1835	char regs[KVM_APIC_REG_SIZE];
1836  };
1837
1838Reads the Local APIC registers and copies them into the input argument.  The
1839data format and layout are the same as documented in the architecture manual.
1840
1841If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1842enabled, then the format of APIC_ID register depends on the APIC mode
1843(reported by MSR_IA32_APICBASE) of its VCPU.  x2APIC stores APIC ID in
1844the APIC_ID register (bytes 32-35).  xAPIC only allows an 8-bit APIC ID
1845which is stored in bits 31-24 of the APIC register, or equivalently in
1846byte 35 of struct kvm_lapic_state's regs field.  KVM_GET_LAPIC must then
1847be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1848
1849If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1850always uses xAPIC format.
1851
1852
18534.58 KVM_SET_LAPIC
1854------------------
1855
1856:Capability: KVM_CAP_IRQCHIP
1857:Architectures: x86
1858:Type: vcpu ioctl
1859:Parameters: struct kvm_lapic_state (in)
1860:Returns: 0 on success, -1 on error
1861
1862::
1863
1864  #define KVM_APIC_REG_SIZE 0x400
1865  struct kvm_lapic_state {
1866	char regs[KVM_APIC_REG_SIZE];
1867  };
1868
1869Copies the input argument into the Local APIC registers.  The data format
1870and layout are the same as documented in the architecture manual.
1871
1872The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1873regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1874See the note in KVM_GET_LAPIC.
1875
1876
18774.59 KVM_IOEVENTFD
1878------------------
1879
1880:Capability: KVM_CAP_IOEVENTFD
1881:Architectures: all
1882:Type: vm ioctl
1883:Parameters: struct kvm_ioeventfd (in)
1884:Returns: 0 on success, !0 on error
1885
1886This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1887within the guest.  A guest write in the registered address will signal the
1888provided event instead of triggering an exit.
1889
1890::
1891
1892  struct kvm_ioeventfd {
1893	__u64 datamatch;
1894	__u64 addr;        /* legal pio/mmio address */
1895	__u32 len;         /* 0, 1, 2, 4, or 8 bytes    */
1896	__s32 fd;
1897	__u32 flags;
1898	__u8  pad[36];
1899  };
1900
1901For the special case of virtio-ccw devices on s390, the ioevent is matched
1902to a subchannel/virtqueue tuple instead.
1903
1904The following flags are defined::
1905
1906  #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1907  #define KVM_IOEVENTFD_FLAG_PIO       (1 << kvm_ioeventfd_flag_nr_pio)
1908  #define KVM_IOEVENTFD_FLAG_DEASSIGN  (1 << kvm_ioeventfd_flag_nr_deassign)
1909  #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1910	(1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1911
1912If datamatch flag is set, the event will be signaled only if the written value
1913to the registered address is equal to datamatch in struct kvm_ioeventfd.
1914
1915For virtio-ccw devices, addr contains the subchannel id and datamatch the
1916virtqueue index.
1917
1918With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1919the kernel will ignore the length of guest write and may get a faster vmexit.
1920The speedup may only apply to specific architectures, but the ioeventfd will
1921work anyway.
1922
19234.60 KVM_DIRTY_TLB
1924------------------
1925
1926:Capability: KVM_CAP_SW_TLB
1927:Architectures: ppc
1928:Type: vcpu ioctl
1929:Parameters: struct kvm_dirty_tlb (in)
1930:Returns: 0 on success, -1 on error
1931
1932::
1933
1934  struct kvm_dirty_tlb {
1935	__u64 bitmap;
1936	__u32 num_dirty;
1937  };
1938
1939This must be called whenever userspace has changed an entry in the shared
1940TLB, prior to calling KVM_RUN on the associated vcpu.
1941
1942The "bitmap" field is the userspace address of an array.  This array
1943consists of a number of bits, equal to the total number of TLB entries as
1944determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1945nearest multiple of 64.
1946
1947Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1948array.
1949
1950The array is little-endian: the bit 0 is the least significant bit of the
1951first byte, bit 8 is the least significant bit of the second byte, etc.
1952This avoids any complications with differing word sizes.
1953
1954The "num_dirty" field is a performance hint for KVM to determine whether it
1955should skip processing the bitmap and just invalidate everything.  It must
1956be set to the number of set bits in the bitmap.
1957
1958
19594.62 KVM_CREATE_SPAPR_TCE
1960-------------------------
1961
1962:Capability: KVM_CAP_SPAPR_TCE
1963:Architectures: powerpc
1964:Type: vm ioctl
1965:Parameters: struct kvm_create_spapr_tce (in)
1966:Returns: file descriptor for manipulating the created TCE table
1967
1968This creates a virtual TCE (translation control entry) table, which
1969is an IOMMU for PAPR-style virtual I/O.  It is used to translate
1970logical addresses used in virtual I/O into guest physical addresses,
1971and provides a scatter/gather capability for PAPR virtual I/O.
1972
1973::
1974
1975  /* for KVM_CAP_SPAPR_TCE */
1976  struct kvm_create_spapr_tce {
1977	__u64 liobn;
1978	__u32 window_size;
1979  };
1980
1981The liobn field gives the logical IO bus number for which to create a
1982TCE table.  The window_size field specifies the size of the DMA window
1983which this TCE table will translate - the table will contain one 64
1984bit TCE entry for every 4kiB of the DMA window.
1985
1986When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1987table has been created using this ioctl(), the kernel will handle it
1988in real mode, updating the TCE table.  H_PUT_TCE calls for other
1989liobns will cause a vm exit and must be handled by userspace.
1990
1991The return value is a file descriptor which can be passed to mmap(2)
1992to map the created TCE table into userspace.  This lets userspace read
1993the entries written by kernel-handled H_PUT_TCE calls, and also lets
1994userspace update the TCE table directly which is useful in some
1995circumstances.
1996
1997
19984.63 KVM_ALLOCATE_RMA
1999---------------------
2000
2001:Capability: KVM_CAP_PPC_RMA
2002:Architectures: powerpc
2003:Type: vm ioctl
2004:Parameters: struct kvm_allocate_rma (out)
2005:Returns: file descriptor for mapping the allocated RMA
2006
2007This allocates a Real Mode Area (RMA) from the pool allocated at boot
2008time by the kernel.  An RMA is a physically-contiguous, aligned region
2009of memory used on older POWER processors to provide the memory which
2010will be accessed by real-mode (MMU off) accesses in a KVM guest.
2011POWER processors support a set of sizes for the RMA that usually
2012includes 64MB, 128MB, 256MB and some larger powers of two.
2013
2014::
2015
2016  /* for KVM_ALLOCATE_RMA */
2017  struct kvm_allocate_rma {
2018	__u64 rma_size;
2019  };
2020
2021The return value is a file descriptor which can be passed to mmap(2)
2022to map the allocated RMA into userspace.  The mapped area can then be
2023passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2024RMA for a virtual machine.  The size of the RMA in bytes (which is
2025fixed at host kernel boot time) is returned in the rma_size field of
2026the argument structure.
2027
2028The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2029is supported; 2 if the processor requires all virtual machines to have
2030an RMA, or 1 if the processor can use an RMA but doesn't require it,
2031because it supports the Virtual RMA (VRMA) facility.
2032
2033
20344.64 KVM_NMI
2035------------
2036
2037:Capability: KVM_CAP_USER_NMI
2038:Architectures: x86
2039:Type: vcpu ioctl
2040:Parameters: none
2041:Returns: 0 on success, -1 on error
2042
2043Queues an NMI on the thread's vcpu.  Note this is well defined only
2044when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2045between the virtual cpu core and virtual local APIC.  After KVM_CREATE_IRQCHIP
2046has been called, this interface is completely emulated within the kernel.
2047
2048To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2049following algorithm:
2050
2051  - pause the vcpu
2052  - read the local APIC's state (KVM_GET_LAPIC)
2053  - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2054  - if so, issue KVM_NMI
2055  - resume the vcpu
2056
2057Some guests configure the LINT1 NMI input to cause a panic, aiding in
2058debugging.
2059
2060
20614.65 KVM_S390_UCAS_MAP
2062----------------------
2063
2064:Capability: KVM_CAP_S390_UCONTROL
2065:Architectures: s390
2066:Type: vcpu ioctl
2067:Parameters: struct kvm_s390_ucas_mapping (in)
2068:Returns: 0 in case of success
2069
2070The parameter is defined like this::
2071
2072	struct kvm_s390_ucas_mapping {
2073		__u64 user_addr;
2074		__u64 vcpu_addr;
2075		__u64 length;
2076	};
2077
2078This ioctl maps the memory at "user_addr" with the length "length" to
2079the vcpu's address space starting at "vcpu_addr". All parameters need to
2080be aligned by 1 megabyte.
2081
2082
20834.66 KVM_S390_UCAS_UNMAP
2084------------------------
2085
2086:Capability: KVM_CAP_S390_UCONTROL
2087:Architectures: s390
2088:Type: vcpu ioctl
2089:Parameters: struct kvm_s390_ucas_mapping (in)
2090:Returns: 0 in case of success
2091
2092The parameter is defined like this::
2093
2094	struct kvm_s390_ucas_mapping {
2095		__u64 user_addr;
2096		__u64 vcpu_addr;
2097		__u64 length;
2098	};
2099
2100This ioctl unmaps the memory in the vcpu's address space starting at
2101"vcpu_addr" with the length "length". The field "user_addr" is ignored.
2102All parameters need to be aligned by 1 megabyte.
2103
2104
21054.67 KVM_S390_VCPU_FAULT
2106------------------------
2107
2108:Capability: KVM_CAP_S390_UCONTROL
2109:Architectures: s390
2110:Type: vcpu ioctl
2111:Parameters: vcpu absolute address (in)
2112:Returns: 0 in case of success
2113
2114This call creates a page table entry on the virtual cpu's address space
2115(for user controlled virtual machines) or the virtual machine's address
2116space (for regular virtual machines). This only works for minor faults,
2117thus it's recommended to access subject memory page via the user page
2118table upfront. This is useful to handle validity intercepts for user
2119controlled virtual machines to fault in the virtual cpu's lowcore pages
2120prior to calling the KVM_RUN ioctl.
2121
2122
21234.68 KVM_SET_ONE_REG
2124--------------------
2125
2126:Capability: KVM_CAP_ONE_REG
2127:Architectures: all
2128:Type: vcpu ioctl
2129:Parameters: struct kvm_one_reg (in)
2130:Returns: 0 on success, negative value on failure
2131
2132Errors:
2133
2134  ======   ============================================================
2135  ENOENT   no such register
2136  EINVAL   invalid register ID, or no such register or used with VMs in
2137           protected virtualization mode on s390
2138  EPERM    (arm64) register access not allowed before vcpu finalization
2139  ======   ============================================================
2140
2141(These error codes are indicative only: do not rely on a specific error
2142code being returned in a specific situation.)
2143
2144::
2145
2146  struct kvm_one_reg {
2147       __u64 id;
2148       __u64 addr;
2149 };
2150
2151Using this ioctl, a single vcpu register can be set to a specific value
2152defined by user space with the passed in struct kvm_one_reg, where id
2153refers to the register identifier as described below and addr is a pointer
2154to a variable with the respective size. There can be architecture agnostic
2155and architecture specific registers. Each have their own range of operation
2156and their own constants and width. To keep track of the implemented
2157registers, find a list below:
2158
2159  ======= =============================== ============
2160  Arch              Register              Width (bits)
2161  ======= =============================== ============
2162  PPC     KVM_REG_PPC_HIOR                64
2163  PPC     KVM_REG_PPC_IAC1                64
2164  PPC     KVM_REG_PPC_IAC2                64
2165  PPC     KVM_REG_PPC_IAC3                64
2166  PPC     KVM_REG_PPC_IAC4                64
2167  PPC     KVM_REG_PPC_DAC1                64
2168  PPC     KVM_REG_PPC_DAC2                64
2169  PPC     KVM_REG_PPC_DABR                64
2170  PPC     KVM_REG_PPC_DSCR                64
2171  PPC     KVM_REG_PPC_PURR                64
2172  PPC     KVM_REG_PPC_SPURR               64
2173  PPC     KVM_REG_PPC_DAR                 64
2174  PPC     KVM_REG_PPC_DSISR               32
2175  PPC     KVM_REG_PPC_AMR                 64
2176  PPC     KVM_REG_PPC_UAMOR               64
2177  PPC     KVM_REG_PPC_MMCR0               64
2178  PPC     KVM_REG_PPC_MMCR1               64
2179  PPC     KVM_REG_PPC_MMCRA               64
2180  PPC     KVM_REG_PPC_MMCR2               64
2181  PPC     KVM_REG_PPC_MMCRS               64
2182  PPC     KVM_REG_PPC_MMCR3               64
2183  PPC     KVM_REG_PPC_SIAR                64
2184  PPC     KVM_REG_PPC_SDAR                64
2185  PPC     KVM_REG_PPC_SIER                64
2186  PPC     KVM_REG_PPC_SIER2               64
2187  PPC     KVM_REG_PPC_SIER3               64
2188  PPC     KVM_REG_PPC_PMC1                32
2189  PPC     KVM_REG_PPC_PMC2                32
2190  PPC     KVM_REG_PPC_PMC3                32
2191  PPC     KVM_REG_PPC_PMC4                32
2192  PPC     KVM_REG_PPC_PMC5                32
2193  PPC     KVM_REG_PPC_PMC6                32
2194  PPC     KVM_REG_PPC_PMC7                32
2195  PPC     KVM_REG_PPC_PMC8                32
2196  PPC     KVM_REG_PPC_FPR0                64
2197  ...
2198  PPC     KVM_REG_PPC_FPR31               64
2199  PPC     KVM_REG_PPC_VR0                 128
2200  ...
2201  PPC     KVM_REG_PPC_VR31                128
2202  PPC     KVM_REG_PPC_VSR0                128
2203  ...
2204  PPC     KVM_REG_PPC_VSR31               128
2205  PPC     KVM_REG_PPC_FPSCR               64
2206  PPC     KVM_REG_PPC_VSCR                32
2207  PPC     KVM_REG_PPC_VPA_ADDR            64
2208  PPC     KVM_REG_PPC_VPA_SLB             128
2209  PPC     KVM_REG_PPC_VPA_DTL             128
2210  PPC     KVM_REG_PPC_EPCR                32
2211  PPC     KVM_REG_PPC_EPR                 32
2212  PPC     KVM_REG_PPC_TCR                 32
2213  PPC     KVM_REG_PPC_TSR                 32
2214  PPC     KVM_REG_PPC_OR_TSR              32
2215  PPC     KVM_REG_PPC_CLEAR_TSR           32
2216  PPC     KVM_REG_PPC_MAS0                32
2217  PPC     KVM_REG_PPC_MAS1                32
2218  PPC     KVM_REG_PPC_MAS2                64
2219  PPC     KVM_REG_PPC_MAS7_3              64
2220  PPC     KVM_REG_PPC_MAS4                32
2221  PPC     KVM_REG_PPC_MAS6                32
2222  PPC     KVM_REG_PPC_MMUCFG              32
2223  PPC     KVM_REG_PPC_TLB0CFG             32
2224  PPC     KVM_REG_PPC_TLB1CFG             32
2225  PPC     KVM_REG_PPC_TLB2CFG             32
2226  PPC     KVM_REG_PPC_TLB3CFG             32
2227  PPC     KVM_REG_PPC_TLB0PS              32
2228  PPC     KVM_REG_PPC_TLB1PS              32
2229  PPC     KVM_REG_PPC_TLB2PS              32
2230  PPC     KVM_REG_PPC_TLB3PS              32
2231  PPC     KVM_REG_PPC_EPTCFG              32
2232  PPC     KVM_REG_PPC_ICP_STATE           64
2233  PPC     KVM_REG_PPC_VP_STATE            128
2234  PPC     KVM_REG_PPC_TB_OFFSET           64
2235  PPC     KVM_REG_PPC_SPMC1               32
2236  PPC     KVM_REG_PPC_SPMC2               32
2237  PPC     KVM_REG_PPC_IAMR                64
2238  PPC     KVM_REG_PPC_TFHAR               64
2239  PPC     KVM_REG_PPC_TFIAR               64
2240  PPC     KVM_REG_PPC_TEXASR              64
2241  PPC     KVM_REG_PPC_FSCR                64
2242  PPC     KVM_REG_PPC_PSPB                32
2243  PPC     KVM_REG_PPC_EBBHR               64
2244  PPC     KVM_REG_PPC_EBBRR               64
2245  PPC     KVM_REG_PPC_BESCR               64
2246  PPC     KVM_REG_PPC_TAR                 64
2247  PPC     KVM_REG_PPC_DPDES               64
2248  PPC     KVM_REG_PPC_DAWR                64
2249  PPC     KVM_REG_PPC_DAWRX               64
2250  PPC     KVM_REG_PPC_CIABR               64
2251  PPC     KVM_REG_PPC_IC                  64
2252  PPC     KVM_REG_PPC_VTB                 64
2253  PPC     KVM_REG_PPC_CSIGR               64
2254  PPC     KVM_REG_PPC_TACR                64
2255  PPC     KVM_REG_PPC_TCSCR               64
2256  PPC     KVM_REG_PPC_PID                 64
2257  PPC     KVM_REG_PPC_ACOP                64
2258  PPC     KVM_REG_PPC_VRSAVE              32
2259  PPC     KVM_REG_PPC_LPCR                32
2260  PPC     KVM_REG_PPC_LPCR_64             64
2261  PPC     KVM_REG_PPC_PPR                 64
2262  PPC     KVM_REG_PPC_ARCH_COMPAT         32
2263  PPC     KVM_REG_PPC_DABRX               32
2264  PPC     KVM_REG_PPC_WORT                64
2265  PPC	  KVM_REG_PPC_SPRG9               64
2266  PPC	  KVM_REG_PPC_DBSR                32
2267  PPC     KVM_REG_PPC_TIDR                64
2268  PPC     KVM_REG_PPC_PSSCR               64
2269  PPC     KVM_REG_PPC_DEC_EXPIRY          64
2270  PPC     KVM_REG_PPC_PTCR                64
2271  PPC     KVM_REG_PPC_TM_GPR0             64
2272  ...
2273  PPC     KVM_REG_PPC_TM_GPR31            64
2274  PPC     KVM_REG_PPC_TM_VSR0             128
2275  ...
2276  PPC     KVM_REG_PPC_TM_VSR63            128
2277  PPC     KVM_REG_PPC_TM_CR               64
2278  PPC     KVM_REG_PPC_TM_LR               64
2279  PPC     KVM_REG_PPC_TM_CTR              64
2280  PPC     KVM_REG_PPC_TM_FPSCR            64
2281  PPC     KVM_REG_PPC_TM_AMR              64
2282  PPC     KVM_REG_PPC_TM_PPR              64
2283  PPC     KVM_REG_PPC_TM_VRSAVE           64
2284  PPC     KVM_REG_PPC_TM_VSCR             32
2285  PPC     KVM_REG_PPC_TM_DSCR             64
2286  PPC     KVM_REG_PPC_TM_TAR              64
2287  PPC     KVM_REG_PPC_TM_XER              64
2288
2289  MIPS    KVM_REG_MIPS_R0                 64
2290  ...
2291  MIPS    KVM_REG_MIPS_R31                64
2292  MIPS    KVM_REG_MIPS_HI                 64
2293  MIPS    KVM_REG_MIPS_LO                 64
2294  MIPS    KVM_REG_MIPS_PC                 64
2295  MIPS    KVM_REG_MIPS_CP0_INDEX          32
2296  MIPS    KVM_REG_MIPS_CP0_ENTRYLO0       64
2297  MIPS    KVM_REG_MIPS_CP0_ENTRYLO1       64
2298  MIPS    KVM_REG_MIPS_CP0_CONTEXT        64
2299  MIPS    KVM_REG_MIPS_CP0_CONTEXTCONFIG  32
2300  MIPS    KVM_REG_MIPS_CP0_USERLOCAL      64
2301  MIPS    KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2302  MIPS    KVM_REG_MIPS_CP0_PAGEMASK       32
2303  MIPS    KVM_REG_MIPS_CP0_PAGEGRAIN      32
2304  MIPS    KVM_REG_MIPS_CP0_SEGCTL0        64
2305  MIPS    KVM_REG_MIPS_CP0_SEGCTL1        64
2306  MIPS    KVM_REG_MIPS_CP0_SEGCTL2        64
2307  MIPS    KVM_REG_MIPS_CP0_PWBASE         64
2308  MIPS    KVM_REG_MIPS_CP0_PWFIELD        64
2309  MIPS    KVM_REG_MIPS_CP0_PWSIZE         64
2310  MIPS    KVM_REG_MIPS_CP0_WIRED          32
2311  MIPS    KVM_REG_MIPS_CP0_PWCTL          32
2312  MIPS    KVM_REG_MIPS_CP0_HWRENA         32
2313  MIPS    KVM_REG_MIPS_CP0_BADVADDR       64
2314  MIPS    KVM_REG_MIPS_CP0_BADINSTR       32
2315  MIPS    KVM_REG_MIPS_CP0_BADINSTRP      32
2316  MIPS    KVM_REG_MIPS_CP0_COUNT          32
2317  MIPS    KVM_REG_MIPS_CP0_ENTRYHI        64
2318  MIPS    KVM_REG_MIPS_CP0_COMPARE        32
2319  MIPS    KVM_REG_MIPS_CP0_STATUS         32
2320  MIPS    KVM_REG_MIPS_CP0_INTCTL         32
2321  MIPS    KVM_REG_MIPS_CP0_CAUSE          32
2322  MIPS    KVM_REG_MIPS_CP0_EPC            64
2323  MIPS    KVM_REG_MIPS_CP0_PRID           32
2324  MIPS    KVM_REG_MIPS_CP0_EBASE          64
2325  MIPS    KVM_REG_MIPS_CP0_CONFIG         32
2326  MIPS    KVM_REG_MIPS_CP0_CONFIG1        32
2327  MIPS    KVM_REG_MIPS_CP0_CONFIG2        32
2328  MIPS    KVM_REG_MIPS_CP0_CONFIG3        32
2329  MIPS    KVM_REG_MIPS_CP0_CONFIG4        32
2330  MIPS    KVM_REG_MIPS_CP0_CONFIG5        32
2331  MIPS    KVM_REG_MIPS_CP0_CONFIG7        32
2332  MIPS    KVM_REG_MIPS_CP0_XCONTEXT       64
2333  MIPS    KVM_REG_MIPS_CP0_ERROREPC       64
2334  MIPS    KVM_REG_MIPS_CP0_KSCRATCH1      64
2335  MIPS    KVM_REG_MIPS_CP0_KSCRATCH2      64
2336  MIPS    KVM_REG_MIPS_CP0_KSCRATCH3      64
2337  MIPS    KVM_REG_MIPS_CP0_KSCRATCH4      64
2338  MIPS    KVM_REG_MIPS_CP0_KSCRATCH5      64
2339  MIPS    KVM_REG_MIPS_CP0_KSCRATCH6      64
2340  MIPS    KVM_REG_MIPS_CP0_MAAR(0..63)    64
2341  MIPS    KVM_REG_MIPS_COUNT_CTL          64
2342  MIPS    KVM_REG_MIPS_COUNT_RESUME       64
2343  MIPS    KVM_REG_MIPS_COUNT_HZ           64
2344  MIPS    KVM_REG_MIPS_FPR_32(0..31)      32
2345  MIPS    KVM_REG_MIPS_FPR_64(0..31)      64
2346  MIPS    KVM_REG_MIPS_VEC_128(0..31)     128
2347  MIPS    KVM_REG_MIPS_FCR_IR             32
2348  MIPS    KVM_REG_MIPS_FCR_CSR            32
2349  MIPS    KVM_REG_MIPS_MSA_IR             32
2350  MIPS    KVM_REG_MIPS_MSA_CSR            32
2351  ======= =============================== ============
2352
2353ARM registers are mapped using the lower 32 bits.  The upper 16 of that
2354is the register group type, or coprocessor number:
2355
2356ARM core registers have the following id bit patterns::
2357
2358  0x4020 0000 0010 <index into the kvm_regs struct:16>
2359
2360ARM 32-bit CP15 registers have the following id bit patterns::
2361
2362  0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2363
2364ARM 64-bit CP15 registers have the following id bit patterns::
2365
2366  0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2367
2368ARM CCSIDR registers are demultiplexed by CSSELR value::
2369
2370  0x4020 0000 0011 00 <csselr:8>
2371
2372ARM 32-bit VFP control registers have the following id bit patterns::
2373
2374  0x4020 0000 0012 1 <regno:12>
2375
2376ARM 64-bit FP registers have the following id bit patterns::
2377
2378  0x4030 0000 0012 0 <regno:12>
2379
2380ARM firmware pseudo-registers have the following bit pattern::
2381
2382  0x4030 0000 0014 <regno:16>
2383
2384
2385arm64 registers are mapped using the lower 32 bits. The upper 16 of
2386that is the register group type, or coprocessor number:
2387
2388arm64 core/FP-SIMD registers have the following id bit patterns. Note
2389that the size of the access is variable, as the kvm_regs structure
2390contains elements ranging from 32 to 128 bits. The index is a 32bit
2391value in the kvm_regs structure seen as a 32bit array::
2392
2393  0x60x0 0000 0010 <index into the kvm_regs struct:16>
2394
2395Specifically:
2396
2397======================= ========= ===== =======================================
2398    Encoding            Register  Bits  kvm_regs member
2399======================= ========= ===== =======================================
2400  0x6030 0000 0010 0000 X0          64  regs.regs[0]
2401  0x6030 0000 0010 0002 X1          64  regs.regs[1]
2402  ...
2403  0x6030 0000 0010 003c X30         64  regs.regs[30]
2404  0x6030 0000 0010 003e SP          64  regs.sp
2405  0x6030 0000 0010 0040 PC          64  regs.pc
2406  0x6030 0000 0010 0042 PSTATE      64  regs.pstate
2407  0x6030 0000 0010 0044 SP_EL1      64  sp_el1
2408  0x6030 0000 0010 0046 ELR_EL1     64  elr_el1
2409  0x6030 0000 0010 0048 SPSR_EL1    64  spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2410  0x6030 0000 0010 004a SPSR_ABT    64  spsr[KVM_SPSR_ABT]
2411  0x6030 0000 0010 004c SPSR_UND    64  spsr[KVM_SPSR_UND]
2412  0x6030 0000 0010 004e SPSR_IRQ    64  spsr[KVM_SPSR_IRQ]
2413  0x6060 0000 0010 0050 SPSR_FIQ    64  spsr[KVM_SPSR_FIQ]
2414  0x6040 0000 0010 0054 V0         128  fp_regs.vregs[0]    [1]_
2415  0x6040 0000 0010 0058 V1         128  fp_regs.vregs[1]    [1]_
2416  ...
2417  0x6040 0000 0010 00d0 V31        128  fp_regs.vregs[31]   [1]_
2418  0x6020 0000 0010 00d4 FPSR        32  fp_regs.fpsr
2419  0x6020 0000 0010 00d5 FPCR        32  fp_regs.fpcr
2420======================= ========= ===== =======================================
2421
2422.. [1] These encodings are not accepted for SVE-enabled vcpus.  See
2423       KVM_ARM_VCPU_INIT.
2424
2425       The equivalent register content can be accessed via bits [127:0] of
2426       the corresponding SVE Zn registers instead for vcpus that have SVE
2427       enabled (see below).
2428
2429arm64 CCSIDR registers are demultiplexed by CSSELR value::
2430
2431  0x6020 0000 0011 00 <csselr:8>
2432
2433arm64 system registers have the following id bit patterns::
2434
2435  0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2436
2437.. warning::
2438
2439     Two system register IDs do not follow the specified pattern.  These
2440     are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2441     system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively.  These
2442     two had their values accidentally swapped, which means TIMER_CVAL is
2443     derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2444     derived from the register encoding for CNTV_CVAL_EL0.  As this is
2445     API, it must remain this way.
2446
2447arm64 firmware pseudo-registers have the following bit pattern::
2448
2449  0x6030 0000 0014 <regno:16>
2450
2451arm64 SVE registers have the following bit patterns::
2452
2453  0x6080 0000 0015 00 <n:5> <slice:5>   Zn bits[2048*slice + 2047 : 2048*slice]
2454  0x6050 0000 0015 04 <n:4> <slice:5>   Pn bits[256*slice + 255 : 256*slice]
2455  0x6050 0000 0015 060 <slice:5>        FFR bits[256*slice + 255 : 256*slice]
2456  0x6060 0000 0015 ffff                 KVM_REG_ARM64_SVE_VLS pseudo-register
2457
2458Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2459ENOENT.  max_vq is the vcpu's maximum supported vector length in 128-bit
2460quadwords: see [2]_ below.
2461
2462These registers are only accessible on vcpus for which SVE is enabled.
2463See KVM_ARM_VCPU_INIT for details.
2464
2465In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2466accessible until the vcpu's SVE configuration has been finalized
2467using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).  See KVM_ARM_VCPU_INIT
2468and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2469
2470KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2471lengths supported by the vcpu to be discovered and configured by
2472userspace.  When transferred to or from user memory via KVM_GET_ONE_REG
2473or KVM_SET_ONE_REG, the value of this register is of type
2474__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2475follows::
2476
2477  __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2478
2479  if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2480      ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2481		((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2482	/* Vector length vq * 16 bytes supported */
2483  else
2484	/* Vector length vq * 16 bytes not supported */
2485
2486.. [2] The maximum value vq for which the above condition is true is
2487       max_vq.  This is the maximum vector length available to the guest on
2488       this vcpu, and determines which register slices are visible through
2489       this ioctl interface.
2490
2491(See Documentation/arm64/sve.rst for an explanation of the "vq"
2492nomenclature.)
2493
2494KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2495KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2496the host supports.
2497
2498Userspace may subsequently modify it if desired until the vcpu's SVE
2499configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2500
2501Apart from simply removing all vector lengths from the host set that
2502exceed some value, support for arbitrarily chosen sets of vector lengths
2503is hardware-dependent and may not be available.  Attempting to configure
2504an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2505EINVAL.
2506
2507After the vcpu's SVE configuration is finalized, further attempts to
2508write this register will fail with EPERM.
2509
2510
2511MIPS registers are mapped using the lower 32 bits.  The upper 16 of that is
2512the register group type:
2513
2514MIPS core registers (see above) have the following id bit patterns::
2515
2516  0x7030 0000 0000 <reg:16>
2517
2518MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2519patterns depending on whether they're 32-bit or 64-bit registers::
2520
2521  0x7020 0000 0001 00 <reg:5> <sel:3>   (32-bit)
2522  0x7030 0000 0001 00 <reg:5> <sel:3>   (64-bit)
2523
2524Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2525versions of the EntryLo registers regardless of the word size of the host
2526hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2527with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2528the PFNX field starting at bit 30.
2529
2530MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2531patterns::
2532
2533  0x7030 0000 0001 01 <reg:8>
2534
2535MIPS KVM control registers (see above) have the following id bit patterns::
2536
2537  0x7030 0000 0002 <reg:16>
2538
2539MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2540id bit patterns depending on the size of the register being accessed. They are
2541always accessed according to the current guest FPU mode (Status.FR and
2542Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2543if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2544registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2545overlap the FPU registers::
2546
2547  0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2548  0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2549  0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2550
2551MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2552following id bit patterns::
2553
2554  0x7020 0000 0003 01 <0:3> <reg:5>
2555
2556MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2557following id bit patterns::
2558
2559  0x7020 0000 0003 02 <0:3> <reg:5>
2560
2561
25624.69 KVM_GET_ONE_REG
2563--------------------
2564
2565:Capability: KVM_CAP_ONE_REG
2566:Architectures: all
2567:Type: vcpu ioctl
2568:Parameters: struct kvm_one_reg (in and out)
2569:Returns: 0 on success, negative value on failure
2570
2571Errors include:
2572
2573  ======== ============================================================
2574  ENOENT   no such register
2575  EINVAL   invalid register ID, or no such register or used with VMs in
2576           protected virtualization mode on s390
2577  EPERM    (arm64) register access not allowed before vcpu finalization
2578  ======== ============================================================
2579
2580(These error codes are indicative only: do not rely on a specific error
2581code being returned in a specific situation.)
2582
2583This ioctl allows to receive the value of a single register implemented
2584in a vcpu. The register to read is indicated by the "id" field of the
2585kvm_one_reg struct passed in. On success, the register value can be found
2586at the memory location pointed to by "addr".
2587
2588The list of registers accessible using this interface is identical to the
2589list in 4.68.
2590
2591
25924.70 KVM_KVMCLOCK_CTRL
2593----------------------
2594
2595:Capability: KVM_CAP_KVMCLOCK_CTRL
2596:Architectures: Any that implement pvclocks (currently x86 only)
2597:Type: vcpu ioctl
2598:Parameters: None
2599:Returns: 0 on success, -1 on error
2600
2601This ioctl sets a flag accessible to the guest indicating that the specified
2602vCPU has been paused by the host userspace.
2603
2604The host will set a flag in the pvclock structure that is checked from the
2605soft lockup watchdog.  The flag is part of the pvclock structure that is
2606shared between guest and host, specifically the second bit of the flags
2607field of the pvclock_vcpu_time_info structure.  It will be set exclusively by
2608the host and read/cleared exclusively by the guest.  The guest operation of
2609checking and clearing the flag must be an atomic operation so
2610load-link/store-conditional, or equivalent must be used.  There are two cases
2611where the guest will clear the flag: when the soft lockup watchdog timer resets
2612itself or when a soft lockup is detected.  This ioctl can be called any time
2613after pausing the vcpu, but before it is resumed.
2614
2615
26164.71 KVM_SIGNAL_MSI
2617-------------------
2618
2619:Capability: KVM_CAP_SIGNAL_MSI
2620:Architectures: x86 arm arm64
2621:Type: vm ioctl
2622:Parameters: struct kvm_msi (in)
2623:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2624
2625Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2626MSI messages.
2627
2628::
2629
2630  struct kvm_msi {
2631	__u32 address_lo;
2632	__u32 address_hi;
2633	__u32 data;
2634	__u32 flags;
2635	__u32 devid;
2636	__u8  pad[12];
2637  };
2638
2639flags:
2640  KVM_MSI_VALID_DEVID: devid contains a valid value.  The per-VM
2641  KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2642  the device ID.  If this capability is not available, userspace
2643  should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2644
2645If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2646for the device that wrote the MSI message.  For PCI, this is usually a
2647BFD identifier in the lower 16 bits.
2648
2649On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2650feature of KVM_CAP_X2APIC_API capability is enabled.  If it is enabled,
2651address_hi bits 31-8 provide bits 31-8 of the destination id.  Bits 7-0 of
2652address_hi must be zero.
2653
2654
26554.71 KVM_CREATE_PIT2
2656--------------------
2657
2658:Capability: KVM_CAP_PIT2
2659:Architectures: x86
2660:Type: vm ioctl
2661:Parameters: struct kvm_pit_config (in)
2662:Returns: 0 on success, -1 on error
2663
2664Creates an in-kernel device model for the i8254 PIT. This call is only valid
2665after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2666parameters have to be passed::
2667
2668  struct kvm_pit_config {
2669	__u32 flags;
2670	__u32 pad[15];
2671  };
2672
2673Valid flags are::
2674
2675  #define KVM_PIT_SPEAKER_DUMMY     1 /* emulate speaker port stub */
2676
2677PIT timer interrupts may use a per-VM kernel thread for injection. If it
2678exists, this thread will have a name of the following pattern::
2679
2680  kvm-pit/<owner-process-pid>
2681
2682When running a guest with elevated priorities, the scheduling parameters of
2683this thread may have to be adjusted accordingly.
2684
2685This IOCTL replaces the obsolete KVM_CREATE_PIT.
2686
2687
26884.72 KVM_GET_PIT2
2689-----------------
2690
2691:Capability: KVM_CAP_PIT_STATE2
2692:Architectures: x86
2693:Type: vm ioctl
2694:Parameters: struct kvm_pit_state2 (out)
2695:Returns: 0 on success, -1 on error
2696
2697Retrieves the state of the in-kernel PIT model. Only valid after
2698KVM_CREATE_PIT2. The state is returned in the following structure::
2699
2700  struct kvm_pit_state2 {
2701	struct kvm_pit_channel_state channels[3];
2702	__u32 flags;
2703	__u32 reserved[9];
2704  };
2705
2706Valid flags are::
2707
2708  /* disable PIT in HPET legacy mode */
2709  #define KVM_PIT_FLAGS_HPET_LEGACY  0x00000001
2710
2711This IOCTL replaces the obsolete KVM_GET_PIT.
2712
2713
27144.73 KVM_SET_PIT2
2715-----------------
2716
2717:Capability: KVM_CAP_PIT_STATE2
2718:Architectures: x86
2719:Type: vm ioctl
2720:Parameters: struct kvm_pit_state2 (in)
2721:Returns: 0 on success, -1 on error
2722
2723Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2724See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2725
2726This IOCTL replaces the obsolete KVM_SET_PIT.
2727
2728
27294.74 KVM_PPC_GET_SMMU_INFO
2730--------------------------
2731
2732:Capability: KVM_CAP_PPC_GET_SMMU_INFO
2733:Architectures: powerpc
2734:Type: vm ioctl
2735:Parameters: None
2736:Returns: 0 on success, -1 on error
2737
2738This populates and returns a structure describing the features of
2739the "Server" class MMU emulation supported by KVM.
2740This can in turn be used by userspace to generate the appropriate
2741device-tree properties for the guest operating system.
2742
2743The structure contains some global information, followed by an
2744array of supported segment page sizes::
2745
2746      struct kvm_ppc_smmu_info {
2747	     __u64 flags;
2748	     __u32 slb_size;
2749	     __u32 pad;
2750	     struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2751      };
2752
2753The supported flags are:
2754
2755    - KVM_PPC_PAGE_SIZES_REAL:
2756        When that flag is set, guest page sizes must "fit" the backing
2757        store page sizes. When not set, any page size in the list can
2758        be used regardless of how they are backed by userspace.
2759
2760    - KVM_PPC_1T_SEGMENTS
2761        The emulated MMU supports 1T segments in addition to the
2762        standard 256M ones.
2763
2764    - KVM_PPC_NO_HASH
2765	This flag indicates that HPT guests are not supported by KVM,
2766	thus all guests must use radix MMU mode.
2767
2768The "slb_size" field indicates how many SLB entries are supported
2769
2770The "sps" array contains 8 entries indicating the supported base
2771page sizes for a segment in increasing order. Each entry is defined
2772as follow::
2773
2774   struct kvm_ppc_one_seg_page_size {
2775	__u32 page_shift;	/* Base page shift of segment (or 0) */
2776	__u32 slb_enc;		/* SLB encoding for BookS */
2777	struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2778   };
2779
2780An entry with a "page_shift" of 0 is unused. Because the array is
2781organized in increasing order, a lookup can stop when encoutering
2782such an entry.
2783
2784The "slb_enc" field provides the encoding to use in the SLB for the
2785page size. The bits are in positions such as the value can directly
2786be OR'ed into the "vsid" argument of the slbmte instruction.
2787
2788The "enc" array is a list which for each of those segment base page
2789size provides the list of supported actual page sizes (which can be
2790only larger or equal to the base page size), along with the
2791corresponding encoding in the hash PTE. Similarly, the array is
27928 entries sorted by increasing sizes and an entry with a "0" shift
2793is an empty entry and a terminator::
2794
2795   struct kvm_ppc_one_page_size {
2796	__u32 page_shift;	/* Page shift (or 0) */
2797	__u32 pte_enc;		/* Encoding in the HPTE (>>12) */
2798   };
2799
2800The "pte_enc" field provides a value that can OR'ed into the hash
2801PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2802into the hash PTE second double word).
2803
28044.75 KVM_IRQFD
2805--------------
2806
2807:Capability: KVM_CAP_IRQFD
2808:Architectures: x86 s390 arm arm64
2809:Type: vm ioctl
2810:Parameters: struct kvm_irqfd (in)
2811:Returns: 0 on success, -1 on error
2812
2813Allows setting an eventfd to directly trigger a guest interrupt.
2814kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2815kvm_irqfd.gsi specifies the irqchip pin toggled by this event.  When
2816an event is triggered on the eventfd, an interrupt is injected into
2817the guest using the specified gsi pin.  The irqfd is removed using
2818the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2819and kvm_irqfd.gsi.
2820
2821With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2822mechanism allowing emulation of level-triggered, irqfd-based
2823interrupts.  When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2824additional eventfd in the kvm_irqfd.resamplefd field.  When operating
2825in resample mode, posting of an interrupt through kvm_irq.fd asserts
2826the specified gsi in the irqchip.  When the irqchip is resampled, such
2827as from an EOI, the gsi is de-asserted and the user is notified via
2828kvm_irqfd.resamplefd.  It is the user's responsibility to re-queue
2829the interrupt if the device making use of it still requires service.
2830Note that closing the resamplefd is not sufficient to disable the
2831irqfd.  The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2832and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2833
2834On arm/arm64, gsi routing being supported, the following can happen:
2835
2836- in case no routing entry is associated to this gsi, injection fails
2837- in case the gsi is associated to an irqchip routing entry,
2838  irqchip.pin + 32 corresponds to the injected SPI ID.
2839- in case the gsi is associated to an MSI routing entry, the MSI
2840  message and device ID are translated into an LPI (support restricted
2841  to GICv3 ITS in-kernel emulation).
2842
28434.76 KVM_PPC_ALLOCATE_HTAB
2844--------------------------
2845
2846:Capability: KVM_CAP_PPC_ALLOC_HTAB
2847:Architectures: powerpc
2848:Type: vm ioctl
2849:Parameters: Pointer to u32 containing hash table order (in/out)
2850:Returns: 0 on success, -1 on error
2851
2852This requests the host kernel to allocate an MMU hash table for a
2853guest using the PAPR paravirtualization interface.  This only does
2854anything if the kernel is configured to use the Book 3S HV style of
2855virtualization.  Otherwise the capability doesn't exist and the ioctl
2856returns an ENOTTY error.  The rest of this description assumes Book 3S
2857HV.
2858
2859There must be no vcpus running when this ioctl is called; if there
2860are, it will do nothing and return an EBUSY error.
2861
2862The parameter is a pointer to a 32-bit unsigned integer variable
2863containing the order (log base 2) of the desired size of the hash
2864table, which must be between 18 and 46.  On successful return from the
2865ioctl, the value will not be changed by the kernel.
2866
2867If no hash table has been allocated when any vcpu is asked to run
2868(with the KVM_RUN ioctl), the host kernel will allocate a
2869default-sized hash table (16 MB).
2870
2871If this ioctl is called when a hash table has already been allocated,
2872with a different order from the existing hash table, the existing hash
2873table will be freed and a new one allocated.  If this is ioctl is
2874called when a hash table has already been allocated of the same order
2875as specified, the kernel will clear out the existing hash table (zero
2876all HPTEs).  In either case, if the guest is using the virtualized
2877real-mode area (VRMA) facility, the kernel will re-create the VMRA
2878HPTEs on the next KVM_RUN of any vcpu.
2879
28804.77 KVM_S390_INTERRUPT
2881-----------------------
2882
2883:Capability: basic
2884:Architectures: s390
2885:Type: vm ioctl, vcpu ioctl
2886:Parameters: struct kvm_s390_interrupt (in)
2887:Returns: 0 on success, -1 on error
2888
2889Allows to inject an interrupt to the guest. Interrupts can be floating
2890(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2891
2892Interrupt parameters are passed via kvm_s390_interrupt::
2893
2894  struct kvm_s390_interrupt {
2895	__u32 type;
2896	__u32 parm;
2897	__u64 parm64;
2898  };
2899
2900type can be one of the following:
2901
2902KVM_S390_SIGP_STOP (vcpu)
2903    - sigp stop; optional flags in parm
2904KVM_S390_PROGRAM_INT (vcpu)
2905    - program check; code in parm
2906KVM_S390_SIGP_SET_PREFIX (vcpu)
2907    - sigp set prefix; prefix address in parm
2908KVM_S390_RESTART (vcpu)
2909    - restart
2910KVM_S390_INT_CLOCK_COMP (vcpu)
2911    - clock comparator interrupt
2912KVM_S390_INT_CPU_TIMER (vcpu)
2913    - CPU timer interrupt
2914KVM_S390_INT_VIRTIO (vm)
2915    - virtio external interrupt; external interrupt
2916      parameters in parm and parm64
2917KVM_S390_INT_SERVICE (vm)
2918    - sclp external interrupt; sclp parameter in parm
2919KVM_S390_INT_EMERGENCY (vcpu)
2920    - sigp emergency; source cpu in parm
2921KVM_S390_INT_EXTERNAL_CALL (vcpu)
2922    - sigp external call; source cpu in parm
2923KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
2924    - compound value to indicate an
2925      I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2926      I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2927      interruption subclass)
2928KVM_S390_MCHK (vm, vcpu)
2929    - machine check interrupt; cr 14 bits in parm, machine check interrupt
2930      code in parm64 (note that machine checks needing further payload are not
2931      supported by this ioctl)
2932
2933This is an asynchronous vcpu ioctl and can be invoked from any thread.
2934
29354.78 KVM_PPC_GET_HTAB_FD
2936------------------------
2937
2938:Capability: KVM_CAP_PPC_HTAB_FD
2939:Architectures: powerpc
2940:Type: vm ioctl
2941:Parameters: Pointer to struct kvm_get_htab_fd (in)
2942:Returns: file descriptor number (>= 0) on success, -1 on error
2943
2944This returns a file descriptor that can be used either to read out the
2945entries in the guest's hashed page table (HPT), or to write entries to
2946initialize the HPT.  The returned fd can only be written to if the
2947KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2948can only be read if that bit is clear.  The argument struct looks like
2949this::
2950
2951  /* For KVM_PPC_GET_HTAB_FD */
2952  struct kvm_get_htab_fd {
2953	__u64	flags;
2954	__u64	start_index;
2955	__u64	reserved[2];
2956  };
2957
2958  /* Values for kvm_get_htab_fd.flags */
2959  #define KVM_GET_HTAB_BOLTED_ONLY	((__u64)0x1)
2960  #define KVM_GET_HTAB_WRITE		((__u64)0x2)
2961
2962The 'start_index' field gives the index in the HPT of the entry at
2963which to start reading.  It is ignored when writing.
2964
2965Reads on the fd will initially supply information about all
2966"interesting" HPT entries.  Interesting entries are those with the
2967bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2968all entries.  When the end of the HPT is reached, the read() will
2969return.  If read() is called again on the fd, it will start again from
2970the beginning of the HPT, but will only return HPT entries that have
2971changed since they were last read.
2972
2973Data read or written is structured as a header (8 bytes) followed by a
2974series of valid HPT entries (16 bytes) each.  The header indicates how
2975many valid HPT entries there are and how many invalid entries follow
2976the valid entries.  The invalid entries are not represented explicitly
2977in the stream.  The header format is::
2978
2979  struct kvm_get_htab_header {
2980	__u32	index;
2981	__u16	n_valid;
2982	__u16	n_invalid;
2983  };
2984
2985Writes to the fd create HPT entries starting at the index given in the
2986header; first 'n_valid' valid entries with contents from the data
2987written, then 'n_invalid' invalid entries, invalidating any previously
2988valid entries found.
2989
29904.79 KVM_CREATE_DEVICE
2991----------------------
2992
2993:Capability: KVM_CAP_DEVICE_CTRL
2994:Type: vm ioctl
2995:Parameters: struct kvm_create_device (in/out)
2996:Returns: 0 on success, -1 on error
2997
2998Errors:
2999
3000  ======  =======================================================
3001  ENODEV  The device type is unknown or unsupported
3002  EEXIST  Device already created, and this type of device may not
3003          be instantiated multiple times
3004  ======  =======================================================
3005
3006  Other error conditions may be defined by individual device types or
3007  have their standard meanings.
3008
3009Creates an emulated device in the kernel.  The file descriptor returned
3010in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3011
3012If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3013device type is supported (not necessarily whether it can be created
3014in the current vm).
3015
3016Individual devices should not define flags.  Attributes should be used
3017for specifying any behavior that is not implied by the device type
3018number.
3019
3020::
3021
3022  struct kvm_create_device {
3023	__u32	type;	/* in: KVM_DEV_TYPE_xxx */
3024	__u32	fd;	/* out: device handle */
3025	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
3026  };
3027
30284.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3029--------------------------------------------
3030
3031:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3032             KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3033:Type: device ioctl, vm ioctl, vcpu ioctl
3034:Parameters: struct kvm_device_attr
3035:Returns: 0 on success, -1 on error
3036
3037Errors:
3038
3039  =====   =============================================================
3040  ENXIO   The group or attribute is unknown/unsupported for this device
3041          or hardware support is missing.
3042  EPERM   The attribute cannot (currently) be accessed this way
3043          (e.g. read-only attribute, or attribute that only makes
3044          sense when the device is in a different state)
3045  =====   =============================================================
3046
3047  Other error conditions may be defined by individual device types.
3048
3049Gets/sets a specified piece of device configuration and/or state.  The
3050semantics are device-specific.  See individual device documentation in
3051the "devices" directory.  As with ONE_REG, the size of the data
3052transferred is defined by the particular attribute.
3053
3054::
3055
3056  struct kvm_device_attr {
3057	__u32	flags;		/* no flags currently defined */
3058	__u32	group;		/* device-defined */
3059	__u64	attr;		/* group-defined */
3060	__u64	addr;		/* userspace address of attr data */
3061  };
3062
30634.81 KVM_HAS_DEVICE_ATTR
3064------------------------
3065
3066:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3067	     KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3068:Type: device ioctl, vm ioctl, vcpu ioctl
3069:Parameters: struct kvm_device_attr
3070:Returns: 0 on success, -1 on error
3071
3072Errors:
3073
3074  =====   =============================================================
3075  ENXIO   The group or attribute is unknown/unsupported for this device
3076          or hardware support is missing.
3077  =====   =============================================================
3078
3079Tests whether a device supports a particular attribute.  A successful
3080return indicates the attribute is implemented.  It does not necessarily
3081indicate that the attribute can be read or written in the device's
3082current state.  "addr" is ignored.
3083
30844.82 KVM_ARM_VCPU_INIT
3085----------------------
3086
3087:Capability: basic
3088:Architectures: arm, arm64
3089:Type: vcpu ioctl
3090:Parameters: struct kvm_vcpu_init (in)
3091:Returns: 0 on success; -1 on error
3092
3093Errors:
3094
3095  ======     =================================================================
3096  EINVAL     the target is unknown, or the combination of features is invalid.
3097  ENOENT     a features bit specified is unknown.
3098  ======     =================================================================
3099
3100This tells KVM what type of CPU to present to the guest, and what
3101optional features it should have.  This will cause a reset of the cpu
3102registers to their initial values.  If this is not called, KVM_RUN will
3103return ENOEXEC for that vcpu.
3104
3105Note that because some registers reflect machine topology, all vcpus
3106should be created before this ioctl is invoked.
3107
3108Userspace can call this function multiple times for a given vcpu, including
3109after the vcpu has been run. This will reset the vcpu to its initial
3110state. All calls to this function after the initial call must use the same
3111target and same set of feature flags, otherwise EINVAL will be returned.
3112
3113Possible features:
3114
3115	- KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3116	  Depends on KVM_CAP_ARM_PSCI.  If not set, the CPU will be powered on
3117	  and execute guest code when KVM_RUN is called.
3118	- KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3119	  Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3120	- KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3121          backward compatible with v0.2) for the CPU.
3122	  Depends on KVM_CAP_ARM_PSCI_0_2.
3123	- KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3124	  Depends on KVM_CAP_ARM_PMU_V3.
3125
3126	- KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3127	  for arm64 only.
3128	  Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3129	  If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3130	  both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3131	  KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3132	  requested.
3133
3134	- KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3135	  for arm64 only.
3136	  Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3137	  If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3138	  both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3139	  KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3140	  requested.
3141
3142	- KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3143	  Depends on KVM_CAP_ARM_SVE.
3144	  Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3145
3146	   * After KVM_ARM_VCPU_INIT:
3147
3148	      - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3149	        initial value of this pseudo-register indicates the best set of
3150	        vector lengths possible for a vcpu on this host.
3151
3152	   * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3153
3154	      - KVM_RUN and KVM_GET_REG_LIST are not available;
3155
3156	      - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3157	        the scalable archietctural SVE registers
3158	        KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3159	        KVM_REG_ARM64_SVE_FFR;
3160
3161	      - KVM_REG_ARM64_SVE_VLS may optionally be written using
3162	        KVM_SET_ONE_REG, to modify the set of vector lengths available
3163	        for the vcpu.
3164
3165	   * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3166
3167	      - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3168	        no longer be written using KVM_SET_ONE_REG.
3169
31704.83 KVM_ARM_PREFERRED_TARGET
3171-----------------------------
3172
3173:Capability: basic
3174:Architectures: arm, arm64
3175:Type: vm ioctl
3176:Parameters: struct kvm_vcpu_init (out)
3177:Returns: 0 on success; -1 on error
3178
3179Errors:
3180
3181  ======     ==========================================
3182  ENODEV     no preferred target available for the host
3183  ======     ==========================================
3184
3185This queries KVM for preferred CPU target type which can be emulated
3186by KVM on underlying host.
3187
3188The ioctl returns struct kvm_vcpu_init instance containing information
3189about preferred CPU target type and recommended features for it.  The
3190kvm_vcpu_init->features bitmap returned will have feature bits set if
3191the preferred target recommends setting these features, but this is
3192not mandatory.
3193
3194The information returned by this ioctl can be used to prepare an instance
3195of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3196VCPU matching underlying host.
3197
3198
31994.84 KVM_GET_REG_LIST
3200---------------------
3201
3202:Capability: basic
3203:Architectures: arm, arm64, mips
3204:Type: vcpu ioctl
3205:Parameters: struct kvm_reg_list (in/out)
3206:Returns: 0 on success; -1 on error
3207
3208Errors:
3209
3210  =====      ==============================================================
3211  E2BIG      the reg index list is too big to fit in the array specified by
3212             the user (the number required will be written into n).
3213  =====      ==============================================================
3214
3215::
3216
3217  struct kvm_reg_list {
3218	__u64 n; /* number of registers in reg[] */
3219	__u64 reg[0];
3220  };
3221
3222This ioctl returns the guest registers that are supported for the
3223KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3224
3225
32264.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3227-----------------------------------------
3228
3229:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3230:Architectures: arm, arm64
3231:Type: vm ioctl
3232:Parameters: struct kvm_arm_device_address (in)
3233:Returns: 0 on success, -1 on error
3234
3235Errors:
3236
3237  ======  ============================================
3238  ENODEV  The device id is unknown
3239  ENXIO   Device not supported on current system
3240  EEXIST  Address already set
3241  E2BIG   Address outside guest physical address space
3242  EBUSY   Address overlaps with other device range
3243  ======  ============================================
3244
3245::
3246
3247  struct kvm_arm_device_addr {
3248	__u64 id;
3249	__u64 addr;
3250  };
3251
3252Specify a device address in the guest's physical address space where guests
3253can access emulated or directly exposed devices, which the host kernel needs
3254to know about. The id field is an architecture specific identifier for a
3255specific device.
3256
3257ARM/arm64 divides the id field into two parts, a device id and an
3258address type id specific to the individual device::
3259
3260  bits:  | 63        ...       32 | 31    ...    16 | 15    ...    0 |
3261  field: |        0x00000000      |     device id   |  addr type id  |
3262
3263ARM/arm64 currently only require this when using the in-kernel GIC
3264support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3265as the device id.  When setting the base address for the guest's
3266mapping of the VGIC virtual CPU and distributor interface, the ioctl
3267must be called after calling KVM_CREATE_IRQCHIP, but before calling
3268KVM_RUN on any of the VCPUs.  Calling this ioctl twice for any of the
3269base addresses will return -EEXIST.
3270
3271Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3272should be used instead.
3273
3274
32754.86 KVM_PPC_RTAS_DEFINE_TOKEN
3276------------------------------
3277
3278:Capability: KVM_CAP_PPC_RTAS
3279:Architectures: ppc
3280:Type: vm ioctl
3281:Parameters: struct kvm_rtas_token_args
3282:Returns: 0 on success, -1 on error
3283
3284Defines a token value for a RTAS (Run Time Abstraction Services)
3285service in order to allow it to be handled in the kernel.  The
3286argument struct gives the name of the service, which must be the name
3287of a service that has a kernel-side implementation.  If the token
3288value is non-zero, it will be associated with that service, and
3289subsequent RTAS calls by the guest specifying that token will be
3290handled by the kernel.  If the token value is 0, then any token
3291associated with the service will be forgotten, and subsequent RTAS
3292calls by the guest for that service will be passed to userspace to be
3293handled.
3294
32954.87 KVM_SET_GUEST_DEBUG
3296------------------------
3297
3298:Capability: KVM_CAP_SET_GUEST_DEBUG
3299:Architectures: x86, s390, ppc, arm64
3300:Type: vcpu ioctl
3301:Parameters: struct kvm_guest_debug (in)
3302:Returns: 0 on success; -1 on error
3303
3304::
3305
3306  struct kvm_guest_debug {
3307       __u32 control;
3308       __u32 pad;
3309       struct kvm_guest_debug_arch arch;
3310  };
3311
3312Set up the processor specific debug registers and configure vcpu for
3313handling guest debug events. There are two parts to the structure, the
3314first a control bitfield indicates the type of debug events to handle
3315when running. Common control bits are:
3316
3317  - KVM_GUESTDBG_ENABLE:        guest debugging is enabled
3318  - KVM_GUESTDBG_SINGLESTEP:    the next run should single-step
3319
3320The top 16 bits of the control field are architecture specific control
3321flags which can include the following:
3322
3323  - KVM_GUESTDBG_USE_SW_BP:     using software breakpoints [x86, arm64]
3324  - KVM_GUESTDBG_USE_HW_BP:     using hardware breakpoints [x86, s390, arm64]
3325  - KVM_GUESTDBG_INJECT_DB:     inject DB type exception [x86]
3326  - KVM_GUESTDBG_INJECT_BP:     inject BP type exception [x86]
3327  - KVM_GUESTDBG_EXIT_PENDING:  trigger an immediate guest exit [s390]
3328
3329For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3330are enabled in memory so we need to ensure breakpoint exceptions are
3331correctly trapped and the KVM run loop exits at the breakpoint and not
3332running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3333we need to ensure the guest vCPUs architecture specific registers are
3334updated to the correct (supplied) values.
3335
3336The second part of the structure is architecture specific and
3337typically contains a set of debug registers.
3338
3339For arm64 the number of debug registers is implementation defined and
3340can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3341KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3342indicating the number of supported registers.
3343
3344For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3345the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3346
3347When debug events exit the main run loop with the reason
3348KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3349structure containing architecture specific debug information.
3350
33514.88 KVM_GET_EMULATED_CPUID
3352---------------------------
3353
3354:Capability: KVM_CAP_EXT_EMUL_CPUID
3355:Architectures: x86
3356:Type: system ioctl
3357:Parameters: struct kvm_cpuid2 (in/out)
3358:Returns: 0 on success, -1 on error
3359
3360::
3361
3362  struct kvm_cpuid2 {
3363	__u32 nent;
3364	__u32 flags;
3365	struct kvm_cpuid_entry2 entries[0];
3366  };
3367
3368The member 'flags' is used for passing flags from userspace.
3369
3370::
3371
3372  #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
3373  #define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1) /* deprecated */
3374  #define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2) /* deprecated */
3375
3376  struct kvm_cpuid_entry2 {
3377	__u32 function;
3378	__u32 index;
3379	__u32 flags;
3380	__u32 eax;
3381	__u32 ebx;
3382	__u32 ecx;
3383	__u32 edx;
3384	__u32 padding[3];
3385  };
3386
3387This ioctl returns x86 cpuid features which are emulated by
3388kvm.Userspace can use the information returned by this ioctl to query
3389which features are emulated by kvm instead of being present natively.
3390
3391Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3392structure with the 'nent' field indicating the number of entries in
3393the variable-size array 'entries'. If the number of entries is too low
3394to describe the cpu capabilities, an error (E2BIG) is returned. If the
3395number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3396is returned. If the number is just right, the 'nent' field is adjusted
3397to the number of valid entries in the 'entries' array, which is then
3398filled.
3399
3400The entries returned are the set CPUID bits of the respective features
3401which kvm emulates, as returned by the CPUID instruction, with unknown
3402or unsupported feature bits cleared.
3403
3404Features like x2apic, for example, may not be present in the host cpu
3405but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3406emulated efficiently and thus not included here.
3407
3408The fields in each entry are defined as follows:
3409
3410  function:
3411	 the eax value used to obtain the entry
3412  index:
3413	 the ecx value used to obtain the entry (for entries that are
3414         affected by ecx)
3415  flags:
3416    an OR of zero or more of the following:
3417
3418        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3419           if the index field is valid
3420
3421   eax, ebx, ecx, edx:
3422
3423         the values returned by the cpuid instruction for
3424         this function/index combination
3425
34264.89 KVM_S390_MEM_OP
3427--------------------
3428
3429:Capability: KVM_CAP_S390_MEM_OP
3430:Architectures: s390
3431:Type: vcpu ioctl
3432:Parameters: struct kvm_s390_mem_op (in)
3433:Returns: = 0 on success,
3434          < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3435          > 0 if an exception occurred while walking the page tables
3436
3437Read or write data from/to the logical (virtual) memory of a VCPU.
3438
3439Parameters are specified via the following structure::
3440
3441  struct kvm_s390_mem_op {
3442	__u64 gaddr;		/* the guest address */
3443	__u64 flags;		/* flags */
3444	__u32 size;		/* amount of bytes */
3445	__u32 op;		/* type of operation */
3446	__u64 buf;		/* buffer in userspace */
3447	__u8 ar;		/* the access register number */
3448	__u8 reserved[31];	/* should be set to 0 */
3449  };
3450
3451The type of operation is specified in the "op" field. It is either
3452KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
3453KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
3454KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
3455whether the corresponding memory access would create an access exception
3456(without touching the data in the memory at the destination). In case an
3457access exception occurred while walking the MMU tables of the guest, the
3458ioctl returns a positive error number to indicate the type of exception.
3459This exception is also raised directly at the corresponding VCPU if the
3460flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
3461
3462The start address of the memory region has to be specified in the "gaddr"
3463field, and the length of the region in the "size" field (which must not
3464be 0). The maximum value for "size" can be obtained by checking the
3465KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3466userspace application where the read data should be written to for
3467KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written is
3468stored for a KVM_S390_MEMOP_LOGICAL_WRITE. When KVM_S390_MEMOP_F_CHECK_ONLY
3469is specified, "buf" is unused and can be NULL. "ar" designates the access
3470register number to be used; the valid range is 0..15.
3471
3472The "reserved" field is meant for future extensions. It is not used by
3473KVM with the currently defined set of flags.
3474
34754.90 KVM_S390_GET_SKEYS
3476-----------------------
3477
3478:Capability: KVM_CAP_S390_SKEYS
3479:Architectures: s390
3480:Type: vm ioctl
3481:Parameters: struct kvm_s390_skeys
3482:Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
3483          keys, negative value on error
3484
3485This ioctl is used to get guest storage key values on the s390
3486architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3487
3488  struct kvm_s390_skeys {
3489	__u64 start_gfn;
3490	__u64 count;
3491	__u64 skeydata_addr;
3492	__u32 flags;
3493	__u32 reserved[9];
3494  };
3495
3496The start_gfn field is the number of the first guest frame whose storage keys
3497you want to get.
3498
3499The count field is the number of consecutive frames (starting from start_gfn)
3500whose storage keys to get. The count field must be at least 1 and the maximum
3501allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3502will cause the ioctl to return -EINVAL.
3503
3504The skeydata_addr field is the address to a buffer large enough to hold count
3505bytes. This buffer will be filled with storage key data by the ioctl.
3506
35074.91 KVM_S390_SET_SKEYS
3508-----------------------
3509
3510:Capability: KVM_CAP_S390_SKEYS
3511:Architectures: s390
3512:Type: vm ioctl
3513:Parameters: struct kvm_s390_skeys
3514:Returns: 0 on success, negative value on error
3515
3516This ioctl is used to set guest storage key values on the s390
3517architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3518See section on KVM_S390_GET_SKEYS for struct definition.
3519
3520The start_gfn field is the number of the first guest frame whose storage keys
3521you want to set.
3522
3523The count field is the number of consecutive frames (starting from start_gfn)
3524whose storage keys to get. The count field must be at least 1 and the maximum
3525allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
3526will cause the ioctl to return -EINVAL.
3527
3528The skeydata_addr field is the address to a buffer containing count bytes of
3529storage keys. Each byte in the buffer will be set as the storage key for a
3530single frame starting at start_gfn for count frames.
3531
3532Note: If any architecturally invalid key value is found in the given data then
3533the ioctl will return -EINVAL.
3534
35354.92 KVM_S390_IRQ
3536-----------------
3537
3538:Capability: KVM_CAP_S390_INJECT_IRQ
3539:Architectures: s390
3540:Type: vcpu ioctl
3541:Parameters: struct kvm_s390_irq (in)
3542:Returns: 0 on success, -1 on error
3543
3544Errors:
3545
3546
3547  ======  =================================================================
3548  EINVAL  interrupt type is invalid
3549          type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3550          type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3551          than the maximum of VCPUs
3552  EBUSY   type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3553          type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3554          type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3555          is already pending
3556  ======  =================================================================
3557
3558Allows to inject an interrupt to the guest.
3559
3560Using struct kvm_s390_irq as a parameter allows
3561to inject additional payload which is not
3562possible via KVM_S390_INTERRUPT.
3563
3564Interrupt parameters are passed via kvm_s390_irq::
3565
3566  struct kvm_s390_irq {
3567	__u64 type;
3568	union {
3569		struct kvm_s390_io_info io;
3570		struct kvm_s390_ext_info ext;
3571		struct kvm_s390_pgm_info pgm;
3572		struct kvm_s390_emerg_info emerg;
3573		struct kvm_s390_extcall_info extcall;
3574		struct kvm_s390_prefix_info prefix;
3575		struct kvm_s390_stop_info stop;
3576		struct kvm_s390_mchk_info mchk;
3577		char reserved[64];
3578	} u;
3579  };
3580
3581type can be one of the following:
3582
3583- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3584- KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3585- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3586- KVM_S390_RESTART - restart; no parameters
3587- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3588- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3589- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3590- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3591- KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3592
3593This is an asynchronous vcpu ioctl and can be invoked from any thread.
3594
35954.94 KVM_S390_GET_IRQ_STATE
3596---------------------------
3597
3598:Capability: KVM_CAP_S390_IRQ_STATE
3599:Architectures: s390
3600:Type: vcpu ioctl
3601:Parameters: struct kvm_s390_irq_state (out)
3602:Returns: >= number of bytes copied into buffer,
3603          -EINVAL if buffer size is 0,
3604          -ENOBUFS if buffer size is too small to fit all pending interrupts,
3605          -EFAULT if the buffer address was invalid
3606
3607This ioctl allows userspace to retrieve the complete state of all currently
3608pending interrupts in a single buffer. Use cases include migration
3609and introspection. The parameter structure contains the address of a
3610userspace buffer and its length::
3611
3612  struct kvm_s390_irq_state {
3613	__u64 buf;
3614	__u32 flags;        /* will stay unused for compatibility reasons */
3615	__u32 len;
3616	__u32 reserved[4];  /* will stay unused for compatibility reasons */
3617  };
3618
3619Userspace passes in the above struct and for each pending interrupt a
3620struct kvm_s390_irq is copied to the provided buffer.
3621
3622The structure contains a flags and a reserved field for future extensions. As
3623the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
3624reserved, these fields can not be used in the future without breaking
3625compatibility.
3626
3627If -ENOBUFS is returned the buffer provided was too small and userspace
3628may retry with a bigger buffer.
3629
36304.95 KVM_S390_SET_IRQ_STATE
3631---------------------------
3632
3633:Capability: KVM_CAP_S390_IRQ_STATE
3634:Architectures: s390
3635:Type: vcpu ioctl
3636:Parameters: struct kvm_s390_irq_state (in)
3637:Returns: 0 on success,
3638          -EFAULT if the buffer address was invalid,
3639          -EINVAL for an invalid buffer length (see below),
3640          -EBUSY if there were already interrupts pending,
3641          errors occurring when actually injecting the
3642          interrupt. See KVM_S390_IRQ.
3643
3644This ioctl allows userspace to set the complete state of all cpu-local
3645interrupts currently pending for the vcpu. It is intended for restoring
3646interrupt state after a migration. The input parameter is a userspace buffer
3647containing a struct kvm_s390_irq_state::
3648
3649  struct kvm_s390_irq_state {
3650	__u64 buf;
3651	__u32 flags;        /* will stay unused for compatibility reasons */
3652	__u32 len;
3653	__u32 reserved[4];  /* will stay unused for compatibility reasons */
3654  };
3655
3656The restrictions for flags and reserved apply as well.
3657(see KVM_S390_GET_IRQ_STATE)
3658
3659The userspace memory referenced by buf contains a struct kvm_s390_irq
3660for each interrupt to be injected into the guest.
3661If one of the interrupts could not be injected for some reason the
3662ioctl aborts.
3663
3664len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3665and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3666which is the maximum number of possibly pending cpu-local interrupts.
3667
36684.96 KVM_SMI
3669------------
3670
3671:Capability: KVM_CAP_X86_SMM
3672:Architectures: x86
3673:Type: vcpu ioctl
3674:Parameters: none
3675:Returns: 0 on success, -1 on error
3676
3677Queues an SMI on the thread's vcpu.
3678
36794.97 KVM_CAP_PPC_MULTITCE
3680-------------------------
3681
3682:Capability: KVM_CAP_PPC_MULTITCE
3683:Architectures: ppc
3684:Type: vm
3685
3686This capability means the kernel is capable of handling hypercalls
3687H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
3688space. This significantly accelerates DMA operations for PPC KVM guests.
3689User space should expect that its handlers for these hypercalls
3690are not going to be called if user space previously registered LIOBN
3691in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
3692
3693In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
3694user space might have to advertise it for the guest. For example,
3695IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
3696present in the "ibm,hypertas-functions" device-tree property.
3697
3698The hypercalls mentioned above may or may not be processed successfully
3699in the kernel based fast path. If they can not be handled by the kernel,
3700they will get passed on to user space. So user space still has to have
3701an implementation for these despite the in kernel acceleration.
3702
3703This capability is always enabled.
3704
37054.98 KVM_CREATE_SPAPR_TCE_64
3706----------------------------
3707
3708:Capability: KVM_CAP_SPAPR_TCE_64
3709:Architectures: powerpc
3710:Type: vm ioctl
3711:Parameters: struct kvm_create_spapr_tce_64 (in)
3712:Returns: file descriptor for manipulating the created TCE table
3713
3714This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
3715windows, described in 4.62 KVM_CREATE_SPAPR_TCE
3716
3717This capability uses extended struct in ioctl interface::
3718
3719  /* for KVM_CAP_SPAPR_TCE_64 */
3720  struct kvm_create_spapr_tce_64 {
3721	__u64 liobn;
3722	__u32 page_shift;
3723	__u32 flags;
3724	__u64 offset;	/* in pages */
3725	__u64 size; 	/* in pages */
3726  };
3727
3728The aim of extension is to support an additional bigger DMA window with
3729a variable page size.
3730KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
3731a bus offset of the corresponding DMA window, @size and @offset are numbers
3732of IOMMU pages.
3733
3734@flags are not used at the moment.
3735
3736The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
3737
37384.99 KVM_REINJECT_CONTROL
3739-------------------------
3740
3741:Capability: KVM_CAP_REINJECT_CONTROL
3742:Architectures: x86
3743:Type: vm ioctl
3744:Parameters: struct kvm_reinject_control (in)
3745:Returns: 0 on success,
3746         -EFAULT if struct kvm_reinject_control cannot be read,
3747         -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
3748
3749i8254 (PIT) has two modes, reinject and !reinject.  The default is reinject,
3750where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
3751vector(s) that i8254 injects.  Reinject mode dequeues a tick and injects its
3752interrupt whenever there isn't a pending interrupt from i8254.
3753!reinject mode injects an interrupt as soon as a tick arrives.
3754
3755::
3756
3757  struct kvm_reinject_control {
3758	__u8 pit_reinject;
3759	__u8 reserved[31];
3760  };
3761
3762pit_reinject = 0 (!reinject mode) is recommended, unless running an old
3763operating system that uses the PIT for timing (e.g. Linux 2.4.x).
3764
37654.100 KVM_PPC_CONFIGURE_V3_MMU
3766------------------------------
3767
3768:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
3769:Architectures: ppc
3770:Type: vm ioctl
3771:Parameters: struct kvm_ppc_mmuv3_cfg (in)
3772:Returns: 0 on success,
3773         -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
3774         -EINVAL if the configuration is invalid
3775
3776This ioctl controls whether the guest will use radix or HPT (hashed
3777page table) translation, and sets the pointer to the process table for
3778the guest.
3779
3780::
3781
3782  struct kvm_ppc_mmuv3_cfg {
3783	__u64	flags;
3784	__u64	process_table;
3785  };
3786
3787There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
3788KVM_PPC_MMUV3_GTSE.  KVM_PPC_MMUV3_RADIX, if set, configures the guest
3789to use radix tree translation, and if clear, to use HPT translation.
3790KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
3791to be able to use the global TLB and SLB invalidation instructions;
3792if clear, the guest may not use these instructions.
3793
3794The process_table field specifies the address and size of the guest
3795process table, which is in the guest's space.  This field is formatted
3796as the second doubleword of the partition table entry, as defined in
3797the Power ISA V3.00, Book III section 5.7.6.1.
3798
37994.101 KVM_PPC_GET_RMMU_INFO
3800---------------------------
3801
3802:Capability: KVM_CAP_PPC_RADIX_MMU
3803:Architectures: ppc
3804:Type: vm ioctl
3805:Parameters: struct kvm_ppc_rmmu_info (out)
3806:Returns: 0 on success,
3807	 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
3808	 -EINVAL if no useful information can be returned
3809
3810This ioctl returns a structure containing two things: (a) a list
3811containing supported radix tree geometries, and (b) a list that maps
3812page sizes to put in the "AP" (actual page size) field for the tlbie
3813(TLB invalidate entry) instruction.
3814
3815::
3816
3817  struct kvm_ppc_rmmu_info {
3818	struct kvm_ppc_radix_geom {
3819		__u8	page_shift;
3820		__u8	level_bits[4];
3821		__u8	pad[3];
3822	}	geometries[8];
3823	__u32	ap_encodings[8];
3824  };
3825
3826The geometries[] field gives up to 8 supported geometries for the
3827radix page table, in terms of the log base 2 of the smallest page
3828size, and the number of bits indexed at each level of the tree, from
3829the PTE level up to the PGD level in that order.  Any unused entries
3830will have 0 in the page_shift field.
3831
3832The ap_encodings gives the supported page sizes and their AP field
3833encodings, encoded with the AP value in the top 3 bits and the log
3834base 2 of the page size in the bottom 6 bits.
3835
38364.102 KVM_PPC_RESIZE_HPT_PREPARE
3837--------------------------------
3838
3839:Capability: KVM_CAP_SPAPR_RESIZE_HPT
3840:Architectures: powerpc
3841:Type: vm ioctl
3842:Parameters: struct kvm_ppc_resize_hpt (in)
3843:Returns: 0 on successful completion,
3844	 >0 if a new HPT is being prepared, the value is an estimated
3845         number of milliseconds until preparation is complete,
3846         -EFAULT if struct kvm_reinject_control cannot be read,
3847	 -EINVAL if the supplied shift or flags are invalid,
3848	 -ENOMEM if unable to allocate the new HPT,
3849	 -ENOSPC if there was a hash collision
3850
3851::
3852
3853  struct kvm_ppc_rmmu_info {
3854	struct kvm_ppc_radix_geom {
3855		__u8	page_shift;
3856		__u8	level_bits[4];
3857		__u8	pad[3];
3858	}	geometries[8];
3859	__u32	ap_encodings[8];
3860  };
3861
3862The geometries[] field gives up to 8 supported geometries for the
3863radix page table, in terms of the log base 2 of the smallest page
3864size, and the number of bits indexed at each level of the tree, from
3865the PTE level up to the PGD level in that order.  Any unused entries
3866will have 0 in the page_shift field.
3867
3868The ap_encodings gives the supported page sizes and their AP field
3869encodings, encoded with the AP value in the top 3 bits and the log
3870base 2 of the page size in the bottom 6 bits.
3871
38724.102 KVM_PPC_RESIZE_HPT_PREPARE
3873--------------------------------
3874
3875:Capability: KVM_CAP_SPAPR_RESIZE_HPT
3876:Architectures: powerpc
3877:Type: vm ioctl
3878:Parameters: struct kvm_ppc_resize_hpt (in)
3879:Returns: 0 on successful completion,
3880	 >0 if a new HPT is being prepared, the value is an estimated
3881         number of milliseconds until preparation is complete,
3882         -EFAULT if struct kvm_reinject_control cannot be read,
3883	 -EINVAL if the supplied shift or flags are invalid,when moving existing
3884         HPT entries to the new HPT,
3885	 -EIO on other error conditions
3886
3887Used to implement the PAPR extension for runtime resizing of a guest's
3888Hashed Page Table (HPT).  Specifically this starts, stops or monitors
3889the preparation of a new potential HPT for the guest, essentially
3890implementing the H_RESIZE_HPT_PREPARE hypercall.
3891
3892If called with shift > 0 when there is no pending HPT for the guest,
3893this begins preparation of a new pending HPT of size 2^(shift) bytes.
3894It then returns a positive integer with the estimated number of
3895milliseconds until preparation is complete.
3896
3897If called when there is a pending HPT whose size does not match that
3898requested in the parameters, discards the existing pending HPT and
3899creates a new one as above.
3900
3901If called when there is a pending HPT of the size requested, will:
3902
3903  * If preparation of the pending HPT is already complete, return 0
3904  * If preparation of the pending HPT has failed, return an error
3905    code, then discard the pending HPT.
3906  * If preparation of the pending HPT is still in progress, return an
3907    estimated number of milliseconds until preparation is complete.
3908
3909If called with shift == 0, discards any currently pending HPT and
3910returns 0 (i.e. cancels any in-progress preparation).
3911
3912flags is reserved for future expansion, currently setting any bits in
3913flags will result in an -EINVAL.
3914
3915Normally this will be called repeatedly with the same parameters until
3916it returns <= 0.  The first call will initiate preparation, subsequent
3917ones will monitor preparation until it completes or fails.
3918
3919::
3920
3921  struct kvm_ppc_resize_hpt {
3922	__u64 flags;
3923	__u32 shift;
3924	__u32 pad;
3925  };
3926
39274.103 KVM_PPC_RESIZE_HPT_COMMIT
3928-------------------------------
3929
3930:Capability: KVM_CAP_SPAPR_RESIZE_HPT
3931:Architectures: powerpc
3932:Type: vm ioctl
3933:Parameters: struct kvm_ppc_resize_hpt (in)
3934:Returns: 0 on successful completion,
3935         -EFAULT if struct kvm_reinject_control cannot be read,
3936	 -EINVAL if the supplied shift or flags are invalid,
3937	 -ENXIO is there is no pending HPT, or the pending HPT doesn't
3938         have the requested size,
3939	 -EBUSY if the pending HPT is not fully prepared,
3940	 -ENOSPC if there was a hash collision when moving existing
3941         HPT entries to the new HPT,
3942	 -EIO on other error conditions
3943
3944Used to implement the PAPR extension for runtime resizing of a guest's
3945Hashed Page Table (HPT).  Specifically this requests that the guest be
3946transferred to working with the new HPT, essentially implementing the
3947H_RESIZE_HPT_COMMIT hypercall.
3948
3949This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
3950returned 0 with the same parameters.  In other cases
3951KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
3952-EBUSY, though others may be possible if the preparation was started,
3953but failed).
3954
3955This will have undefined effects on the guest if it has not already
3956placed itself in a quiescent state where no vcpu will make MMU enabled
3957memory accesses.
3958
3959On succsful completion, the pending HPT will become the guest's active
3960HPT and the previous HPT will be discarded.
3961
3962On failure, the guest will still be operating on its previous HPT.
3963
3964::
3965
3966  struct kvm_ppc_resize_hpt {
3967	__u64 flags;
3968	__u32 shift;
3969	__u32 pad;
3970  };
3971
39724.104 KVM_X86_GET_MCE_CAP_SUPPORTED
3973-----------------------------------
3974
3975:Capability: KVM_CAP_MCE
3976:Architectures: x86
3977:Type: system ioctl
3978:Parameters: u64 mce_cap (out)
3979:Returns: 0 on success, -1 on error
3980
3981Returns supported MCE capabilities. The u64 mce_cap parameter
3982has the same format as the MSR_IA32_MCG_CAP register. Supported
3983capabilities will have the corresponding bits set.
3984
39854.105 KVM_X86_SETUP_MCE
3986-----------------------
3987
3988:Capability: KVM_CAP_MCE
3989:Architectures: x86
3990:Type: vcpu ioctl
3991:Parameters: u64 mcg_cap (in)
3992:Returns: 0 on success,
3993         -EFAULT if u64 mcg_cap cannot be read,
3994         -EINVAL if the requested number of banks is invalid,
3995         -EINVAL if requested MCE capability is not supported.
3996
3997Initializes MCE support for use. The u64 mcg_cap parameter
3998has the same format as the MSR_IA32_MCG_CAP register and
3999specifies which capabilities should be enabled. The maximum
4000supported number of error-reporting banks can be retrieved when
4001checking for KVM_CAP_MCE. The supported capabilities can be
4002retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4003
40044.106 KVM_X86_SET_MCE
4005---------------------
4006
4007:Capability: KVM_CAP_MCE
4008:Architectures: x86
4009:Type: vcpu ioctl
4010:Parameters: struct kvm_x86_mce (in)
4011:Returns: 0 on success,
4012         -EFAULT if struct kvm_x86_mce cannot be read,
4013         -EINVAL if the bank number is invalid,
4014         -EINVAL if VAL bit is not set in status field.
4015
4016Inject a machine check error (MCE) into the guest. The input
4017parameter is::
4018
4019  struct kvm_x86_mce {
4020	__u64 status;
4021	__u64 addr;
4022	__u64 misc;
4023	__u64 mcg_status;
4024	__u8 bank;
4025	__u8 pad1[7];
4026	__u64 pad2[3];
4027  };
4028
4029If the MCE being reported is an uncorrected error, KVM will
4030inject it as an MCE exception into the guest. If the guest
4031MCG_STATUS register reports that an MCE is in progress, KVM
4032causes an KVM_EXIT_SHUTDOWN vmexit.
4033
4034Otherwise, if the MCE is a corrected error, KVM will just
4035store it in the corresponding bank (provided this bank is
4036not holding a previously reported uncorrected error).
4037
40384.107 KVM_S390_GET_CMMA_BITS
4039----------------------------
4040
4041:Capability: KVM_CAP_S390_CMMA_MIGRATION
4042:Architectures: s390
4043:Type: vm ioctl
4044:Parameters: struct kvm_s390_cmma_log (in, out)
4045:Returns: 0 on success, a negative value on error
4046
4047This ioctl is used to get the values of the CMMA bits on the s390
4048architecture. It is meant to be used in two scenarios:
4049
4050- During live migration to save the CMMA values. Live migration needs
4051  to be enabled via the KVM_REQ_START_MIGRATION VM property.
4052- To non-destructively peek at the CMMA values, with the flag
4053  KVM_S390_CMMA_PEEK set.
4054
4055The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4056values are written to a buffer whose location is indicated via the "values"
4057member in the kvm_s390_cmma_log struct.  The values in the input struct are
4058also updated as needed.
4059
4060Each CMMA value takes up one byte.
4061
4062::
4063
4064  struct kvm_s390_cmma_log {
4065	__u64 start_gfn;
4066	__u32 count;
4067	__u32 flags;
4068	union {
4069		__u64 remaining;
4070		__u64 mask;
4071	};
4072	__u64 values;
4073  };
4074
4075start_gfn is the number of the first guest frame whose CMMA values are
4076to be retrieved,
4077
4078count is the length of the buffer in bytes,
4079
4080values points to the buffer where the result will be written to.
4081
4082If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4083KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4084other ioctls.
4085
4086The result is written in the buffer pointed to by the field values, and
4087the values of the input parameter are updated as follows.
4088
4089Depending on the flags, different actions are performed. The only
4090supported flag so far is KVM_S390_CMMA_PEEK.
4091
4092The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4093start_gfn will indicate the first page frame whose CMMA bits were dirty.
4094It is not necessarily the same as the one passed as input, as clean pages
4095are skipped.
4096
4097count will indicate the number of bytes actually written in the buffer.
4098It can (and very often will) be smaller than the input value, since the
4099buffer is only filled until 16 bytes of clean values are found (which
4100are then not copied in the buffer). Since a CMMA migration block needs
4101the base address and the length, for a total of 16 bytes, we will send
4102back some clean data if there is some dirty data afterwards, as long as
4103the size of the clean data does not exceed the size of the header. This
4104allows to minimize the amount of data to be saved or transferred over
4105the network at the expense of more roundtrips to userspace. The next
4106invocation of the ioctl will skip over all the clean values, saving
4107potentially more than just the 16 bytes we found.
4108
4109If KVM_S390_CMMA_PEEK is set:
4110the existing storage attributes are read even when not in migration
4111mode, and no other action is performed;
4112
4113the output start_gfn will be equal to the input start_gfn,
4114
4115the output count will be equal to the input count, except if the end of
4116memory has been reached.
4117
4118In both cases:
4119the field "remaining" will indicate the total number of dirty CMMA values
4120still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4121not enabled.
4122
4123mask is unused.
4124
4125values points to the userspace buffer where the result will be stored.
4126
4127This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4128complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4129KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4130-EFAULT if the userspace address is invalid or if no page table is
4131present for the addresses (e.g. when using hugepages).
4132
41334.108 KVM_S390_SET_CMMA_BITS
4134----------------------------
4135
4136:Capability: KVM_CAP_S390_CMMA_MIGRATION
4137:Architectures: s390
4138:Type: vm ioctl
4139:Parameters: struct kvm_s390_cmma_log (in)
4140:Returns: 0 on success, a negative value on error
4141
4142This ioctl is used to set the values of the CMMA bits on the s390
4143architecture. It is meant to be used during live migration to restore
4144the CMMA values, but there are no restrictions on its use.
4145The ioctl takes parameters via the kvm_s390_cmma_values struct.
4146Each CMMA value takes up one byte.
4147
4148::
4149
4150  struct kvm_s390_cmma_log {
4151	__u64 start_gfn;
4152	__u32 count;
4153	__u32 flags;
4154	union {
4155		__u64 remaining;
4156		__u64 mask;
4157 	};
4158	__u64 values;
4159  };
4160
4161start_gfn indicates the starting guest frame number,
4162
4163count indicates how many values are to be considered in the buffer,
4164
4165flags is not used and must be 0.
4166
4167mask indicates which PGSTE bits are to be considered.
4168
4169remaining is not used.
4170
4171values points to the buffer in userspace where to store the values.
4172
4173This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4174complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4175the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4176if the flags field was not 0, with -EFAULT if the userspace address is
4177invalid, if invalid pages are written to (e.g. after the end of memory)
4178or if no page table is present for the addresses (e.g. when using
4179hugepages).
4180
41814.109 KVM_PPC_GET_CPU_CHAR
4182--------------------------
4183
4184:Capability: KVM_CAP_PPC_GET_CPU_CHAR
4185:Architectures: powerpc
4186:Type: vm ioctl
4187:Parameters: struct kvm_ppc_cpu_char (out)
4188:Returns: 0 on successful completion,
4189	 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4190
4191This ioctl gives userspace information about certain characteristics
4192of the CPU relating to speculative execution of instructions and
4193possible information leakage resulting from speculative execution (see
4194CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754).  The information is
4195returned in struct kvm_ppc_cpu_char, which looks like this::
4196
4197  struct kvm_ppc_cpu_char {
4198	__u64	character;		/* characteristics of the CPU */
4199	__u64	behaviour;		/* recommended software behaviour */
4200	__u64	character_mask;		/* valid bits in character */
4201	__u64	behaviour_mask;		/* valid bits in behaviour */
4202  };
4203
4204For extensibility, the character_mask and behaviour_mask fields
4205indicate which bits of character and behaviour have been filled in by
4206the kernel.  If the set of defined bits is extended in future then
4207userspace will be able to tell whether it is running on a kernel that
4208knows about the new bits.
4209
4210The character field describes attributes of the CPU which can help
4211with preventing inadvertent information disclosure - specifically,
4212whether there is an instruction to flash-invalidate the L1 data cache
4213(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4214to a mode where entries can only be used by the thread that created
4215them, whether the bcctr[l] instruction prevents speculation, and
4216whether a speculation barrier instruction (ori 31,31,0) is provided.
4217
4218The behaviour field describes actions that software should take to
4219prevent inadvertent information disclosure, and thus describes which
4220vulnerabilities the hardware is subject to; specifically whether the
4221L1 data cache should be flushed when returning to user mode from the
4222kernel, and whether a speculation barrier should be placed between an
4223array bounds check and the array access.
4224
4225These fields use the same bit definitions as the new
4226H_GET_CPU_CHARACTERISTICS hypercall.
4227
42284.110 KVM_MEMORY_ENCRYPT_OP
4229---------------------------
4230
4231:Capability: basic
4232:Architectures: x86
4233:Type: vm
4234:Parameters: an opaque platform specific structure (in/out)
4235:Returns: 0 on success; -1 on error
4236
4237If the platform supports creating encrypted VMs then this ioctl can be used
4238for issuing platform-specific memory encryption commands to manage those
4239encrypted VMs.
4240
4241Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4242(SEV) commands on AMD Processors. The SEV commands are defined in
4243Documentation/virt/kvm/amd-memory-encryption.rst.
4244
42454.111 KVM_MEMORY_ENCRYPT_REG_REGION
4246-----------------------------------
4247
4248:Capability: basic
4249:Architectures: x86
4250:Type: system
4251:Parameters: struct kvm_enc_region (in)
4252:Returns: 0 on success; -1 on error
4253
4254This ioctl can be used to register a guest memory region which may
4255contain encrypted data (e.g. guest RAM, SMRAM etc).
4256
4257It is used in the SEV-enabled guest. When encryption is enabled, a guest
4258memory region may contain encrypted data. The SEV memory encryption
4259engine uses a tweak such that two identical plaintext pages, each at
4260different locations will have differing ciphertexts. So swapping or
4261moving ciphertext of those pages will not result in plaintext being
4262swapped. So relocating (or migrating) physical backing pages for the SEV
4263guest will require some additional steps.
4264
4265Note: The current SEV key management spec does not provide commands to
4266swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4267memory region registered with the ioctl.
4268
42694.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4270-------------------------------------
4271
4272:Capability: basic
4273:Architectures: x86
4274:Type: system
4275:Parameters: struct kvm_enc_region (in)
4276:Returns: 0 on success; -1 on error
4277
4278This ioctl can be used to unregister the guest memory region registered
4279with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4280
42814.113 KVM_HYPERV_EVENTFD
4282------------------------
4283
4284:Capability: KVM_CAP_HYPERV_EVENTFD
4285:Architectures: x86
4286:Type: vm ioctl
4287:Parameters: struct kvm_hyperv_eventfd (in)
4288
4289This ioctl (un)registers an eventfd to receive notifications from the guest on
4290the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4291causing a user exit.  SIGNAL_EVENT hypercall with non-zero event flag number
4292(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4293
4294::
4295
4296  struct kvm_hyperv_eventfd {
4297	__u32 conn_id;
4298	__s32 fd;
4299	__u32 flags;
4300	__u32 padding[3];
4301  };
4302
4303The conn_id field should fit within 24 bits::
4304
4305  #define KVM_HYPERV_CONN_ID_MASK		0x00ffffff
4306
4307The acceptable values for the flags field are::
4308
4309  #define KVM_HYPERV_EVENTFD_DEASSIGN	(1 << 0)
4310
4311:Returns: 0 on success,
4312 	  -EINVAL if conn_id or flags is outside the allowed range,
4313	  -ENOENT on deassign if the conn_id isn't registered,
4314	  -EEXIST on assign if the conn_id is already registered
4315
43164.114 KVM_GET_NESTED_STATE
4317--------------------------
4318
4319:Capability: KVM_CAP_NESTED_STATE
4320:Architectures: x86
4321:Type: vcpu ioctl
4322:Parameters: struct kvm_nested_state (in/out)
4323:Returns: 0 on success, -1 on error
4324
4325Errors:
4326
4327  =====      =============================================================
4328  E2BIG      the total state size exceeds the value of 'size' specified by
4329             the user; the size required will be written into size.
4330  =====      =============================================================
4331
4332::
4333
4334  struct kvm_nested_state {
4335	__u16 flags;
4336	__u16 format;
4337	__u32 size;
4338
4339	union {
4340		struct kvm_vmx_nested_state_hdr vmx;
4341		struct kvm_svm_nested_state_hdr svm;
4342
4343		/* Pad the header to 128 bytes.  */
4344		__u8 pad[120];
4345	} hdr;
4346
4347	union {
4348		struct kvm_vmx_nested_state_data vmx[0];
4349		struct kvm_svm_nested_state_data svm[0];
4350	} data;
4351  };
4352
4353  #define KVM_STATE_NESTED_GUEST_MODE		0x00000001
4354  #define KVM_STATE_NESTED_RUN_PENDING		0x00000002
4355  #define KVM_STATE_NESTED_EVMCS		0x00000004
4356
4357  #define KVM_STATE_NESTED_FORMAT_VMX		0
4358  #define KVM_STATE_NESTED_FORMAT_SVM		1
4359
4360  #define KVM_STATE_NESTED_VMX_VMCS_SIZE	0x1000
4361
4362  #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE	0x00000001
4363  #define KVM_STATE_NESTED_VMX_SMM_VMXON	0x00000002
4364
4365  #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4366
4367  struct kvm_vmx_nested_state_hdr {
4368	__u64 vmxon_pa;
4369	__u64 vmcs12_pa;
4370
4371	struct {
4372		__u16 flags;
4373	} smm;
4374
4375	__u32 flags;
4376	__u64 preemption_timer_deadline;
4377  };
4378
4379  struct kvm_vmx_nested_state_data {
4380	__u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4381	__u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4382  };
4383
4384This ioctl copies the vcpu's nested virtualization state from the kernel to
4385userspace.
4386
4387The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4388to the KVM_CHECK_EXTENSION ioctl().
4389
43904.115 KVM_SET_NESTED_STATE
4391--------------------------
4392
4393:Capability: KVM_CAP_NESTED_STATE
4394:Architectures: x86
4395:Type: vcpu ioctl
4396:Parameters: struct kvm_nested_state (in)
4397:Returns: 0 on success, -1 on error
4398
4399This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4400For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4401
44024.116 KVM_(UN)REGISTER_COALESCED_MMIO
4403-------------------------------------
4404
4405:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4406	     KVM_CAP_COALESCED_PIO (for coalesced pio)
4407:Architectures: all
4408:Type: vm ioctl
4409:Parameters: struct kvm_coalesced_mmio_zone
4410:Returns: 0 on success, < 0 on error
4411
4412Coalesced I/O is a performance optimization that defers hardware
4413register write emulation so that userspace exits are avoided.  It is
4414typically used to reduce the overhead of emulating frequently accessed
4415hardware registers.
4416
4417When a hardware register is configured for coalesced I/O, write accesses
4418do not exit to userspace and their value is recorded in a ring buffer
4419that is shared between kernel and userspace.
4420
4421Coalesced I/O is used if one or more write accesses to a hardware
4422register can be deferred until a read or a write to another hardware
4423register on the same device.  This last access will cause a vmexit and
4424userspace will process accesses from the ring buffer before emulating
4425it. That will avoid exiting to userspace on repeated writes.
4426
4427Coalesced pio is based on coalesced mmio. There is little difference
4428between coalesced mmio and pio except that coalesced pio records accesses
4429to I/O ports.
4430
44314.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4432------------------------------------
4433
4434:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4435:Architectures: x86, arm, arm64, mips
4436:Type: vm ioctl
4437:Parameters: struct kvm_clear_dirty_log (in)
4438:Returns: 0 on success, -1 on error
4439
4440::
4441
4442  /* for KVM_CLEAR_DIRTY_LOG */
4443  struct kvm_clear_dirty_log {
4444	__u32 slot;
4445	__u32 num_pages;
4446	__u64 first_page;
4447	union {
4448		void __user *dirty_bitmap; /* one bit per page */
4449		__u64 padding;
4450	};
4451  };
4452
4453The ioctl clears the dirty status of pages in a memory slot, according to
4454the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4455field.  Bit 0 of the bitmap corresponds to page "first_page" in the
4456memory slot, and num_pages is the size in bits of the input bitmap.
4457first_page must be a multiple of 64; num_pages must also be a multiple of
445864 unless first_page + num_pages is the size of the memory slot.  For each
4459bit that is set in the input bitmap, the corresponding page is marked "clean"
4460in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4461(for example via write-protection, or by clearing the dirty bit in
4462a page table entry).
4463
4464If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4465the address space for which you want to clear the dirty status.  See
4466KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4467
4468This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4469is enabled; for more information, see the description of the capability.
4470However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4471that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4472
44734.118 KVM_GET_SUPPORTED_HV_CPUID
4474--------------------------------
4475
4476:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4477:Architectures: x86
4478:Type: system ioctl, vcpu ioctl
4479:Parameters: struct kvm_cpuid2 (in/out)
4480:Returns: 0 on success, -1 on error
4481
4482::
4483
4484  struct kvm_cpuid2 {
4485	__u32 nent;
4486	__u32 padding;
4487	struct kvm_cpuid_entry2 entries[0];
4488  };
4489
4490  struct kvm_cpuid_entry2 {
4491	__u32 function;
4492	__u32 index;
4493	__u32 flags;
4494	__u32 eax;
4495	__u32 ebx;
4496	__u32 ecx;
4497	__u32 edx;
4498	__u32 padding[3];
4499  };
4500
4501This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4502KVM.  Userspace can use the information returned by this ioctl to construct
4503cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4504Windows or Hyper-V guests).
4505
4506CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4507Functional Specification (TLFS). These leaves can't be obtained with
4508KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4509leaves (0x40000000, 0x40000001).
4510
4511Currently, the following list of CPUID leaves are returned:
4512 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4513 - HYPERV_CPUID_INTERFACE
4514 - HYPERV_CPUID_VERSION
4515 - HYPERV_CPUID_FEATURES
4516 - HYPERV_CPUID_ENLIGHTMENT_INFO
4517 - HYPERV_CPUID_IMPLEMENT_LIMITS
4518 - HYPERV_CPUID_NESTED_FEATURES
4519 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4520 - HYPERV_CPUID_SYNDBG_INTERFACE
4521 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4522
4523Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4524with the 'nent' field indicating the number of entries in the variable-size
4525array 'entries'.  If the number of entries is too low to describe all Hyper-V
4526feature leaves, an error (E2BIG) is returned. If the number is more or equal
4527to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4528number of valid entries in the 'entries' array, which is then filled.
4529
4530'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4531userspace should not expect to get any particular value there.
4532
4533Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4534system ioctl which exposes all supported feature bits unconditionally, vcpu
4535version has the following quirks:
4536- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4537  feature bit are only exposed when Enlightened VMCS was previously enabled
4538  on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4539- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4540  (presumes KVM_CREATE_IRQCHIP has already been called).
4541
45424.119 KVM_ARM_VCPU_FINALIZE
4543---------------------------
4544
4545:Architectures: arm, arm64
4546:Type: vcpu ioctl
4547:Parameters: int feature (in)
4548:Returns: 0 on success, -1 on error
4549
4550Errors:
4551
4552  ======     ==============================================================
4553  EPERM      feature not enabled, needs configuration, or already finalized
4554  EINVAL     feature unknown or not present
4555  ======     ==============================================================
4556
4557Recognised values for feature:
4558
4559  =====      ===========================================
4560  arm64      KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4561  =====      ===========================================
4562
4563Finalizes the configuration of the specified vcpu feature.
4564
4565The vcpu must already have been initialised, enabling the affected feature, by
4566means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4567features[].
4568
4569For affected vcpu features, this is a mandatory step that must be performed
4570before the vcpu is fully usable.
4571
4572Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4573configured by use of ioctls such as KVM_SET_ONE_REG.  The exact configuration
4574that should be performaned and how to do it are feature-dependent.
4575
4576Other calls that depend on a particular feature being finalized, such as
4577KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4578-EPERM unless the feature has already been finalized by means of a
4579KVM_ARM_VCPU_FINALIZE call.
4580
4581See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4582using this ioctl.
4583
45844.120 KVM_SET_PMU_EVENT_FILTER
4585------------------------------
4586
4587:Capability: KVM_CAP_PMU_EVENT_FILTER
4588:Architectures: x86
4589:Type: vm ioctl
4590:Parameters: struct kvm_pmu_event_filter (in)
4591:Returns: 0 on success, -1 on error
4592
4593::
4594
4595  struct kvm_pmu_event_filter {
4596	__u32 action;
4597	__u32 nevents;
4598	__u32 fixed_counter_bitmap;
4599	__u32 flags;
4600	__u32 pad[4];
4601	__u64 events[0];
4602  };
4603
4604This ioctl restricts the set of PMU events that the guest can program.
4605The argument holds a list of events which will be allowed or denied.
4606The eventsel+umask of each event the guest attempts to program is compared
4607against the events field to determine whether the guest should have access.
4608The events field only controls general purpose counters; fixed purpose
4609counters are controlled by the fixed_counter_bitmap.
4610
4611No flags are defined yet, the field must be zero.
4612
4613Valid values for 'action'::
4614
4615  #define KVM_PMU_EVENT_ALLOW 0
4616  #define KVM_PMU_EVENT_DENY 1
4617
46184.121 KVM_PPC_SVM_OFF
4619---------------------
4620
4621:Capability: basic
4622:Architectures: powerpc
4623:Type: vm ioctl
4624:Parameters: none
4625:Returns: 0 on successful completion,
4626
4627Errors:
4628
4629  ======     ================================================================
4630  EINVAL     if ultravisor failed to terminate the secure guest
4631  ENOMEM     if hypervisor failed to allocate new radix page tables for guest
4632  ======     ================================================================
4633
4634This ioctl is used to turn off the secure mode of the guest or transition
4635the guest from secure mode to normal mode. This is invoked when the guest
4636is reset. This has no effect if called for a normal guest.
4637
4638This ioctl issues an ultravisor call to terminate the secure guest,
4639unpins the VPA pages and releases all the device pages that are used to
4640track the secure pages by hypervisor.
4641
46424.122 KVM_S390_NORMAL_RESET
4643---------------------------
4644
4645:Capability: KVM_CAP_S390_VCPU_RESETS
4646:Architectures: s390
4647:Type: vcpu ioctl
4648:Parameters: none
4649:Returns: 0
4650
4651This ioctl resets VCPU registers and control structures according to
4652the cpu reset definition in the POP (Principles Of Operation).
4653
46544.123 KVM_S390_INITIAL_RESET
4655----------------------------
4656
4657:Capability: none
4658:Architectures: s390
4659:Type: vcpu ioctl
4660:Parameters: none
4661:Returns: 0
4662
4663This ioctl resets VCPU registers and control structures according to
4664the initial cpu reset definition in the POP. However, the cpu is not
4665put into ESA mode. This reset is a superset of the normal reset.
4666
46674.124 KVM_S390_CLEAR_RESET
4668--------------------------
4669
4670:Capability: KVM_CAP_S390_VCPU_RESETS
4671:Architectures: s390
4672:Type: vcpu ioctl
4673:Parameters: none
4674:Returns: 0
4675
4676This ioctl resets VCPU registers and control structures according to
4677the clear cpu reset definition in the POP. However, the cpu is not put
4678into ESA mode. This reset is a superset of the initial reset.
4679
4680
46814.125 KVM_S390_PV_COMMAND
4682-------------------------
4683
4684:Capability: KVM_CAP_S390_PROTECTED
4685:Architectures: s390
4686:Type: vm ioctl
4687:Parameters: struct kvm_pv_cmd
4688:Returns: 0 on success, < 0 on error
4689
4690::
4691
4692  struct kvm_pv_cmd {
4693	__u32 cmd;	/* Command to be executed */
4694	__u16 rc;	/* Ultravisor return code */
4695	__u16 rrc;	/* Ultravisor return reason code */
4696	__u64 data;	/* Data or address */
4697	__u32 flags;    /* flags for future extensions. Must be 0 for now */
4698	__u32 reserved[3];
4699  };
4700
4701cmd values:
4702
4703KVM_PV_ENABLE
4704  Allocate memory and register the VM with the Ultravisor, thereby
4705  donating memory to the Ultravisor that will become inaccessible to
4706  KVM. All existing CPUs are converted to protected ones. After this
4707  command has succeeded, any CPU added via hotplug will become
4708  protected during its creation as well.
4709
4710  Errors:
4711
4712  =====      =============================
4713  EINTR      an unmasked signal is pending
4714  =====      =============================
4715
4716KVM_PV_DISABLE
4717
4718  Deregister the VM from the Ultravisor and reclaim the memory that
4719  had been donated to the Ultravisor, making it usable by the kernel
4720  again.  All registered VCPUs are converted back to non-protected
4721  ones.
4722
4723KVM_PV_VM_SET_SEC_PARMS
4724  Pass the image header from VM memory to the Ultravisor in
4725  preparation of image unpacking and verification.
4726
4727KVM_PV_VM_UNPACK
4728  Unpack (protect and decrypt) a page of the encrypted boot image.
4729
4730KVM_PV_VM_VERIFY
4731  Verify the integrity of the unpacked image. Only if this succeeds,
4732  KVM is allowed to start protected VCPUs.
4733
47344.126 KVM_X86_SET_MSR_FILTER
4735----------------------------
4736
4737:Capability: KVM_X86_SET_MSR_FILTER
4738:Architectures: x86
4739:Type: vm ioctl
4740:Parameters: struct kvm_msr_filter
4741:Returns: 0 on success, < 0 on error
4742
4743::
4744
4745  struct kvm_msr_filter_range {
4746  #define KVM_MSR_FILTER_READ  (1 << 0)
4747  #define KVM_MSR_FILTER_WRITE (1 << 1)
4748	__u32 flags;
4749	__u32 nmsrs; /* number of msrs in bitmap */
4750	__u32 base;  /* MSR index the bitmap starts at */
4751	__u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4752  };
4753
4754  #define KVM_MSR_FILTER_MAX_RANGES 16
4755  struct kvm_msr_filter {
4756  #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4757  #define KVM_MSR_FILTER_DEFAULT_DENY  (1 << 0)
4758	__u32 flags;
4759	struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4760  };
4761
4762flags values for ``struct kvm_msr_filter_range``:
4763
4764``KVM_MSR_FILTER_READ``
4765
4766  Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4767  indicates that a read should immediately fail, while a 1 indicates that
4768  a read for a particular MSR should be handled regardless of the default
4769  filter action.
4770
4771``KVM_MSR_FILTER_WRITE``
4772
4773  Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4774  indicates that a write should immediately fail, while a 1 indicates that
4775  a write for a particular MSR should be handled regardless of the default
4776  filter action.
4777
4778``KVM_MSR_FILTER_READ | KVM_MSR_FILTER_WRITE``
4779
4780  Filter both read and write accesses to MSRs using the given bitmap. A 0
4781  in the bitmap indicates that both reads and writes should immediately fail,
4782  while a 1 indicates that reads and writes for a particular MSR are not
4783  filtered by this range.
4784
4785flags values for ``struct kvm_msr_filter``:
4786
4787``KVM_MSR_FILTER_DEFAULT_ALLOW``
4788
4789  If no filter range matches an MSR index that is getting accessed, KVM will
4790  fall back to allowing access to the MSR.
4791
4792``KVM_MSR_FILTER_DEFAULT_DENY``
4793
4794  If no filter range matches an MSR index that is getting accessed, KVM will
4795  fall back to rejecting access to the MSR. In this mode, all MSRs that should
4796  be processed by KVM need to explicitly be marked as allowed in the bitmaps.
4797
4798This ioctl allows user space to define up to 16 bitmaps of MSR ranges to
4799specify whether a certain MSR access should be explicitly filtered for or not.
4800
4801If this ioctl has never been invoked, MSR accesses are not guarded and the
4802default KVM in-kernel emulation behavior is fully preserved.
4803
4804Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4805filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4806an error.
4807
4808As soon as the filtering is in place, every MSR access is processed through
4809the filtering except for accesses to the x2APIC MSRs (from 0x800 to 0x8ff);
4810x2APIC MSRs are always allowed, independent of the ``default_allow`` setting,
4811and their behavior depends on the ``X2APIC_ENABLE`` bit of the APIC base
4812register.
4813
4814If a bit is within one of the defined ranges, read and write accesses are
4815guarded by the bitmap's value for the MSR index if the kind of access
4816is included in the ``struct kvm_msr_filter_range`` flags.  If no range
4817cover this particular access, the behavior is determined by the flags
4818field in the kvm_msr_filter struct: ``KVM_MSR_FILTER_DEFAULT_ALLOW``
4819and ``KVM_MSR_FILTER_DEFAULT_DENY``.
4820
4821Each bitmap range specifies a range of MSRs to potentially allow access on.
4822The range goes from MSR index [base .. base+nmsrs]. The flags field
4823indicates whether reads, writes or both reads and writes are filtered
4824by setting a 1 bit in the bitmap for the corresponding MSR index.
4825
4826If an MSR access is not permitted through the filtering, it generates a
4827#GP inside the guest. When combined with KVM_CAP_X86_USER_SPACE_MSR, that
4828allows user space to deflect and potentially handle various MSR accesses
4829into user space.
4830
4831If a vCPU is in running state while this ioctl is invoked, the vCPU may
4832experience inconsistent filtering behavior on MSR accesses.
4833
4834
48355. The kvm_run structure
4836========================
4837
4838Application code obtains a pointer to the kvm_run structure by
4839mmap()ing a vcpu fd.  From that point, application code can control
4840execution by changing fields in kvm_run prior to calling the KVM_RUN
4841ioctl, and obtain information about the reason KVM_RUN returned by
4842looking up structure members.
4843
4844::
4845
4846  struct kvm_run {
4847	/* in */
4848	__u8 request_interrupt_window;
4849
4850Request that KVM_RUN return when it becomes possible to inject external
4851interrupts into the guest.  Useful in conjunction with KVM_INTERRUPT.
4852
4853::
4854
4855	__u8 immediate_exit;
4856
4857This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
4858exits immediately, returning -EINTR.  In the common scenario where a
4859signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
4860to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
4861Rather than blocking the signal outside KVM_RUN, userspace can set up
4862a signal handler that sets run->immediate_exit to a non-zero value.
4863
4864This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
4865
4866::
4867
4868	__u8 padding1[6];
4869
4870	/* out */
4871	__u32 exit_reason;
4872
4873When KVM_RUN has returned successfully (return value 0), this informs
4874application code why KVM_RUN has returned.  Allowable values for this
4875field are detailed below.
4876
4877::
4878
4879	__u8 ready_for_interrupt_injection;
4880
4881If request_interrupt_window has been specified, this field indicates
4882an interrupt can be injected now with KVM_INTERRUPT.
4883
4884::
4885
4886	__u8 if_flag;
4887
4888The value of the current interrupt flag.  Only valid if in-kernel
4889local APIC is not used.
4890
4891::
4892
4893	__u16 flags;
4894
4895More architecture-specific flags detailing state of the VCPU that may
4896affect the device's behavior.  The only currently defined flag is
4897KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
4898VCPU is in system management mode.
4899
4900::
4901
4902	/* in (pre_kvm_run), out (post_kvm_run) */
4903	__u64 cr8;
4904
4905The value of the cr8 register.  Only valid if in-kernel local APIC is
4906not used.  Both input and output.
4907
4908::
4909
4910	__u64 apic_base;
4911
4912The value of the APIC BASE msr.  Only valid if in-kernel local
4913APIC is not used.  Both input and output.
4914
4915::
4916
4917	union {
4918		/* KVM_EXIT_UNKNOWN */
4919		struct {
4920			__u64 hardware_exit_reason;
4921		} hw;
4922
4923If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
4924reasons.  Further architecture-specific information is available in
4925hardware_exit_reason.
4926
4927::
4928
4929		/* KVM_EXIT_FAIL_ENTRY */
4930		struct {
4931			__u64 hardware_entry_failure_reason;
4932			__u32 cpu; /* if KVM_LAST_CPU */
4933		} fail_entry;
4934
4935If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
4936to unknown reasons.  Further architecture-specific information is
4937available in hardware_entry_failure_reason.
4938
4939::
4940
4941		/* KVM_EXIT_EXCEPTION */
4942		struct {
4943			__u32 exception;
4944			__u32 error_code;
4945		} ex;
4946
4947Unused.
4948
4949::
4950
4951		/* KVM_EXIT_IO */
4952		struct {
4953  #define KVM_EXIT_IO_IN  0
4954  #define KVM_EXIT_IO_OUT 1
4955			__u8 direction;
4956			__u8 size; /* bytes */
4957			__u16 port;
4958			__u32 count;
4959			__u64 data_offset; /* relative to kvm_run start */
4960		} io;
4961
4962If exit_reason is KVM_EXIT_IO, then the vcpu has
4963executed a port I/O instruction which could not be satisfied by kvm.
4964data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
4965where kvm expects application code to place the data for the next
4966KVM_RUN invocation (KVM_EXIT_IO_IN).  Data format is a packed array.
4967
4968::
4969
4970		/* KVM_EXIT_DEBUG */
4971		struct {
4972			struct kvm_debug_exit_arch arch;
4973		} debug;
4974
4975If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
4976for which architecture specific information is returned.
4977
4978::
4979
4980		/* KVM_EXIT_MMIO */
4981		struct {
4982			__u64 phys_addr;
4983			__u8  data[8];
4984			__u32 len;
4985			__u8  is_write;
4986		} mmio;
4987
4988If exit_reason is KVM_EXIT_MMIO, then the vcpu has
4989executed a memory-mapped I/O instruction which could not be satisfied
4990by kvm.  The 'data' member contains the written data if 'is_write' is
4991true, and should be filled by application code otherwise.
4992
4993The 'data' member contains, in its first 'len' bytes, the value as it would
4994appear if the VCPU performed a load or store of the appropriate width directly
4995to the byte array.
4996
4997.. note::
4998
4999      For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR,
5000      KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
5001      operations are complete (and guest state is consistent) only after userspace
5002      has re-entered the kernel with KVM_RUN.  The kernel side will first finish
5003      incomplete operations and then check for pending signals.  Userspace
5004      can re-enter the guest with an unmasked signal pending to complete
5005      pending operations.
5006
5007::
5008
5009		/* KVM_EXIT_HYPERCALL */
5010		struct {
5011			__u64 nr;
5012			__u64 args[6];
5013			__u64 ret;
5014			__u32 longmode;
5015			__u32 pad;
5016		} hypercall;
5017
5018Unused.  This was once used for 'hypercall to userspace'.  To implement
5019such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
5020
5021.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
5022
5023::
5024
5025		/* KVM_EXIT_TPR_ACCESS */
5026		struct {
5027			__u64 rip;
5028			__u32 is_write;
5029			__u32 pad;
5030		} tpr_access;
5031
5032To be documented (KVM_TPR_ACCESS_REPORTING).
5033
5034::
5035
5036		/* KVM_EXIT_S390_SIEIC */
5037		struct {
5038			__u8 icptcode;
5039			__u64 mask; /* psw upper half */
5040			__u64 addr; /* psw lower half */
5041			__u16 ipa;
5042			__u32 ipb;
5043		} s390_sieic;
5044
5045s390 specific.
5046
5047::
5048
5049		/* KVM_EXIT_S390_RESET */
5050  #define KVM_S390_RESET_POR       1
5051  #define KVM_S390_RESET_CLEAR     2
5052  #define KVM_S390_RESET_SUBSYSTEM 4
5053  #define KVM_S390_RESET_CPU_INIT  8
5054  #define KVM_S390_RESET_IPL       16
5055		__u64 s390_reset_flags;
5056
5057s390 specific.
5058
5059::
5060
5061		/* KVM_EXIT_S390_UCONTROL */
5062		struct {
5063			__u64 trans_exc_code;
5064			__u32 pgm_code;
5065		} s390_ucontrol;
5066
5067s390 specific. A page fault has occurred for a user controlled virtual
5068machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
5069resolved by the kernel.
5070The program code and the translation exception code that were placed
5071in the cpu's lowcore are presented here as defined by the z Architecture
5072Principles of Operation Book in the Chapter for Dynamic Address Translation
5073(DAT)
5074
5075::
5076
5077		/* KVM_EXIT_DCR */
5078		struct {
5079			__u32 dcrn;
5080			__u32 data;
5081			__u8  is_write;
5082		} dcr;
5083
5084Deprecated - was used for 440 KVM.
5085
5086::
5087
5088		/* KVM_EXIT_OSI */
5089		struct {
5090			__u64 gprs[32];
5091		} osi;
5092
5093MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
5094hypercalls and exit with this exit struct that contains all the guest gprs.
5095
5096If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
5097Userspace can now handle the hypercall and when it's done modify the gprs as
5098necessary. Upon guest entry all guest GPRs will then be replaced by the values
5099in this struct.
5100
5101::
5102
5103		/* KVM_EXIT_PAPR_HCALL */
5104		struct {
5105			__u64 nr;
5106			__u64 ret;
5107			__u64 args[9];
5108		} papr_hcall;
5109
5110This is used on 64-bit PowerPC when emulating a pSeries partition,
5111e.g. with the 'pseries' machine type in qemu.  It occurs when the
5112guest does a hypercall using the 'sc 1' instruction.  The 'nr' field
5113contains the hypercall number (from the guest R3), and 'args' contains
5114the arguments (from the guest R4 - R12).  Userspace should put the
5115return code in 'ret' and any extra returned values in args[].
5116The possible hypercalls are defined in the Power Architecture Platform
5117Requirements (PAPR) document available from www.power.org (free
5118developer registration required to access it).
5119
5120::
5121
5122		/* KVM_EXIT_S390_TSCH */
5123		struct {
5124			__u16 subchannel_id;
5125			__u16 subchannel_nr;
5126			__u32 io_int_parm;
5127			__u32 io_int_word;
5128			__u32 ipb;
5129			__u8 dequeued;
5130		} s390_tsch;
5131
5132s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
5133and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
5134interrupt for the target subchannel has been dequeued and subchannel_id,
5135subchannel_nr, io_int_parm and io_int_word contain the parameters for that
5136interrupt. ipb is needed for instruction parameter decoding.
5137
5138::
5139
5140		/* KVM_EXIT_EPR */
5141		struct {
5142			__u32 epr;
5143		} epr;
5144
5145On FSL BookE PowerPC chips, the interrupt controller has a fast patch
5146interrupt acknowledge path to the core. When the core successfully
5147delivers an interrupt, it automatically populates the EPR register with
5148the interrupt vector number and acknowledges the interrupt inside
5149the interrupt controller.
5150
5151In case the interrupt controller lives in user space, we need to do
5152the interrupt acknowledge cycle through it to fetch the next to be
5153delivered interrupt vector using this exit.
5154
5155It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
5156external interrupt has just been delivered into the guest. User space
5157should put the acknowledged interrupt vector into the 'epr' field.
5158
5159::
5160
5161		/* KVM_EXIT_SYSTEM_EVENT */
5162		struct {
5163  #define KVM_SYSTEM_EVENT_SHUTDOWN       1
5164  #define KVM_SYSTEM_EVENT_RESET          2
5165  #define KVM_SYSTEM_EVENT_CRASH          3
5166			__u32 type;
5167			__u64 flags;
5168		} system_event;
5169
5170If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
5171a system-level event using some architecture specific mechanism (hypercall
5172or some special instruction). In case of ARM/ARM64, this is triggered using
5173HVC instruction based PSCI call from the vcpu. The 'type' field describes
5174the system-level event type. The 'flags' field describes architecture
5175specific flags for the system-level event.
5176
5177Valid values for 'type' are:
5178
5179 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
5180   VM. Userspace is not obliged to honour this, and if it does honour
5181   this does not need to destroy the VM synchronously (ie it may call
5182   KVM_RUN again before shutdown finally occurs).
5183 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
5184   As with SHUTDOWN, userspace can choose to ignore the request, or
5185   to schedule the reset to occur in the future and may call KVM_RUN again.
5186 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
5187   has requested a crash condition maintenance. Userspace can choose
5188   to ignore the request, or to gather VM memory core dump and/or
5189   reset/shutdown of the VM.
5190
5191::
5192
5193		/* KVM_EXIT_IOAPIC_EOI */
5194		struct {
5195			__u8 vector;
5196		} eoi;
5197
5198Indicates that the VCPU's in-kernel local APIC received an EOI for a
5199level-triggered IOAPIC interrupt.  This exit only triggers when the
5200IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
5201the userspace IOAPIC should process the EOI and retrigger the interrupt if
5202it is still asserted.  Vector is the LAPIC interrupt vector for which the
5203EOI was received.
5204
5205::
5206
5207		struct kvm_hyperv_exit {
5208  #define KVM_EXIT_HYPERV_SYNIC          1
5209  #define KVM_EXIT_HYPERV_HCALL          2
5210  #define KVM_EXIT_HYPERV_SYNDBG         3
5211			__u32 type;
5212			__u32 pad1;
5213			union {
5214				struct {
5215					__u32 msr;
5216					__u32 pad2;
5217					__u64 control;
5218					__u64 evt_page;
5219					__u64 msg_page;
5220				} synic;
5221				struct {
5222					__u64 input;
5223					__u64 result;
5224					__u64 params[2];
5225				} hcall;
5226				struct {
5227					__u32 msr;
5228					__u32 pad2;
5229					__u64 control;
5230					__u64 status;
5231					__u64 send_page;
5232					__u64 recv_page;
5233					__u64 pending_page;
5234				} syndbg;
5235			} u;
5236		};
5237		/* KVM_EXIT_HYPERV */
5238                struct kvm_hyperv_exit hyperv;
5239
5240Indicates that the VCPU exits into userspace to process some tasks
5241related to Hyper-V emulation.
5242
5243Valid values for 'type' are:
5244
5245	- KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
5246
5247Hyper-V SynIC state change. Notification is used to remap SynIC
5248event/message pages and to enable/disable SynIC messages/events processing
5249in userspace.
5250
5251	- KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
5252
5253Hyper-V Synthetic debugger state change. Notification is used to either update
5254the pending_page location or to send a control command (send the buffer located
5255in send_page or recv a buffer to recv_page).
5256
5257::
5258
5259		/* KVM_EXIT_ARM_NISV */
5260		struct {
5261			__u64 esr_iss;
5262			__u64 fault_ipa;
5263		} arm_nisv;
5264
5265Used on arm and arm64 systems. If a guest accesses memory not in a memslot,
5266KVM will typically return to userspace and ask it to do MMIO emulation on its
5267behalf. However, for certain classes of instructions, no instruction decode
5268(direction, length of memory access) is provided, and fetching and decoding
5269the instruction from the VM is overly complicated to live in the kernel.
5270
5271Historically, when this situation occurred, KVM would print a warning and kill
5272the VM. KVM assumed that if the guest accessed non-memslot memory, it was
5273trying to do I/O, which just couldn't be emulated, and the warning message was
5274phrased accordingly. However, what happened more often was that a guest bug
5275caused access outside the guest memory areas which should lead to a more
5276meaningful warning message and an external abort in the guest, if the access
5277did not fall within an I/O window.
5278
5279Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
5280this capability at VM creation. Once this is done, these types of errors will
5281instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
5282the HSR (arm) and ESR_EL2 (arm64) in the esr_iss field, and the faulting IPA
5283in the fault_ipa field. Userspace can either fix up the access if it's
5284actually an I/O access by decoding the instruction from guest memory (if it's
5285very brave) and continue executing the guest, or it can decide to suspend,
5286dump, or restart the guest.
5287
5288Note that KVM does not skip the faulting instruction as it does for
5289KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
5290if it decides to decode and emulate the instruction.
5291
5292::
5293
5294		/* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
5295		struct {
5296			__u8 error; /* user -> kernel */
5297			__u8 pad[7];
5298			__u32 reason; /* kernel -> user */
5299			__u32 index; /* kernel -> user */
5300			__u64 data; /* kernel <-> user */
5301		} msr;
5302
5303Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
5304enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
5305will instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
5306exit for writes.
5307
5308The "reason" field specifies why the MSR trap occurred. User space will only
5309receive MSR exit traps when a particular reason was requested during through
5310ENABLE_CAP. Currently valid exit reasons are:
5311
5312	KVM_MSR_EXIT_REASON_UNKNOWN - access to MSR that is unknown to KVM
5313	KVM_MSR_EXIT_REASON_INVAL - access to invalid MSRs or reserved bits
5314	KVM_MSR_EXIT_REASON_FILTER - access blocked by KVM_X86_SET_MSR_FILTER
5315
5316For KVM_EXIT_X86_RDMSR, the "index" field tells user space which MSR the guest
5317wants to read. To respond to this request with a successful read, user space
5318writes the respective data into the "data" field and must continue guest
5319execution to ensure the read data is transferred into guest register state.
5320
5321If the RDMSR request was unsuccessful, user space indicates that with a "1" in
5322the "error" field. This will inject a #GP into the guest when the VCPU is
5323executed again.
5324
5325For KVM_EXIT_X86_WRMSR, the "index" field tells user space which MSR the guest
5326wants to write. Once finished processing the event, user space must continue
5327vCPU execution. If the MSR write was unsuccessful, user space also sets the
5328"error" field to "1".
5329
5330::
5331
5332		/* Fix the size of the union. */
5333		char padding[256];
5334	};
5335
5336	/*
5337	 * shared registers between kvm and userspace.
5338	 * kvm_valid_regs specifies the register classes set by the host
5339	 * kvm_dirty_regs specified the register classes dirtied by userspace
5340	 * struct kvm_sync_regs is architecture specific, as well as the
5341	 * bits for kvm_valid_regs and kvm_dirty_regs
5342	 */
5343	__u64 kvm_valid_regs;
5344	__u64 kvm_dirty_regs;
5345	union {
5346		struct kvm_sync_regs regs;
5347		char padding[SYNC_REGS_SIZE_BYTES];
5348	} s;
5349
5350If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
5351certain guest registers without having to call SET/GET_*REGS. Thus we can
5352avoid some system call overhead if userspace has to handle the exit.
5353Userspace can query the validity of the structure by checking
5354kvm_valid_regs for specific bits. These bits are architecture specific
5355and usually define the validity of a groups of registers. (e.g. one bit
5356for general purpose registers)
5357
5358Please note that the kernel is allowed to use the kvm_run structure as the
5359primary storage for certain register types. Therefore, the kernel may use the
5360values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
5361
5362::
5363
5364  };
5365
5366
5367
53686. Capabilities that can be enabled on vCPUs
5369============================================
5370
5371There are certain capabilities that change the behavior of the virtual CPU or
5372the virtual machine when enabled. To enable them, please see section 4.37.
5373Below you can find a list of capabilities and what their effect on the vCPU or
5374the virtual machine is when enabling them.
5375
5376The following information is provided along with the description:
5377
5378  Architectures:
5379      which instruction set architectures provide this ioctl.
5380      x86 includes both i386 and x86_64.
5381
5382  Target:
5383      whether this is a per-vcpu or per-vm capability.
5384
5385  Parameters:
5386      what parameters are accepted by the capability.
5387
5388  Returns:
5389      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
5390      are not detailed, but errors with specific meanings are.
5391
5392
53936.1 KVM_CAP_PPC_OSI
5394-------------------
5395
5396:Architectures: ppc
5397:Target: vcpu
5398:Parameters: none
5399:Returns: 0 on success; -1 on error
5400
5401This capability enables interception of OSI hypercalls that otherwise would
5402be treated as normal system calls to be injected into the guest. OSI hypercalls
5403were invented by Mac-on-Linux to have a standardized communication mechanism
5404between the guest and the host.
5405
5406When this capability is enabled, KVM_EXIT_OSI can occur.
5407
5408
54096.2 KVM_CAP_PPC_PAPR
5410--------------------
5411
5412:Architectures: ppc
5413:Target: vcpu
5414:Parameters: none
5415:Returns: 0 on success; -1 on error
5416
5417This capability enables interception of PAPR hypercalls. PAPR hypercalls are
5418done using the hypercall instruction "sc 1".
5419
5420It also sets the guest privilege level to "supervisor" mode. Usually the guest
5421runs in "hypervisor" privilege mode with a few missing features.
5422
5423In addition to the above, it changes the semantics of SDR1. In this mode, the
5424HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
5425HTAB invisible to the guest.
5426
5427When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
5428
5429
54306.3 KVM_CAP_SW_TLB
5431------------------
5432
5433:Architectures: ppc
5434:Target: vcpu
5435:Parameters: args[0] is the address of a struct kvm_config_tlb
5436:Returns: 0 on success; -1 on error
5437
5438::
5439
5440  struct kvm_config_tlb {
5441	__u64 params;
5442	__u64 array;
5443	__u32 mmu_type;
5444	__u32 array_len;
5445  };
5446
5447Configures the virtual CPU's TLB array, establishing a shared memory area
5448between userspace and KVM.  The "params" and "array" fields are userspace
5449addresses of mmu-type-specific data structures.  The "array_len" field is an
5450safety mechanism, and should be set to the size in bytes of the memory that
5451userspace has reserved for the array.  It must be at least the size dictated
5452by "mmu_type" and "params".
5453
5454While KVM_RUN is active, the shared region is under control of KVM.  Its
5455contents are undefined, and any modification by userspace results in
5456boundedly undefined behavior.
5457
5458On return from KVM_RUN, the shared region will reflect the current state of
5459the guest's TLB.  If userspace makes any changes, it must call KVM_DIRTY_TLB
5460to tell KVM which entries have been changed, prior to calling KVM_RUN again
5461on this vcpu.
5462
5463For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
5464
5465 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
5466 - The "array" field points to an array of type "struct
5467   kvm_book3e_206_tlb_entry".
5468 - The array consists of all entries in the first TLB, followed by all
5469   entries in the second TLB.
5470 - Within a TLB, entries are ordered first by increasing set number.  Within a
5471   set, entries are ordered by way (increasing ESEL).
5472 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
5473   where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
5474 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
5475   hardware ignores this value for TLB0.
5476
54776.4 KVM_CAP_S390_CSS_SUPPORT
5478----------------------------
5479
5480:Architectures: s390
5481:Target: vcpu
5482:Parameters: none
5483:Returns: 0 on success; -1 on error
5484
5485This capability enables support for handling of channel I/O instructions.
5486
5487TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
5488handled in-kernel, while the other I/O instructions are passed to userspace.
5489
5490When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
5491SUBCHANNEL intercepts.
5492
5493Note that even though this capability is enabled per-vcpu, the complete
5494virtual machine is affected.
5495
54966.5 KVM_CAP_PPC_EPR
5497-------------------
5498
5499:Architectures: ppc
5500:Target: vcpu
5501:Parameters: args[0] defines whether the proxy facility is active
5502:Returns: 0 on success; -1 on error
5503
5504This capability enables or disables the delivery of interrupts through the
5505external proxy facility.
5506
5507When enabled (args[0] != 0), every time the guest gets an external interrupt
5508delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
5509to receive the topmost interrupt vector.
5510
5511When disabled (args[0] == 0), behavior is as if this facility is unsupported.
5512
5513When this capability is enabled, KVM_EXIT_EPR can occur.
5514
55156.6 KVM_CAP_IRQ_MPIC
5516--------------------
5517
5518:Architectures: ppc
5519:Parameters: args[0] is the MPIC device fd;
5520             args[1] is the MPIC CPU number for this vcpu
5521
5522This capability connects the vcpu to an in-kernel MPIC device.
5523
55246.7 KVM_CAP_IRQ_XICS
5525--------------------
5526
5527:Architectures: ppc
5528:Target: vcpu
5529:Parameters: args[0] is the XICS device fd;
5530             args[1] is the XICS CPU number (server ID) for this vcpu
5531
5532This capability connects the vcpu to an in-kernel XICS device.
5533
55346.8 KVM_CAP_S390_IRQCHIP
5535------------------------
5536
5537:Architectures: s390
5538:Target: vm
5539:Parameters: none
5540
5541This capability enables the in-kernel irqchip for s390. Please refer to
5542"4.24 KVM_CREATE_IRQCHIP" for details.
5543
55446.9 KVM_CAP_MIPS_FPU
5545--------------------
5546
5547:Architectures: mips
5548:Target: vcpu
5549:Parameters: args[0] is reserved for future use (should be 0).
5550
5551This capability allows the use of the host Floating Point Unit by the guest. It
5552allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
5553done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
5554accessed (depending on the current guest FPU register mode), and the Status.FR,
5555Config5.FRE bits are accessible via the KVM API and also from the guest,
5556depending on them being supported by the FPU.
5557
55586.10 KVM_CAP_MIPS_MSA
5559---------------------
5560
5561:Architectures: mips
5562:Target: vcpu
5563:Parameters: args[0] is reserved for future use (should be 0).
5564
5565This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
5566It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
5567Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
5568registers can be accessed, and the Config5.MSAEn bit is accessible via the
5569KVM API and also from the guest.
5570
55716.74 KVM_CAP_SYNC_REGS
5572----------------------
5573
5574:Architectures: s390, x86
5575:Target: s390: always enabled, x86: vcpu
5576:Parameters: none
5577:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
5578          sets are supported
5579          (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
5580
5581As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
5582KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
5583without having to call SET/GET_*REGS". This reduces overhead by eliminating
5584repeated ioctl calls for setting and/or getting register values. This is
5585particularly important when userspace is making synchronous guest state
5586modifications, e.g. when emulating and/or intercepting instructions in
5587userspace.
5588
5589For s390 specifics, please refer to the source code.
5590
5591For x86:
5592
5593- the register sets to be copied out to kvm_run are selectable
5594  by userspace (rather that all sets being copied out for every exit).
5595- vcpu_events are available in addition to regs and sregs.
5596
5597For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
5598function as an input bit-array field set by userspace to indicate the
5599specific register sets to be copied out on the next exit.
5600
5601To indicate when userspace has modified values that should be copied into
5602the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
5603This is done using the same bitflags as for the 'kvm_valid_regs' field.
5604If the dirty bit is not set, then the register set values will not be copied
5605into the vCPU even if they've been modified.
5606
5607Unused bitfields in the bitarrays must be set to zero.
5608
5609::
5610
5611  struct kvm_sync_regs {
5612        struct kvm_regs regs;
5613        struct kvm_sregs sregs;
5614        struct kvm_vcpu_events events;
5615  };
5616
56176.75 KVM_CAP_PPC_IRQ_XIVE
5618-------------------------
5619
5620:Architectures: ppc
5621:Target: vcpu
5622:Parameters: args[0] is the XIVE device fd;
5623             args[1] is the XIVE CPU number (server ID) for this vcpu
5624
5625This capability connects the vcpu to an in-kernel XIVE device.
5626
56277. Capabilities that can be enabled on VMs
5628==========================================
5629
5630There are certain capabilities that change the behavior of the virtual
5631machine when enabled. To enable them, please see section 4.37. Below
5632you can find a list of capabilities and what their effect on the VM
5633is when enabling them.
5634
5635The following information is provided along with the description:
5636
5637  Architectures:
5638      which instruction set architectures provide this ioctl.
5639      x86 includes both i386 and x86_64.
5640
5641  Parameters:
5642      what parameters are accepted by the capability.
5643
5644  Returns:
5645      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
5646      are not detailed, but errors with specific meanings are.
5647
5648
56497.1 KVM_CAP_PPC_ENABLE_HCALL
5650----------------------------
5651
5652:Architectures: ppc
5653:Parameters: args[0] is the sPAPR hcall number;
5654	     args[1] is 0 to disable, 1 to enable in-kernel handling
5655
5656This capability controls whether individual sPAPR hypercalls (hcalls)
5657get handled by the kernel or not.  Enabling or disabling in-kernel
5658handling of an hcall is effective across the VM.  On creation, an
5659initial set of hcalls are enabled for in-kernel handling, which
5660consists of those hcalls for which in-kernel handlers were implemented
5661before this capability was implemented.  If disabled, the kernel will
5662not to attempt to handle the hcall, but will always exit to userspace
5663to handle it.  Note that it may not make sense to enable some and
5664disable others of a group of related hcalls, but KVM does not prevent
5665userspace from doing that.
5666
5667If the hcall number specified is not one that has an in-kernel
5668implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
5669error.
5670
56717.2 KVM_CAP_S390_USER_SIGP
5672--------------------------
5673
5674:Architectures: s390
5675:Parameters: none
5676
5677This capability controls which SIGP orders will be handled completely in user
5678space. With this capability enabled, all fast orders will be handled completely
5679in the kernel:
5680
5681- SENSE
5682- SENSE RUNNING
5683- EXTERNAL CALL
5684- EMERGENCY SIGNAL
5685- CONDITIONAL EMERGENCY SIGNAL
5686
5687All other orders will be handled completely in user space.
5688
5689Only privileged operation exceptions will be checked for in the kernel (or even
5690in the hardware prior to interception). If this capability is not enabled, the
5691old way of handling SIGP orders is used (partially in kernel and user space).
5692
56937.3 KVM_CAP_S390_VECTOR_REGISTERS
5694---------------------------------
5695
5696:Architectures: s390
5697:Parameters: none
5698:Returns: 0 on success, negative value on error
5699
5700Allows use of the vector registers introduced with z13 processor, and
5701provides for the synchronization between host and user space.  Will
5702return -EINVAL if the machine does not support vectors.
5703
57047.4 KVM_CAP_S390_USER_STSI
5705--------------------------
5706
5707:Architectures: s390
5708:Parameters: none
5709
5710This capability allows post-handlers for the STSI instruction. After
5711initial handling in the kernel, KVM exits to user space with
5712KVM_EXIT_S390_STSI to allow user space to insert further data.
5713
5714Before exiting to userspace, kvm handlers should fill in s390_stsi field of
5715vcpu->run::
5716
5717  struct {
5718	__u64 addr;
5719	__u8 ar;
5720	__u8 reserved;
5721	__u8 fc;
5722	__u8 sel1;
5723	__u16 sel2;
5724  } s390_stsi;
5725
5726  @addr - guest address of STSI SYSIB
5727  @fc   - function code
5728  @sel1 - selector 1
5729  @sel2 - selector 2
5730  @ar   - access register number
5731
5732KVM handlers should exit to userspace with rc = -EREMOTE.
5733
57347.5 KVM_CAP_SPLIT_IRQCHIP
5735-------------------------
5736
5737:Architectures: x86
5738:Parameters: args[0] - number of routes reserved for userspace IOAPICs
5739:Returns: 0 on success, -1 on error
5740
5741Create a local apic for each processor in the kernel. This can be used
5742instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
5743IOAPIC and PIC (and also the PIT, even though this has to be enabled
5744separately).
5745
5746This capability also enables in kernel routing of interrupt requests;
5747when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
5748used in the IRQ routing table.  The first args[0] MSI routes are reserved
5749for the IOAPIC pins.  Whenever the LAPIC receives an EOI for these routes,
5750a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
5751
5752Fails if VCPU has already been created, or if the irqchip is already in the
5753kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
5754
57557.6 KVM_CAP_S390_RI
5756-------------------
5757
5758:Architectures: s390
5759:Parameters: none
5760
5761Allows use of runtime-instrumentation introduced with zEC12 processor.
5762Will return -EINVAL if the machine does not support runtime-instrumentation.
5763Will return -EBUSY if a VCPU has already been created.
5764
57657.7 KVM_CAP_X2APIC_API
5766----------------------
5767
5768:Architectures: x86
5769:Parameters: args[0] - features that should be enabled
5770:Returns: 0 on success, -EINVAL when args[0] contains invalid features
5771
5772Valid feature flags in args[0] are::
5773
5774  #define KVM_X2APIC_API_USE_32BIT_IDS            (1ULL << 0)
5775  #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK  (1ULL << 1)
5776
5777Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
5778KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
5779allowing the use of 32-bit APIC IDs.  See KVM_CAP_X2APIC_API in their
5780respective sections.
5781
5782KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
5783in logical mode or with more than 255 VCPUs.  Otherwise, KVM treats 0xff
5784as a broadcast even in x2APIC mode in order to support physical x2APIC
5785without interrupt remapping.  This is undesirable in logical mode,
5786where 0xff represents CPUs 0-7 in cluster 0.
5787
57887.8 KVM_CAP_S390_USER_INSTR0
5789----------------------------
5790
5791:Architectures: s390
5792:Parameters: none
5793
5794With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
5795be intercepted and forwarded to user space. User space can use this
5796mechanism e.g. to realize 2-byte software breakpoints. The kernel will
5797not inject an operating exception for these instructions, user space has
5798to take care of that.
5799
5800This capability can be enabled dynamically even if VCPUs were already
5801created and are running.
5802
58037.9 KVM_CAP_S390_GS
5804-------------------
5805
5806:Architectures: s390
5807:Parameters: none
5808:Returns: 0 on success; -EINVAL if the machine does not support
5809          guarded storage; -EBUSY if a VCPU has already been created.
5810
5811Allows use of guarded storage for the KVM guest.
5812
58137.10 KVM_CAP_S390_AIS
5814---------------------
5815
5816:Architectures: s390
5817:Parameters: none
5818
5819Allow use of adapter-interruption suppression.
5820:Returns: 0 on success; -EBUSY if a VCPU has already been created.
5821
58227.11 KVM_CAP_PPC_SMT
5823--------------------
5824
5825:Architectures: ppc
5826:Parameters: vsmt_mode, flags
5827
5828Enabling this capability on a VM provides userspace with a way to set
5829the desired virtual SMT mode (i.e. the number of virtual CPUs per
5830virtual core).  The virtual SMT mode, vsmt_mode, must be a power of 2
5831between 1 and 8.  On POWER8, vsmt_mode must also be no greater than
5832the number of threads per subcore for the host.  Currently flags must
5833be 0.  A successful call to enable this capability will result in
5834vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
5835subsequently queried for the VM.  This capability is only supported by
5836HV KVM, and can only be set before any VCPUs have been created.
5837The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
5838modes are available.
5839
58407.12 KVM_CAP_PPC_FWNMI
5841----------------------
5842
5843:Architectures: ppc
5844:Parameters: none
5845
5846With this capability a machine check exception in the guest address
5847space will cause KVM to exit the guest with NMI exit reason. This
5848enables QEMU to build error log and branch to guest kernel registered
5849machine check handling routine. Without this capability KVM will
5850branch to guests' 0x200 interrupt vector.
5851
58527.13 KVM_CAP_X86_DISABLE_EXITS
5853------------------------------
5854
5855:Architectures: x86
5856:Parameters: args[0] defines which exits are disabled
5857:Returns: 0 on success, -EINVAL when args[0] contains invalid exits
5858
5859Valid bits in args[0] are::
5860
5861  #define KVM_X86_DISABLE_EXITS_MWAIT            (1 << 0)
5862  #define KVM_X86_DISABLE_EXITS_HLT              (1 << 1)
5863  #define KVM_X86_DISABLE_EXITS_PAUSE            (1 << 2)
5864  #define KVM_X86_DISABLE_EXITS_CSTATE           (1 << 3)
5865
5866Enabling this capability on a VM provides userspace with a way to no
5867longer intercept some instructions for improved latency in some
5868workloads, and is suggested when vCPUs are associated to dedicated
5869physical CPUs.  More bits can be added in the future; userspace can
5870just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
5871all such vmexits.
5872
5873Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
5874
58757.14 KVM_CAP_S390_HPAGE_1M
5876--------------------------
5877
5878:Architectures: s390
5879:Parameters: none
5880:Returns: 0 on success, -EINVAL if hpage module parameter was not set
5881	  or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
5882	  flag set
5883
5884With this capability the KVM support for memory backing with 1m pages
5885through hugetlbfs can be enabled for a VM. After the capability is
5886enabled, cmma can't be enabled anymore and pfmfi and the storage key
5887interpretation are disabled. If cmma has already been enabled or the
5888hpage module parameter is not set to 1, -EINVAL is returned.
5889
5890While it is generally possible to create a huge page backed VM without
5891this capability, the VM will not be able to run.
5892
58937.15 KVM_CAP_MSR_PLATFORM_INFO
5894------------------------------
5895
5896:Architectures: x86
5897:Parameters: args[0] whether feature should be enabled or not
5898
5899With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
5900a #GP would be raised when the guest tries to access. Currently, this
5901capability does not enable write permissions of this MSR for the guest.
5902
59037.16 KVM_CAP_PPC_NESTED_HV
5904--------------------------
5905
5906:Architectures: ppc
5907:Parameters: none
5908:Returns: 0 on success, -EINVAL when the implementation doesn't support
5909	  nested-HV virtualization.
5910
5911HV-KVM on POWER9 and later systems allows for "nested-HV"
5912virtualization, which provides a way for a guest VM to run guests that
5913can run using the CPU's supervisor mode (privileged non-hypervisor
5914state).  Enabling this capability on a VM depends on the CPU having
5915the necessary functionality and on the facility being enabled with a
5916kvm-hv module parameter.
5917
59187.17 KVM_CAP_EXCEPTION_PAYLOAD
5919------------------------------
5920
5921:Architectures: x86
5922:Parameters: args[0] whether feature should be enabled or not
5923
5924With this capability enabled, CR2 will not be modified prior to the
5925emulated VM-exit when L1 intercepts a #PF exception that occurs in
5926L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
5927the emulated VM-exit when L1 intercepts a #DB exception that occurs in
5928L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
5929#DB) exception for L2, exception.has_payload will be set and the
5930faulting address (or the new DR6 bits*) will be reported in the
5931exception_payload field. Similarly, when userspace injects a #PF (or
5932#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
5933exception.has_payload and to put the faulting address - or the new DR6
5934bits\ [#]_ - in the exception_payload field.
5935
5936This capability also enables exception.pending in struct
5937kvm_vcpu_events, which allows userspace to distinguish between pending
5938and injected exceptions.
5939
5940
5941.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
5942       will clear DR6.RTM.
5943
59447.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
5945
5946:Architectures: x86, arm, arm64, mips
5947:Parameters: args[0] whether feature should be enabled or not
5948
5949Valid flags are::
5950
5951  #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE   (1 << 0)
5952  #define KVM_DIRTY_LOG_INITIALLY_SET           (1 << 1)
5953
5954With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
5955automatically clear and write-protect all pages that are returned as dirty.
5956Rather, userspace will have to do this operation separately using
5957KVM_CLEAR_DIRTY_LOG.
5958
5959At the cost of a slightly more complicated operation, this provides better
5960scalability and responsiveness for two reasons.  First,
5961KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
5962than requiring to sync a full memslot; this ensures that KVM does not
5963take spinlocks for an extended period of time.  Second, in some cases a
5964large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
5965userspace actually using the data in the page.  Pages can be modified
5966during this time, which is inefficient for both the guest and userspace:
5967the guest will incur a higher penalty due to write protection faults,
5968while userspace can see false reports of dirty pages.  Manual reprotection
5969helps reducing this time, improving guest performance and reducing the
5970number of dirty log false positives.
5971
5972With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
5973will be initialized to 1 when created.  This also improves performance because
5974dirty logging can be enabled gradually in small chunks on the first call
5975to KVM_CLEAR_DIRTY_LOG.  KVM_DIRTY_LOG_INITIALLY_SET depends on
5976KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
5977x86 and arm64 for now).
5978
5979KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
5980KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
5981it hard or impossible to use it correctly.  The availability of
5982KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
5983Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
5984
59857.19 KVM_CAP_PPC_SECURE_GUEST
5986------------------------------
5987
5988:Architectures: ppc
5989
5990This capability indicates that KVM is running on a host that has
5991ultravisor firmware and thus can support a secure guest.  On such a
5992system, a guest can ask the ultravisor to make it a secure guest,
5993one whose memory is inaccessible to the host except for pages which
5994are explicitly requested to be shared with the host.  The ultravisor
5995notifies KVM when a guest requests to become a secure guest, and KVM
5996has the opportunity to veto the transition.
5997
5998If present, this capability can be enabled for a VM, meaning that KVM
5999will allow the transition to secure guest mode.  Otherwise KVM will
6000veto the transition.
6001
60027.20 KVM_CAP_HALT_POLL
6003----------------------
6004
6005:Architectures: all
6006:Target: VM
6007:Parameters: args[0] is the maximum poll time in nanoseconds
6008:Returns: 0 on success; -1 on error
6009
6010This capability overrides the kvm module parameter halt_poll_ns for the
6011target VM.
6012
6013VCPU polling allows a VCPU to poll for wakeup events instead of immediately
6014scheduling during guest halts. The maximum time a VCPU can spend polling is
6015controlled by the kvm module parameter halt_poll_ns. This capability allows
6016the maximum halt time to specified on a per-VM basis, effectively overriding
6017the module parameter for the target VM.
6018
60197.21 KVM_CAP_X86_USER_SPACE_MSR
6020-------------------------------
6021
6022:Architectures: x86
6023:Target: VM
6024:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
6025:Returns: 0 on success; -1 on error
6026
6027This capability enables trapping of #GP invoking RDMSR and WRMSR instructions
6028into user space.
6029
6030When a guest requests to read or write an MSR, KVM may not implement all MSRs
6031that are relevant to a respective system. It also does not differentiate by
6032CPU type.
6033
6034To allow more fine grained control over MSR handling, user space may enable
6035this capability. With it enabled, MSR accesses that match the mask specified in
6036args[0] and trigger a #GP event inside the guest by KVM will instead trigger
6037KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications which user space
6038can then handle to implement model specific MSR handling and/or user notifications
6039to inform a user that an MSR was not handled.
6040
60418. Other capabilities.
6042======================
6043
6044This section lists capabilities that give information about other
6045features of the KVM implementation.
6046
60478.1 KVM_CAP_PPC_HWRNG
6048---------------------
6049
6050:Architectures: ppc
6051
6052This capability, if KVM_CHECK_EXTENSION indicates that it is
6053available, means that the kernel has an implementation of the
6054H_RANDOM hypercall backed by a hardware random-number generator.
6055If present, the kernel H_RANDOM handler can be enabled for guest use
6056with the KVM_CAP_PPC_ENABLE_HCALL capability.
6057
60588.2 KVM_CAP_HYPERV_SYNIC
6059------------------------
6060
6061:Architectures: x86
6062
6063This capability, if KVM_CHECK_EXTENSION indicates that it is
6064available, means that the kernel has an implementation of the
6065Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
6066used to support Windows Hyper-V based guest paravirt drivers(VMBus).
6067
6068In order to use SynIC, it has to be activated by setting this
6069capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
6070will disable the use of APIC hardware virtualization even if supported
6071by the CPU, as it's incompatible with SynIC auto-EOI behavior.
6072
60738.3 KVM_CAP_PPC_RADIX_MMU
6074-------------------------
6075
6076:Architectures: ppc
6077
6078This capability, if KVM_CHECK_EXTENSION indicates that it is
6079available, means that the kernel can support guests using the
6080radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
6081processor).
6082
60838.4 KVM_CAP_PPC_HASH_MMU_V3
6084---------------------------
6085
6086:Architectures: ppc
6087
6088This capability, if KVM_CHECK_EXTENSION indicates that it is
6089available, means that the kernel can support guests using the
6090hashed page table MMU defined in Power ISA V3.00 (as implemented in
6091the POWER9 processor), including in-memory segment tables.
6092
60938.5 KVM_CAP_MIPS_VZ
6094-------------------
6095
6096:Architectures: mips
6097
6098This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6099it is available, means that full hardware assisted virtualization capabilities
6100of the hardware are available for use through KVM. An appropriate
6101KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
6102utilises it.
6103
6104If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6105available, it means that the VM is using full hardware assisted virtualization
6106capabilities of the hardware. This is useful to check after creating a VM with
6107KVM_VM_MIPS_DEFAULT.
6108
6109The value returned by KVM_CHECK_EXTENSION should be compared against known
6110values (see below). All other values are reserved. This is to allow for the
6111possibility of other hardware assisted virtualization implementations which
6112may be incompatible with the MIPS VZ ASE.
6113
6114==  ==========================================================================
6115 0  The trap & emulate implementation is in use to run guest code in user
6116    mode. Guest virtual memory segments are rearranged to fit the guest in the
6117    user mode address space.
6118
6119 1  The MIPS VZ ASE is in use, providing full hardware assisted
6120    virtualization, including standard guest virtual memory segments.
6121==  ==========================================================================
6122
61238.6 KVM_CAP_MIPS_TE
6124-------------------
6125
6126:Architectures: mips
6127
6128This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
6129it is available, means that the trap & emulate implementation is available to
6130run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
6131assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
6132to KVM_CREATE_VM to create a VM which utilises it.
6133
6134If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
6135available, it means that the VM is using trap & emulate.
6136
61378.7 KVM_CAP_MIPS_64BIT
6138----------------------
6139
6140:Architectures: mips
6141
6142This capability indicates the supported architecture type of the guest, i.e. the
6143supported register and address width.
6144
6145The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
6146kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
6147be checked specifically against known values (see below). All other values are
6148reserved.
6149
6150==  ========================================================================
6151 0  MIPS32 or microMIPS32.
6152    Both registers and addresses are 32-bits wide.
6153    It will only be possible to run 32-bit guest code.
6154
6155 1  MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
6156    Registers are 64-bits wide, but addresses are 32-bits wide.
6157    64-bit guest code may run but cannot access MIPS64 memory segments.
6158    It will also be possible to run 32-bit guest code.
6159
6160 2  MIPS64 or microMIPS64 with access to all address segments.
6161    Both registers and addresses are 64-bits wide.
6162    It will be possible to run 64-bit or 32-bit guest code.
6163==  ========================================================================
6164
61658.9 KVM_CAP_ARM_USER_IRQ
6166------------------------
6167
6168:Architectures: arm, arm64
6169
6170This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
6171that if userspace creates a VM without an in-kernel interrupt controller, it
6172will be notified of changes to the output level of in-kernel emulated devices,
6173which can generate virtual interrupts, presented to the VM.
6174For such VMs, on every return to userspace, the kernel
6175updates the vcpu's run->s.regs.device_irq_level field to represent the actual
6176output level of the device.
6177
6178Whenever kvm detects a change in the device output level, kvm guarantees at
6179least one return to userspace before running the VM.  This exit could either
6180be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
6181userspace can always sample the device output level and re-compute the state of
6182the userspace interrupt controller.  Userspace should always check the state
6183of run->s.regs.device_irq_level on every kvm exit.
6184The value in run->s.regs.device_irq_level can represent both level and edge
6185triggered interrupt signals, depending on the device.  Edge triggered interrupt
6186signals will exit to userspace with the bit in run->s.regs.device_irq_level
6187set exactly once per edge signal.
6188
6189The field run->s.regs.device_irq_level is available independent of
6190run->kvm_valid_regs or run->kvm_dirty_regs bits.
6191
6192If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
6193number larger than 0 indicating the version of this capability is implemented
6194and thereby which bits in run->s.regs.device_irq_level can signal values.
6195
6196Currently the following bits are defined for the device_irq_level bitmap::
6197
6198  KVM_CAP_ARM_USER_IRQ >= 1:
6199
6200    KVM_ARM_DEV_EL1_VTIMER -  EL1 virtual timer
6201    KVM_ARM_DEV_EL1_PTIMER -  EL1 physical timer
6202    KVM_ARM_DEV_PMU        -  ARM PMU overflow interrupt signal
6203
6204Future versions of kvm may implement additional events. These will get
6205indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
6206listed above.
6207
62088.10 KVM_CAP_PPC_SMT_POSSIBLE
6209-----------------------------
6210
6211:Architectures: ppc
6212
6213Querying this capability returns a bitmap indicating the possible
6214virtual SMT modes that can be set using KVM_CAP_PPC_SMT.  If bit N
6215(counting from the right) is set, then a virtual SMT mode of 2^N is
6216available.
6217
62188.11 KVM_CAP_HYPERV_SYNIC2
6219--------------------------
6220
6221:Architectures: x86
6222
6223This capability enables a newer version of Hyper-V Synthetic interrupt
6224controller (SynIC).  The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
6225doesn't clear SynIC message and event flags pages when they are enabled by
6226writing to the respective MSRs.
6227
62288.12 KVM_CAP_HYPERV_VP_INDEX
6229----------------------------
6230
6231:Architectures: x86
6232
6233This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr.  Its
6234value is used to denote the target vcpu for a SynIC interrupt.  For
6235compatibilty, KVM initializes this msr to KVM's internal vcpu index.  When this
6236capability is absent, userspace can still query this msr's value.
6237
62388.13 KVM_CAP_S390_AIS_MIGRATION
6239-------------------------------
6240
6241:Architectures: s390
6242:Parameters: none
6243
6244This capability indicates if the flic device will be able to get/set the
6245AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
6246to discover this without having to create a flic device.
6247
62488.14 KVM_CAP_S390_PSW
6249---------------------
6250
6251:Architectures: s390
6252
6253This capability indicates that the PSW is exposed via the kvm_run structure.
6254
62558.15 KVM_CAP_S390_GMAP
6256----------------------
6257
6258:Architectures: s390
6259
6260This capability indicates that the user space memory used as guest mapping can
6261be anywhere in the user memory address space, as long as the memory slots are
6262aligned and sized to a segment (1MB) boundary.
6263
62648.16 KVM_CAP_S390_COW
6265---------------------
6266
6267:Architectures: s390
6268
6269This capability indicates that the user space memory used as guest mapping can
6270use copy-on-write semantics as well as dirty pages tracking via read-only page
6271tables.
6272
62738.17 KVM_CAP_S390_BPB
6274---------------------
6275
6276:Architectures: s390
6277
6278This capability indicates that kvm will implement the interfaces to handle
6279reset, migration and nested KVM for branch prediction blocking. The stfle
6280facility 82 should not be provided to the guest without this capability.
6281
62828.18 KVM_CAP_HYPERV_TLBFLUSH
6283----------------------------
6284
6285:Architectures: x86
6286
6287This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
6288hypercalls:
6289HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
6290HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
6291
62928.19 KVM_CAP_ARM_INJECT_SERROR_ESR
6293----------------------------------
6294
6295:Architectures: arm, arm64
6296
6297This capability indicates that userspace can specify (via the
6298KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
6299takes a virtual SError interrupt exception.
6300If KVM advertises this capability, userspace can only specify the ISS field for
6301the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
6302CPU when the exception is taken. If this virtual SError is taken to EL1 using
6303AArch64, this value will be reported in the ISS field of ESR_ELx.
6304
6305See KVM_CAP_VCPU_EVENTS for more details.
6306
63078.20 KVM_CAP_HYPERV_SEND_IPI
6308----------------------------
6309
6310:Architectures: x86
6311
6312This capability indicates that KVM supports paravirtualized Hyper-V IPI send
6313hypercalls:
6314HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
6315
63168.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
6317-----------------------------------
6318
6319:Architectures: x86
6320
6321This capability indicates that KVM running on top of Hyper-V hypervisor
6322enables Direct TLB flush for its guests meaning that TLB flush
6323hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
6324Due to the different ABI for hypercall parameters between Hyper-V and
6325KVM, enabling this capability effectively disables all hypercall
6326handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
6327flush hypercalls by Hyper-V) so userspace should disable KVM identification
6328in CPUID and only exposes Hyper-V identification. In this case, guest
6329thinks it's running on Hyper-V and only use Hyper-V hypercalls.
6330
63318.22 KVM_CAP_S390_VCPU_RESETS
6332-----------------------------
6333
6334:Architectures: s390
6335
6336This capability indicates that the KVM_S390_NORMAL_RESET and
6337KVM_S390_CLEAR_RESET ioctls are available.
6338
63398.23 KVM_CAP_S390_PROTECTED
6340---------------------------
6341
6342:Architectures: s390
6343
6344This capability indicates that the Ultravisor has been initialized and
6345KVM can therefore start protected VMs.
6346This capability governs the KVM_S390_PV_COMMAND ioctl and the
6347KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
6348guests when the state change is invalid.
6349
63508.24 KVM_CAP_STEAL_TIME
6351-----------------------
6352
6353:Architectures: arm64, x86
6354
6355This capability indicates that KVM supports steal time accounting.
6356When steal time accounting is supported it may be enabled with
6357architecture-specific interfaces.  This capability and the architecture-
6358specific interfaces must be consistent, i.e. if one says the feature
6359is supported, than the other should as well and vice versa.  For arm64
6360see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
6361For x86 see Documentation/virt/kvm/msr.rst "MSR_KVM_STEAL_TIME".
6362
63638.25 KVM_CAP_S390_DIAG318
6364-------------------------
6365
6366:Architectures: s390
6367
6368This capability enables a guest to set information about its control program
6369(i.e. guest kernel type and version). The information is helpful during
6370system/firmware service events, providing additional data about the guest
6371environments running on the machine.
6372
6373The information is associated with the DIAGNOSE 0x318 instruction, which sets
6374an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
6375a 7-byte Control Program Version Code (CPVC). The CPNC determines what
6376environment the control program is running in (e.g. Linux, z/VM...), and the
6377CPVC is used for information specific to OS (e.g. Linux version, Linux
6378distribution...)
6379
6380If this capability is available, then the CPNC and CPVC can be synchronized
6381between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
6382
63838.26 KVM_CAP_X86_USER_SPACE_MSR
6384-------------------------------
6385
6386:Architectures: x86
6387
6388This capability indicates that KVM supports deflection of MSR reads and
6389writes to user space. It can be enabled on a VM level. If enabled, MSR
6390accesses that would usually trigger a #GP by KVM into the guest will
6391instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
6392KVM_EXIT_X86_WRMSR exit notifications.
6393
63948.27 KVM_X86_SET_MSR_FILTER
6395---------------------------
6396
6397:Architectures: x86
6398
6399This capability indicates that KVM supports that accesses to user defined MSRs
6400may be rejected. With this capability exposed, KVM exports new VM ioctl
6401KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
6402ranges that KVM should reject access to.
6403
6404In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
6405trap and emulate MSRs that are outside of the scope of KVM as well as
6406limit the attack surface on KVM's MSR emulation code.
6407
64088.28 KVM_CAP_ENFORCE_PV_CPUID
6409-----------------------------
6410
6411Architectures: x86
6412
6413When enabled, KVM will disable paravirtual features provided to the
6414guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
6415(0x40000001). Otherwise, a guest may use the paravirtual features
6416regardless of what has actually been exposed through the CPUID leaf.
6417
6418
64198.29 KVM_CAP_DIRTY_LOG_RING
6420---------------------------
6421
6422:Architectures: x86
6423:Parameters: args[0] - size of the dirty log ring
6424
6425KVM is capable of tracking dirty memory using ring buffers that are
6426mmaped into userspace; there is one dirty ring per vcpu.
6427
6428The dirty ring is available to userspace as an array of
6429``struct kvm_dirty_gfn``.  Each dirty entry it's defined as::
6430
6431  struct kvm_dirty_gfn {
6432          __u32 flags;
6433          __u32 slot; /* as_id | slot_id */
6434          __u64 offset;
6435  };
6436
6437The following values are defined for the flags field to define the
6438current state of the entry::
6439
6440  #define KVM_DIRTY_GFN_F_DIRTY           BIT(0)
6441  #define KVM_DIRTY_GFN_F_RESET           BIT(1)
6442  #define KVM_DIRTY_GFN_F_MASK            0x3
6443
6444Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
6445ioctl to enable this capability for the new guest and set the size of
6446the rings.  Enabling the capability is only allowed before creating any
6447vCPU, and the size of the ring must be a power of two.  The larger the
6448ring buffer, the less likely the ring is full and the VM is forced to
6449exit to userspace. The optimal size depends on the workload, but it is
6450recommended that it be at least 64 KiB (4096 entries).
6451
6452Just like for dirty page bitmaps, the buffer tracks writes to
6453all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
6454set in KVM_SET_USER_MEMORY_REGION.  Once a memory region is registered
6455with the flag set, userspace can start harvesting dirty pages from the
6456ring buffer.
6457
6458An entry in the ring buffer can be unused (flag bits ``00``),
6459dirty (flag bits ``01``) or harvested (flag bits ``1X``).  The
6460state machine for the entry is as follows::
6461
6462          dirtied         harvested        reset
6463     00 -----------> 01 -------------> 1X -------+
6464      ^                                          |
6465      |                                          |
6466      +------------------------------------------+
6467
6468To harvest the dirty pages, userspace accesses the mmaped ring buffer
6469to read the dirty GFNs.  If the flags has the DIRTY bit set (at this stage
6470the RESET bit must be cleared), then it means this GFN is a dirty GFN.
6471The userspace should harvest this GFN and mark the flags from state
6472``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
6473to show that this GFN is harvested and waiting for a reset), and move
6474on to the next GFN.  The userspace should continue to do this until the
6475flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
6476all the dirty GFNs that were available.
6477
6478It's not necessary for userspace to harvest the all dirty GFNs at once.
6479However it must collect the dirty GFNs in sequence, i.e., the userspace
6480program cannot skip one dirty GFN to collect the one next to it.
6481
6482After processing one or more entries in the ring buffer, userspace
6483calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
6484it, so that the kernel will reprotect those collected GFNs.
6485Therefore, the ioctl must be called *before* reading the content of
6486the dirty pages.
6487
6488The dirty ring can get full.  When it happens, the KVM_RUN of the
6489vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
6490
6491The dirty ring interface has a major difference comparing to the
6492KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
6493userspace, it's still possible that the kernel has not yet flushed the
6494processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
6495flushing is done by the KVM_GET_DIRTY_LOG ioctl).  To achieve that, one
6496needs to kick the vcpu out of KVM_RUN using a signal.  The resulting
6497vmexit ensures that all dirty GFNs are flushed to the dirty rings.
6498
6499NOTE: the capability KVM_CAP_DIRTY_LOG_RING and the corresponding
6500ioctl KVM_RESET_DIRTY_RINGS are mutual exclusive to the existing ioctls
6501KVM_GET_DIRTY_LOG and KVM_CLEAR_DIRTY_LOG.  After enabling
6502KVM_CAP_DIRTY_LOG_RING with an acceptable dirty ring size, the virtual
6503machine will switch to ring-buffer dirty page tracking and further
6504KVM_GET_DIRTY_LOG or KVM_CLEAR_DIRTY_LOG ioctls will fail.
6505