xref: /openbmc/linux/Documentation/virt/kvm/api.rst (revision a90fa0ad)
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 although 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
154On arm64, the physical address size for a VM (IPA Size limit) is limited
155to 40bits by default. The limit can be configured if the host supports the
156extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
157KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
158identifier, where IPA_Bits is the maximum width of any physical
159address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
160machine type identifier.
161
162e.g, to configure a guest to use 48bit physical address size::
163
164    vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
165
166The requested size (IPA_Bits) must be:
167
168 ==   =========================================================
169  0   Implies default size, 40bits (for backward compatibility)
170  N   Implies N bits, where N is a positive integer such that,
171      32 <= N <= Host_IPA_Limit
172 ==   =========================================================
173
174Host_IPA_Limit is the maximum possible value for IPA_Bits on the host and
175is dependent on the CPU capability and the kernel configuration. The limit can
176be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the KVM_CHECK_EXTENSION
177ioctl() at run-time.
178
179Creation of the VM will fail if the requested IPA size (whether it is
180implicit or explicit) is unsupported on the host.
181
182Please note that configuring the IPA size does not affect the capability
183exposed by the guest CPUs in ID_AA64MMFR0_EL1[PARange]. It only affects
184size of the address translated by the stage2 level (guest physical to
185host physical address translations).
186
187
1884.3 KVM_GET_MSR_INDEX_LIST, KVM_GET_MSR_FEATURE_INDEX_LIST
189----------------------------------------------------------
190
191:Capability: basic, KVM_CAP_GET_MSR_FEATURES for KVM_GET_MSR_FEATURE_INDEX_LIST
192:Architectures: x86
193:Type: system ioctl
194:Parameters: struct kvm_msr_list (in/out)
195:Returns: 0 on success; -1 on error
196
197Errors:
198
199  ======     ============================================================
200  EFAULT     the msr index list cannot be read from or written to
201  E2BIG      the msr index list is too big to fit in the array specified by
202             the user.
203  ======     ============================================================
204
205::
206
207  struct kvm_msr_list {
208	__u32 nmsrs; /* number of msrs in entries */
209	__u32 indices[0];
210  };
211
212The user fills in the size of the indices array in nmsrs, and in return
213kvm adjusts nmsrs to reflect the actual number of msrs and fills in the
214indices array with their numbers.
215
216KVM_GET_MSR_INDEX_LIST returns the guest msrs that are supported.  The list
217varies by kvm version and host processor, but does not change otherwise.
218
219Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
220not returned in the MSR list, as different vcpus can have a different number
221of banks, as set via the KVM_X86_SETUP_MCE ioctl.
222
223KVM_GET_MSR_FEATURE_INDEX_LIST returns the list of MSRs that can be passed
224to the KVM_GET_MSRS system ioctl.  This lets userspace probe host capabilities
225and processor features that are exposed via MSRs (e.g., VMX capabilities).
226This list also varies by kvm version and host processor, but does not change
227otherwise.
228
229
2304.4 KVM_CHECK_EXTENSION
231-----------------------
232
233:Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
234:Architectures: all
235:Type: system ioctl, vm ioctl
236:Parameters: extension identifier (KVM_CAP_*)
237:Returns: 0 if unsupported; 1 (or some other positive integer) if supported
238
239The API allows the application to query about extensions to the core
240kvm API.  Userspace passes an extension identifier (an integer) and
241receives an integer that describes the extension availability.
242Generally 0 means no and 1 means yes, but some extensions may report
243additional information in the integer return value.
244
245Based on their initialization different VMs may have different capabilities.
246It is thus encouraged to use the vm ioctl to query for capabilities (available
247with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
248
2494.5 KVM_GET_VCPU_MMAP_SIZE
250--------------------------
251
252:Capability: basic
253:Architectures: all
254:Type: system ioctl
255:Parameters: none
256:Returns: size of vcpu mmap area, in bytes
257
258The KVM_RUN ioctl (cf.) communicates with userspace via a shared
259memory region.  This ioctl returns the size of that region.  See the
260KVM_RUN documentation for details.
261
262Besides the size of the KVM_RUN communication region, other areas of
263the VCPU file descriptor can be mmap-ed, including:
264
265- if KVM_CAP_COALESCED_MMIO is available, a page at
266  KVM_COALESCED_MMIO_PAGE_OFFSET * PAGE_SIZE; for historical reasons,
267  this page is included in the result of KVM_GET_VCPU_MMAP_SIZE.
268  KVM_CAP_COALESCED_MMIO is not documented yet.
269
270- if KVM_CAP_DIRTY_LOG_RING is available, a number of pages at
271  KVM_DIRTY_LOG_PAGE_OFFSET * PAGE_SIZE.  For more information on
272  KVM_CAP_DIRTY_LOG_RING, see section 8.3.
273
274
2754.7 KVM_CREATE_VCPU
276-------------------
277
278:Capability: basic
279:Architectures: all
280:Type: vm ioctl
281:Parameters: vcpu id (apic id on x86)
282:Returns: vcpu fd on success, -1 on error
283
284This API adds a vcpu to a virtual machine. No more than max_vcpus may be added.
285The vcpu id is an integer in the range [0, max_vcpu_id).
286
287The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
288the KVM_CHECK_EXTENSION ioctl() at run-time.
289The maximum possible value for max_vcpus can be retrieved using the
290KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
291
292If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
293cpus max.
294If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
295same as the value returned from KVM_CAP_NR_VCPUS.
296
297The maximum possible value for max_vcpu_id can be retrieved using the
298KVM_CAP_MAX_VCPU_ID of the KVM_CHECK_EXTENSION ioctl() at run-time.
299
300If the KVM_CAP_MAX_VCPU_ID does not exist, you should assume that max_vcpu_id
301is the same as the value returned from KVM_CAP_MAX_VCPUS.
302
303On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
304threads in one or more virtual CPU cores.  (This is because the
305hardware requires all the hardware threads in a CPU core to be in the
306same partition.)  The KVM_CAP_PPC_SMT capability indicates the number
307of vcpus per virtual core (vcore).  The vcore id is obtained by
308dividing the vcpu id by the number of vcpus per vcore.  The vcpus in a
309given vcore will always be in the same physical core as each other
310(though that might be a different physical core from time to time).
311Userspace can control the threading (SMT) mode of the guest by its
312allocation of vcpu ids.  For example, if userspace wants
313single-threaded guest vcpus, it should make all vcpu ids be a multiple
314of the number of vcpus per vcore.
315
316For virtual cpus that have been created with S390 user controlled virtual
317machines, the resulting vcpu fd can be memory mapped at page offset
318KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
319cpu's hardware control block.
320
321
3224.8 KVM_GET_DIRTY_LOG (vm ioctl)
323--------------------------------
324
325:Capability: basic
326:Architectures: all
327:Type: vm ioctl
328:Parameters: struct kvm_dirty_log (in/out)
329:Returns: 0 on success, -1 on error
330
331::
332
333  /* for KVM_GET_DIRTY_LOG */
334  struct kvm_dirty_log {
335	__u32 slot;
336	__u32 padding;
337	union {
338		void __user *dirty_bitmap; /* one bit per page */
339		__u64 padding;
340	};
341  };
342
343Given a memory slot, return a bitmap containing any pages dirtied
344since the last call to this ioctl.  Bit 0 is the first page in the
345memory slot.  Ensure the entire structure is cleared to avoid padding
346issues.
347
348If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
349the address space for which you want to return the dirty bitmap.  See
350KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
351
352The bits in the dirty bitmap are cleared before the ioctl returns, unless
353KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is enabled.  For more information,
354see the description of the capability.
355
356Note that the Xen shared info page, if configured, shall always be assumed
357to be dirty. KVM will not explicitly mark it such.
358
359
3604.10 KVM_RUN
361------------
362
363:Capability: basic
364:Architectures: all
365:Type: vcpu ioctl
366:Parameters: none
367:Returns: 0 on success, -1 on error
368
369Errors:
370
371  =======    ==============================================================
372  EINTR      an unmasked signal is pending
373  ENOEXEC    the vcpu hasn't been initialized or the guest tried to execute
374             instructions from device memory (arm64)
375  ENOSYS     data abort outside memslots with no syndrome info and
376             KVM_CAP_ARM_NISV_TO_USER not enabled (arm64)
377  EPERM      SVE feature set but not finalized (arm64)
378  =======    ==============================================================
379
380This ioctl is used to run a guest virtual cpu.  While there are no
381explicit parameters, there is an implicit parameter block that can be
382obtained by mmap()ing the vcpu fd at offset 0, with the size given by
383KVM_GET_VCPU_MMAP_SIZE.  The parameter block is formatted as a 'struct
384kvm_run' (see below).
385
386
3874.11 KVM_GET_REGS
388-----------------
389
390:Capability: basic
391:Architectures: all except arm64
392:Type: vcpu ioctl
393:Parameters: struct kvm_regs (out)
394:Returns: 0 on success, -1 on error
395
396Reads the general purpose registers from the vcpu.
397
398::
399
400  /* x86 */
401  struct kvm_regs {
402	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
403	__u64 rax, rbx, rcx, rdx;
404	__u64 rsi, rdi, rsp, rbp;
405	__u64 r8,  r9,  r10, r11;
406	__u64 r12, r13, r14, r15;
407	__u64 rip, rflags;
408  };
409
410  /* mips */
411  struct kvm_regs {
412	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
413	__u64 gpr[32];
414	__u64 hi;
415	__u64 lo;
416	__u64 pc;
417  };
418
419
4204.12 KVM_SET_REGS
421-----------------
422
423:Capability: basic
424:Architectures: all except arm64
425:Type: vcpu ioctl
426:Parameters: struct kvm_regs (in)
427:Returns: 0 on success, -1 on error
428
429Writes the general purpose registers into the vcpu.
430
431See KVM_GET_REGS for the data structure.
432
433
4344.13 KVM_GET_SREGS
435------------------
436
437:Capability: basic
438:Architectures: x86, ppc
439:Type: vcpu ioctl
440:Parameters: struct kvm_sregs (out)
441:Returns: 0 on success, -1 on error
442
443Reads special registers from the vcpu.
444
445::
446
447  /* x86 */
448  struct kvm_sregs {
449	struct kvm_segment cs, ds, es, fs, gs, ss;
450	struct kvm_segment tr, ldt;
451	struct kvm_dtable gdt, idt;
452	__u64 cr0, cr2, cr3, cr4, cr8;
453	__u64 efer;
454	__u64 apic_base;
455	__u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
456  };
457
458  /* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
459
460interrupt_bitmap is a bitmap of pending external interrupts.  At most
461one bit may be set.  This interrupt has been acknowledged by the APIC
462but not yet injected into the cpu core.
463
464
4654.14 KVM_SET_SREGS
466------------------
467
468:Capability: basic
469:Architectures: x86, ppc
470:Type: vcpu ioctl
471:Parameters: struct kvm_sregs (in)
472:Returns: 0 on success, -1 on error
473
474Writes special registers into the vcpu.  See KVM_GET_SREGS for the
475data structures.
476
477
4784.15 KVM_TRANSLATE
479------------------
480
481:Capability: basic
482:Architectures: x86
483:Type: vcpu ioctl
484:Parameters: struct kvm_translation (in/out)
485:Returns: 0 on success, -1 on error
486
487Translates a virtual address according to the vcpu's current address
488translation mode.
489
490::
491
492  struct kvm_translation {
493	/* in */
494	__u64 linear_address;
495
496	/* out */
497	__u64 physical_address;
498	__u8  valid;
499	__u8  writeable;
500	__u8  usermode;
501	__u8  pad[5];
502  };
503
504
5054.16 KVM_INTERRUPT
506------------------
507
508:Capability: basic
509:Architectures: x86, ppc, mips, riscv
510:Type: vcpu ioctl
511:Parameters: struct kvm_interrupt (in)
512:Returns: 0 on success, negative on failure.
513
514Queues a hardware interrupt vector to be injected.
515
516::
517
518  /* for KVM_INTERRUPT */
519  struct kvm_interrupt {
520	/* in */
521	__u32 irq;
522  };
523
524X86:
525^^^^
526
527:Returns:
528
529	========= ===================================
530	  0       on success,
531	 -EEXIST  if an interrupt is already enqueued
532	 -EINVAL  the irq number is invalid
533	 -ENXIO   if the PIC is in the kernel
534	 -EFAULT  if the pointer is invalid
535	========= ===================================
536
537Note 'irq' is an interrupt vector, not an interrupt pin or line. This
538ioctl is useful if the in-kernel PIC is not used.
539
540PPC:
541^^^^
542
543Queues an external interrupt to be injected. This ioctl is overleaded
544with 3 different irq values:
545
546a) KVM_INTERRUPT_SET
547
548   This injects an edge type external interrupt into the guest once it's ready
549   to receive interrupts. When injected, the interrupt is done.
550
551b) KVM_INTERRUPT_UNSET
552
553   This unsets any pending interrupt.
554
555   Only available with KVM_CAP_PPC_UNSET_IRQ.
556
557c) KVM_INTERRUPT_SET_LEVEL
558
559   This injects a level type external interrupt into the guest context. The
560   interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
561   is triggered.
562
563   Only available with KVM_CAP_PPC_IRQ_LEVEL.
564
565Note that any value for 'irq' other than the ones stated above is invalid
566and incurs unexpected behavior.
567
568This is an asynchronous vcpu ioctl and can be invoked from any thread.
569
570MIPS:
571^^^^^
572
573Queues an external interrupt to be injected into the virtual CPU. A negative
574interrupt number dequeues the interrupt.
575
576This is an asynchronous vcpu ioctl and can be invoked from any thread.
577
578RISC-V:
579^^^^^^^
580
581Queues an external interrupt to be injected into the virutal CPU. This ioctl
582is overloaded with 2 different irq values:
583
584a) KVM_INTERRUPT_SET
585
586   This sets external interrupt for a virtual CPU and it will receive
587   once it is ready.
588
589b) KVM_INTERRUPT_UNSET
590
591   This clears pending external interrupt for a virtual CPU.
592
593This is an asynchronous vcpu ioctl and can be invoked from any thread.
594
595
5964.17 KVM_DEBUG_GUEST
597--------------------
598
599:Capability: basic
600:Architectures: none
601:Type: vcpu ioctl
602:Parameters: none)
603:Returns: -1 on error
604
605Support for this has been removed.  Use KVM_SET_GUEST_DEBUG instead.
606
607
6084.18 KVM_GET_MSRS
609-----------------
610
611:Capability: basic (vcpu), KVM_CAP_GET_MSR_FEATURES (system)
612:Architectures: x86
613:Type: system ioctl, vcpu ioctl
614:Parameters: struct kvm_msrs (in/out)
615:Returns: number of msrs successfully returned;
616          -1 on error
617
618When used as a system ioctl:
619Reads the values of MSR-based features that are available for the VM.  This
620is similar to KVM_GET_SUPPORTED_CPUID, but it returns MSR indices and values.
621The list of msr-based features can be obtained using KVM_GET_MSR_FEATURE_INDEX_LIST
622in a system ioctl.
623
624When used as a vcpu ioctl:
625Reads model-specific registers from the vcpu.  Supported msr indices can
626be obtained using KVM_GET_MSR_INDEX_LIST in a system ioctl.
627
628::
629
630  struct kvm_msrs {
631	__u32 nmsrs; /* number of msrs in entries */
632	__u32 pad;
633
634	struct kvm_msr_entry entries[0];
635  };
636
637  struct kvm_msr_entry {
638	__u32 index;
639	__u32 reserved;
640	__u64 data;
641  };
642
643Application code should set the 'nmsrs' member (which indicates the
644size of the entries array) and the 'index' member of each array entry.
645kvm will fill in the 'data' member.
646
647
6484.19 KVM_SET_MSRS
649-----------------
650
651:Capability: basic
652:Architectures: x86
653:Type: vcpu ioctl
654:Parameters: struct kvm_msrs (in)
655:Returns: number of msrs successfully set (see below), -1 on error
656
657Writes model-specific registers to the vcpu.  See KVM_GET_MSRS for the
658data structures.
659
660Application code should set the 'nmsrs' member (which indicates the
661size of the entries array), and the 'index' and 'data' members of each
662array entry.
663
664It tries to set the MSRs in array entries[] one by one. If setting an MSR
665fails, e.g., due to setting reserved bits, the MSR isn't supported/emulated
666by KVM, etc..., it stops processing the MSR list and returns the number of
667MSRs that have been set successfully.
668
669
6704.20 KVM_SET_CPUID
671------------------
672
673:Capability: basic
674:Architectures: x86
675:Type: vcpu ioctl
676:Parameters: struct kvm_cpuid (in)
677:Returns: 0 on success, -1 on error
678
679Defines the vcpu responses to the cpuid instruction.  Applications
680should use the KVM_SET_CPUID2 ioctl if available.
681
682Caveat emptor:
683  - If this IOCTL fails, KVM gives no guarantees that previous valid CPUID
684    configuration (if there is) is not corrupted. Userspace can get a copy
685    of the resulting CPUID configuration through KVM_GET_CPUID2 in case.
686  - Using KVM_SET_CPUID{,2} after KVM_RUN, i.e. changing the guest vCPU model
687    after running the guest, may cause guest instability.
688  - Using heterogeneous CPUID configurations, modulo APIC IDs, topology, etc...
689    may cause guest instability.
690
691::
692
693  struct kvm_cpuid_entry {
694	__u32 function;
695	__u32 eax;
696	__u32 ebx;
697	__u32 ecx;
698	__u32 edx;
699	__u32 padding;
700  };
701
702  /* for KVM_SET_CPUID */
703  struct kvm_cpuid {
704	__u32 nent;
705	__u32 padding;
706	struct kvm_cpuid_entry entries[0];
707  };
708
709
7104.21 KVM_SET_SIGNAL_MASK
711------------------------
712
713:Capability: basic
714:Architectures: all
715:Type: vcpu ioctl
716:Parameters: struct kvm_signal_mask (in)
717:Returns: 0 on success, -1 on error
718
719Defines which signals are blocked during execution of KVM_RUN.  This
720signal mask temporarily overrides the threads signal mask.  Any
721unblocked signal received (except SIGKILL and SIGSTOP, which retain
722their traditional behaviour) will cause KVM_RUN to return with -EINTR.
723
724Note the signal will only be delivered if not blocked by the original
725signal mask.
726
727::
728
729  /* for KVM_SET_SIGNAL_MASK */
730  struct kvm_signal_mask {
731	__u32 len;
732	__u8  sigset[0];
733  };
734
735
7364.22 KVM_GET_FPU
737----------------
738
739:Capability: basic
740:Architectures: x86
741:Type: vcpu ioctl
742:Parameters: struct kvm_fpu (out)
743:Returns: 0 on success, -1 on error
744
745Reads the floating point state from the vcpu.
746
747::
748
749  /* for KVM_GET_FPU and KVM_SET_FPU */
750  struct kvm_fpu {
751	__u8  fpr[8][16];
752	__u16 fcw;
753	__u16 fsw;
754	__u8  ftwx;  /* in fxsave format */
755	__u8  pad1;
756	__u16 last_opcode;
757	__u64 last_ip;
758	__u64 last_dp;
759	__u8  xmm[16][16];
760	__u32 mxcsr;
761	__u32 pad2;
762  };
763
764
7654.23 KVM_SET_FPU
766----------------
767
768:Capability: basic
769:Architectures: x86
770:Type: vcpu ioctl
771:Parameters: struct kvm_fpu (in)
772:Returns: 0 on success, -1 on error
773
774Writes the floating point state to the vcpu.
775
776::
777
778  /* for KVM_GET_FPU and KVM_SET_FPU */
779  struct kvm_fpu {
780	__u8  fpr[8][16];
781	__u16 fcw;
782	__u16 fsw;
783	__u8  ftwx;  /* in fxsave format */
784	__u8  pad1;
785	__u16 last_opcode;
786	__u64 last_ip;
787	__u64 last_dp;
788	__u8  xmm[16][16];
789	__u32 mxcsr;
790	__u32 pad2;
791  };
792
793
7944.24 KVM_CREATE_IRQCHIP
795-----------------------
796
797:Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
798:Architectures: x86, arm64, s390
799:Type: vm ioctl
800:Parameters: none
801:Returns: 0 on success, -1 on error
802
803Creates an interrupt controller model in the kernel.
804On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
805future vcpus to have a local APIC.  IRQ routing for GSIs 0-15 is set to both
806PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
807On arm64, a GICv2 is created. Any other GIC versions require the usage of
808KVM_CREATE_DEVICE, which also supports creating a GICv2.  Using
809KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
810On s390, a dummy irq routing table is created.
811
812Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
813before KVM_CREATE_IRQCHIP can be used.
814
815
8164.25 KVM_IRQ_LINE
817-----------------
818
819:Capability: KVM_CAP_IRQCHIP
820:Architectures: x86, arm64
821:Type: vm ioctl
822:Parameters: struct kvm_irq_level
823:Returns: 0 on success, -1 on error
824
825Sets the level of a GSI input to the interrupt controller model in the kernel.
826On some architectures it is required that an interrupt controller model has
827been previously created with KVM_CREATE_IRQCHIP.  Note that edge-triggered
828interrupts require the level to be set to 1 and then back to 0.
829
830On real hardware, interrupt pins can be active-low or active-high.  This
831does not matter for the level field of struct kvm_irq_level: 1 always
832means active (asserted), 0 means inactive (deasserted).
833
834x86 allows the operating system to program the interrupt polarity
835(active-low/active-high) for level-triggered interrupts, and KVM used
836to consider the polarity.  However, due to bitrot in the handling of
837active-low interrupts, the above convention is now valid on x86 too.
838This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED.  Userspace
839should not present interrupts to the guest as active-low unless this
840capability is present (or unless it is not using the in-kernel irqchip,
841of course).
842
843
844arm64 can signal an interrupt either at the CPU level, or at the
845in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
846use PPIs designated for specific cpus.  The irq field is interpreted
847like this::
848
849  bits:  |  31 ... 28  | 27 ... 24 | 23  ... 16 | 15 ... 0 |
850  field: | vcpu2_index | irq_type  | vcpu_index |  irq_id  |
851
852The irq_type field has the following values:
853
854- irq_type[0]:
855	       out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
856- irq_type[1]:
857	       in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
858               (the vcpu_index field is ignored)
859- irq_type[2]:
860	       in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
861
862(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
863
864In both cases, level is used to assert/deassert the line.
865
866When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
867identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
868must be zero.
869
870Note that on arm64, the KVM_CAP_IRQCHIP capability only conditions
871injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
872be used for a userspace interrupt controller.
873
874::
875
876  struct kvm_irq_level {
877	union {
878		__u32 irq;     /* GSI */
879		__s32 status;  /* not used for KVM_IRQ_LEVEL */
880	};
881	__u32 level;           /* 0 or 1 */
882  };
883
884
8854.26 KVM_GET_IRQCHIP
886--------------------
887
888:Capability: KVM_CAP_IRQCHIP
889:Architectures: x86
890:Type: vm ioctl
891:Parameters: struct kvm_irqchip (in/out)
892:Returns: 0 on success, -1 on error
893
894Reads the state of a kernel interrupt controller created with
895KVM_CREATE_IRQCHIP into a buffer provided by the caller.
896
897::
898
899  struct kvm_irqchip {
900	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
901	__u32 pad;
902        union {
903		char dummy[512];  /* reserving space */
904		struct kvm_pic_state pic;
905		struct kvm_ioapic_state ioapic;
906	} chip;
907  };
908
909
9104.27 KVM_SET_IRQCHIP
911--------------------
912
913:Capability: KVM_CAP_IRQCHIP
914:Architectures: x86
915:Type: vm ioctl
916:Parameters: struct kvm_irqchip (in)
917:Returns: 0 on success, -1 on error
918
919Sets the state of a kernel interrupt controller created with
920KVM_CREATE_IRQCHIP from a buffer provided by the caller.
921
922::
923
924  struct kvm_irqchip {
925	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
926	__u32 pad;
927        union {
928		char dummy[512];  /* reserving space */
929		struct kvm_pic_state pic;
930		struct kvm_ioapic_state ioapic;
931	} chip;
932  };
933
934
9354.28 KVM_XEN_HVM_CONFIG
936-----------------------
937
938:Capability: KVM_CAP_XEN_HVM
939:Architectures: x86
940:Type: vm ioctl
941:Parameters: struct kvm_xen_hvm_config (in)
942:Returns: 0 on success, -1 on error
943
944Sets the MSR that the Xen HVM guest uses to initialize its hypercall
945page, and provides the starting address and size of the hypercall
946blobs in userspace.  When the guest writes the MSR, kvm copies one
947page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
948memory.
949
950::
951
952  struct kvm_xen_hvm_config {
953	__u32 flags;
954	__u32 msr;
955	__u64 blob_addr_32;
956	__u64 blob_addr_64;
957	__u8 blob_size_32;
958	__u8 blob_size_64;
959	__u8 pad2[30];
960  };
961
962If certain flags are returned from the KVM_CAP_XEN_HVM check, they may
963be set in the flags field of this ioctl:
964
965The KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL flag requests KVM to generate
966the contents of the hypercall page automatically; hypercalls will be
967intercepted and passed to userspace through KVM_EXIT_XEN.  In this
968ase, all of the blob size and address fields must be zero.
969
970The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates to KVM that userspace
971will always use the KVM_XEN_HVM_EVTCHN_SEND ioctl to deliver event
972channel interrupts rather than manipulating the guest's shared_info
973structures directly. This, in turn, may allow KVM to enable features
974such as intercepting the SCHEDOP_poll hypercall to accelerate PV
975spinlock operation for the guest. Userspace may still use the ioctl
976to deliver events if it was advertised, even if userspace does not
977send this indication that it will always do so
978
979No other flags are currently valid in the struct kvm_xen_hvm_config.
980
9814.29 KVM_GET_CLOCK
982------------------
983
984:Capability: KVM_CAP_ADJUST_CLOCK
985:Architectures: x86
986:Type: vm ioctl
987:Parameters: struct kvm_clock_data (out)
988:Returns: 0 on success, -1 on error
989
990Gets the current timestamp of kvmclock as seen by the current guest. In
991conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
992such as migration.
993
994When KVM_CAP_ADJUST_CLOCK is passed to KVM_CHECK_EXTENSION, it returns the
995set of bits that KVM can return in struct kvm_clock_data's flag member.
996
997The following flags are defined:
998
999KVM_CLOCK_TSC_STABLE
1000  If set, the returned value is the exact kvmclock
1001  value seen by all VCPUs at the instant when KVM_GET_CLOCK was called.
1002  If clear, the returned value is simply CLOCK_MONOTONIC plus a constant
1003  offset; the offset can be modified with KVM_SET_CLOCK.  KVM will try
1004  to make all VCPUs follow this clock, but the exact value read by each
1005  VCPU could differ, because the host TSC is not stable.
1006
1007KVM_CLOCK_REALTIME
1008  If set, the `realtime` field in the kvm_clock_data
1009  structure is populated with the value of the host's real time
1010  clocksource at the instant when KVM_GET_CLOCK was called. If clear,
1011  the `realtime` field does not contain a value.
1012
1013KVM_CLOCK_HOST_TSC
1014  If set, the `host_tsc` field in the kvm_clock_data
1015  structure is populated with the value of the host's timestamp counter (TSC)
1016  at the instant when KVM_GET_CLOCK was called. If clear, the `host_tsc` field
1017  does not contain a value.
1018
1019::
1020
1021  struct kvm_clock_data {
1022	__u64 clock;  /* kvmclock current value */
1023	__u32 flags;
1024	__u32 pad0;
1025	__u64 realtime;
1026	__u64 host_tsc;
1027	__u32 pad[4];
1028  };
1029
1030
10314.30 KVM_SET_CLOCK
1032------------------
1033
1034:Capability: KVM_CAP_ADJUST_CLOCK
1035:Architectures: x86
1036:Type: vm ioctl
1037:Parameters: struct kvm_clock_data (in)
1038:Returns: 0 on success, -1 on error
1039
1040Sets the current timestamp of kvmclock to the value specified in its parameter.
1041In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
1042such as migration.
1043
1044The following flags can be passed:
1045
1046KVM_CLOCK_REALTIME
1047  If set, KVM will compare the value of the `realtime` field
1048  with the value of the host's real time clocksource at the instant when
1049  KVM_SET_CLOCK was called. The difference in elapsed time is added to the final
1050  kvmclock value that will be provided to guests.
1051
1052Other flags returned by ``KVM_GET_CLOCK`` are accepted but ignored.
1053
1054::
1055
1056  struct kvm_clock_data {
1057	__u64 clock;  /* kvmclock current value */
1058	__u32 flags;
1059	__u32 pad0;
1060	__u64 realtime;
1061	__u64 host_tsc;
1062	__u32 pad[4];
1063  };
1064
1065
10664.31 KVM_GET_VCPU_EVENTS
1067------------------------
1068
1069:Capability: KVM_CAP_VCPU_EVENTS
1070:Extended by: KVM_CAP_INTR_SHADOW
1071:Architectures: x86, arm64
1072:Type: vcpu ioctl
1073:Parameters: struct kvm_vcpu_event (out)
1074:Returns: 0 on success, -1 on error
1075
1076X86:
1077^^^^
1078
1079Gets currently pending exceptions, interrupts, and NMIs as well as related
1080states of the vcpu.
1081
1082::
1083
1084  struct kvm_vcpu_events {
1085	struct {
1086		__u8 injected;
1087		__u8 nr;
1088		__u8 has_error_code;
1089		__u8 pending;
1090		__u32 error_code;
1091	} exception;
1092	struct {
1093		__u8 injected;
1094		__u8 nr;
1095		__u8 soft;
1096		__u8 shadow;
1097	} interrupt;
1098	struct {
1099		__u8 injected;
1100		__u8 pending;
1101		__u8 masked;
1102		__u8 pad;
1103	} nmi;
1104	__u32 sipi_vector;
1105	__u32 flags;
1106	struct {
1107		__u8 smm;
1108		__u8 pending;
1109		__u8 smm_inside_nmi;
1110		__u8 latched_init;
1111	} smi;
1112	__u8 reserved[27];
1113	__u8 exception_has_payload;
1114	__u64 exception_payload;
1115  };
1116
1117The following bits are defined in the flags field:
1118
1119- KVM_VCPUEVENT_VALID_SHADOW may be set to signal that
1120  interrupt.shadow contains a valid state.
1121
1122- KVM_VCPUEVENT_VALID_SMM may be set to signal that smi contains a
1123  valid state.
1124
1125- KVM_VCPUEVENT_VALID_PAYLOAD may be set to signal that the
1126  exception_has_payload, exception_payload, and exception.pending
1127  fields contain a valid state. This bit will be set whenever
1128  KVM_CAP_EXCEPTION_PAYLOAD is enabled.
1129
1130- KVM_VCPUEVENT_VALID_TRIPLE_FAULT may be set to signal that the
1131  triple_fault_pending field contains a valid state. This bit will
1132  be set whenever KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled.
1133
1134ARM64:
1135^^^^^^
1136
1137If the guest accesses a device that is being emulated by the host kernel in
1138such a way that a real device would generate a physical SError, KVM may make
1139a virtual SError pending for that VCPU. This system error interrupt remains
1140pending until the guest takes the exception by unmasking PSTATE.A.
1141
1142Running the VCPU may cause it to take a pending SError, or make an access that
1143causes an SError to become pending. The event's description is only valid while
1144the VPCU is not running.
1145
1146This API provides a way to read and write the pending 'event' state that is not
1147visible to the guest. To save, restore or migrate a VCPU the struct representing
1148the state can be read then written using this GET/SET API, along with the other
1149guest-visible registers. It is not possible to 'cancel' an SError that has been
1150made pending.
1151
1152A device being emulated in user-space may also wish to generate an SError. To do
1153this the events structure can be populated by user-space. The current state
1154should be read first, to ensure no existing SError is pending. If an existing
1155SError is pending, the architecture's 'Multiple SError interrupts' rules should
1156be followed. (2.5.3 of DDI0587.a "ARM Reliability, Availability, and
1157Serviceability (RAS) Specification").
1158
1159SError exceptions always have an ESR value. Some CPUs have the ability to
1160specify what the virtual SError's ESR value should be. These systems will
1161advertise KVM_CAP_ARM_INJECT_SERROR_ESR. In this case exception.has_esr will
1162always have a non-zero value when read, and the agent making an SError pending
1163should specify the ISS field in the lower 24 bits of exception.serror_esr. If
1164the system supports KVM_CAP_ARM_INJECT_SERROR_ESR, but user-space sets the events
1165with exception.has_esr as zero, KVM will choose an ESR.
1166
1167Specifying exception.has_esr on a system that does not support it will return
1168-EINVAL. Setting anything other than the lower 24bits of exception.serror_esr
1169will return -EINVAL.
1170
1171It is not possible to read back a pending external abort (injected via
1172KVM_SET_VCPU_EVENTS or otherwise) because such an exception is always delivered
1173directly to the virtual CPU).
1174
1175::
1176
1177  struct kvm_vcpu_events {
1178	struct {
1179		__u8 serror_pending;
1180		__u8 serror_has_esr;
1181		__u8 ext_dabt_pending;
1182		/* Align it to 8 bytes */
1183		__u8 pad[5];
1184		__u64 serror_esr;
1185	} exception;
1186	__u32 reserved[12];
1187  };
1188
11894.32 KVM_SET_VCPU_EVENTS
1190------------------------
1191
1192:Capability: KVM_CAP_VCPU_EVENTS
1193:Extended by: KVM_CAP_INTR_SHADOW
1194:Architectures: x86, arm64
1195:Type: vcpu ioctl
1196:Parameters: struct kvm_vcpu_event (in)
1197:Returns: 0 on success, -1 on error
1198
1199X86:
1200^^^^
1201
1202Set pending exceptions, interrupts, and NMIs as well as related states of the
1203vcpu.
1204
1205See KVM_GET_VCPU_EVENTS for the data structure.
1206
1207Fields that may be modified asynchronously by running VCPUs can be excluded
1208from the update. These fields are nmi.pending, sipi_vector, smi.smm,
1209smi.pending. Keep the corresponding bits in the flags field cleared to
1210suppress overwriting the current in-kernel state. The bits are:
1211
1212===============================  ==================================
1213KVM_VCPUEVENT_VALID_NMI_PENDING  transfer nmi.pending to the kernel
1214KVM_VCPUEVENT_VALID_SIPI_VECTOR  transfer sipi_vector
1215KVM_VCPUEVENT_VALID_SMM          transfer the smi sub-struct.
1216===============================  ==================================
1217
1218If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
1219the flags field to signal that interrupt.shadow contains a valid state and
1220shall be written into the VCPU.
1221
1222KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
1223
1224If KVM_CAP_EXCEPTION_PAYLOAD is enabled, KVM_VCPUEVENT_VALID_PAYLOAD
1225can be set in the flags field to signal that the
1226exception_has_payload, exception_payload, and exception.pending fields
1227contain a valid state and shall be written into the VCPU.
1228
1229If KVM_CAP_X86_TRIPLE_FAULT_EVENT is enabled, KVM_VCPUEVENT_VALID_TRIPLE_FAULT
1230can be set in flags field to signal that the triple_fault field contains
1231a valid state and shall be written into the VCPU.
1232
1233ARM64:
1234^^^^^^
1235
1236User space may need to inject several types of events to the guest.
1237
1238Set the pending SError exception state for this VCPU. It is not possible to
1239'cancel' an Serror that has been made pending.
1240
1241If the guest performed an access to I/O memory which could not be handled by
1242userspace, for example because of missing instruction syndrome decode
1243information or because there is no device mapped at the accessed IPA, then
1244userspace can ask the kernel to inject an external abort using the address
1245from the exiting fault on the VCPU. It is a programming error to set
1246ext_dabt_pending after an exit which was not either KVM_EXIT_MMIO or
1247KVM_EXIT_ARM_NISV. This feature is only available if the system supports
1248KVM_CAP_ARM_INJECT_EXT_DABT. This is a helper which provides commonality in
1249how userspace reports accesses for the above cases to guests, across different
1250userspace implementations. Nevertheless, userspace can still emulate all Arm
1251exceptions by manipulating individual registers using the KVM_SET_ONE_REG API.
1252
1253See KVM_GET_VCPU_EVENTS for the data structure.
1254
1255
12564.33 KVM_GET_DEBUGREGS
1257----------------------
1258
1259:Capability: KVM_CAP_DEBUGREGS
1260:Architectures: x86
1261:Type: vm ioctl
1262:Parameters: struct kvm_debugregs (out)
1263:Returns: 0 on success, -1 on error
1264
1265Reads debug registers from the vcpu.
1266
1267::
1268
1269  struct kvm_debugregs {
1270	__u64 db[4];
1271	__u64 dr6;
1272	__u64 dr7;
1273	__u64 flags;
1274	__u64 reserved[9];
1275  };
1276
1277
12784.34 KVM_SET_DEBUGREGS
1279----------------------
1280
1281:Capability: KVM_CAP_DEBUGREGS
1282:Architectures: x86
1283:Type: vm ioctl
1284:Parameters: struct kvm_debugregs (in)
1285:Returns: 0 on success, -1 on error
1286
1287Writes debug registers into the vcpu.
1288
1289See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
1290yet and must be cleared on entry.
1291
1292
12934.35 KVM_SET_USER_MEMORY_REGION
1294-------------------------------
1295
1296:Capability: KVM_CAP_USER_MEMORY
1297:Architectures: all
1298:Type: vm ioctl
1299:Parameters: struct kvm_userspace_memory_region (in)
1300:Returns: 0 on success, -1 on error
1301
1302::
1303
1304  struct kvm_userspace_memory_region {
1305	__u32 slot;
1306	__u32 flags;
1307	__u64 guest_phys_addr;
1308	__u64 memory_size; /* bytes */
1309	__u64 userspace_addr; /* start of the userspace allocated memory */
1310  };
1311
1312  /* for kvm_userspace_memory_region::flags */
1313  #define KVM_MEM_LOG_DIRTY_PAGES	(1UL << 0)
1314  #define KVM_MEM_READONLY	(1UL << 1)
1315
1316This ioctl allows the user to create, modify or delete a guest physical
1317memory slot.  Bits 0-15 of "slot" specify the slot id and this value
1318should be less than the maximum number of user memory slots supported per
1319VM.  The maximum allowed slots can be queried using KVM_CAP_NR_MEMSLOTS.
1320Slots may not overlap in guest physical address space.
1321
1322If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
1323specifies the address space which is being modified.  They must be
1324less than the value that KVM_CHECK_EXTENSION returns for the
1325KVM_CAP_MULTI_ADDRESS_SPACE capability.  Slots in separate address spaces
1326are unrelated; the restriction on overlapping slots only applies within
1327each address space.
1328
1329Deleting a slot is done by passing zero for memory_size.  When changing
1330an existing slot, it may be moved in the guest physical memory space,
1331or its flags may be modified, but it may not be resized.
1332
1333Memory for the region is taken starting at the address denoted by the
1334field userspace_addr, which must point at user addressable memory for
1335the entire memory slot size.  Any object may back this memory, including
1336anonymous memory, ordinary files, and hugetlbfs.
1337
1338On architectures that support a form of address tagging, userspace_addr must
1339be an untagged address.
1340
1341It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
1342be identical.  This allows large pages in the guest to be backed by large
1343pages in the host.
1344
1345The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
1346KVM_MEM_READONLY.  The former can be set to instruct KVM to keep track of
1347writes to memory within the slot.  See KVM_GET_DIRTY_LOG ioctl to know how to
1348use it.  The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
1349to make a new slot read-only.  In this case, writes to this memory will be
1350posted to userspace as KVM_EXIT_MMIO exits.
1351
1352When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
1353the memory region are automatically reflected into the guest.  For example, an
1354mmap() that affects the region will be made visible immediately.  Another
1355example is madvise(MADV_DROP).
1356
1357Note: On arm64, a write generated by the page-table walker (to update
1358the Access and Dirty flags, for example) never results in a
1359KVM_EXIT_MMIO exit when the slot has the KVM_MEM_READONLY flag. This
1360is because KVM cannot provide the data that would be written by the
1361page-table walker, making it impossible to emulate the access.
1362Instead, an abort (data abort if the cause of the page-table update
1363was a load or a store, instruction abort if it was an instruction
1364fetch) is injected in the guest.
1365
13664.36 KVM_SET_TSS_ADDR
1367---------------------
1368
1369:Capability: KVM_CAP_SET_TSS_ADDR
1370:Architectures: x86
1371:Type: vm ioctl
1372:Parameters: unsigned long tss_address (in)
1373:Returns: 0 on success, -1 on error
1374
1375This ioctl defines the physical address of a three-page region in the guest
1376physical address space.  The region must be within the first 4GB of the
1377guest physical address space and must not conflict with any memory slot
1378or any mmio address.  The guest may malfunction if it accesses this memory
1379region.
1380
1381This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1382because of a quirk in the virtualization implementation (see the internals
1383documentation when it pops into existence).
1384
1385
13864.37 KVM_ENABLE_CAP
1387-------------------
1388
1389:Capability: KVM_CAP_ENABLE_CAP
1390:Architectures: mips, ppc, s390, x86
1391:Type: vcpu ioctl
1392:Parameters: struct kvm_enable_cap (in)
1393:Returns: 0 on success; -1 on error
1394
1395:Capability: KVM_CAP_ENABLE_CAP_VM
1396:Architectures: all
1397:Type: vm ioctl
1398:Parameters: struct kvm_enable_cap (in)
1399:Returns: 0 on success; -1 on error
1400
1401.. note::
1402
1403   Not all extensions are enabled by default. Using this ioctl the application
1404   can enable an extension, making it available to the guest.
1405
1406On systems that do not support this ioctl, it always fails. On systems that
1407do support it, it only works for extensions that are supported for enablement.
1408
1409To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1410be used.
1411
1412::
1413
1414  struct kvm_enable_cap {
1415       /* in */
1416       __u32 cap;
1417
1418The capability that is supposed to get enabled.
1419
1420::
1421
1422       __u32 flags;
1423
1424A bitfield indicating future enhancements. Has to be 0 for now.
1425
1426::
1427
1428       __u64 args[4];
1429
1430Arguments for enabling a feature. If a feature needs initial values to
1431function properly, this is the place to put them.
1432
1433::
1434
1435       __u8  pad[64];
1436  };
1437
1438The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1439for vm-wide capabilities.
1440
14414.38 KVM_GET_MP_STATE
1442---------------------
1443
1444:Capability: KVM_CAP_MP_STATE
1445:Architectures: x86, s390, arm64, riscv
1446:Type: vcpu ioctl
1447:Parameters: struct kvm_mp_state (out)
1448:Returns: 0 on success; -1 on error
1449
1450::
1451
1452  struct kvm_mp_state {
1453	__u32 mp_state;
1454  };
1455
1456Returns the vcpu's current "multiprocessing state" (though also valid on
1457uniprocessor guests).
1458
1459Possible values are:
1460
1461   ==========================    ===============================================
1462   KVM_MP_STATE_RUNNABLE         the vcpu is currently running
1463                                 [x86,arm64,riscv]
1464   KVM_MP_STATE_UNINITIALIZED    the vcpu is an application processor (AP)
1465                                 which has not yet received an INIT signal [x86]
1466   KVM_MP_STATE_INIT_RECEIVED    the vcpu has received an INIT signal, and is
1467                                 now ready for a SIPI [x86]
1468   KVM_MP_STATE_HALTED           the vcpu has executed a HLT instruction and
1469                                 is waiting for an interrupt [x86]
1470   KVM_MP_STATE_SIPI_RECEIVED    the vcpu has just received a SIPI (vector
1471                                 accessible via KVM_GET_VCPU_EVENTS) [x86]
1472   KVM_MP_STATE_STOPPED          the vcpu is stopped [s390,arm64,riscv]
1473   KVM_MP_STATE_CHECK_STOP       the vcpu is in a special error state [s390]
1474   KVM_MP_STATE_OPERATING        the vcpu is operating (running or halted)
1475                                 [s390]
1476   KVM_MP_STATE_LOAD             the vcpu is in a special load/startup state
1477                                 [s390]
1478   KVM_MP_STATE_SUSPENDED        the vcpu is in a suspend state and is waiting
1479                                 for a wakeup event [arm64]
1480   ==========================    ===============================================
1481
1482On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1483in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1484these architectures.
1485
1486For arm64:
1487^^^^^^^^^^
1488
1489If a vCPU is in the KVM_MP_STATE_SUSPENDED state, KVM will emulate the
1490architectural execution of a WFI instruction.
1491
1492If a wakeup event is recognized, KVM will exit to userspace with a
1493KVM_SYSTEM_EVENT exit, where the event type is KVM_SYSTEM_EVENT_WAKEUP. If
1494userspace wants to honor the wakeup, it must set the vCPU's MP state to
1495KVM_MP_STATE_RUNNABLE. If it does not, KVM will continue to await a wakeup
1496event in subsequent calls to KVM_RUN.
1497
1498.. warning::
1499
1500     If userspace intends to keep the vCPU in a SUSPENDED state, it is
1501     strongly recommended that userspace take action to suppress the
1502     wakeup event (such as masking an interrupt). Otherwise, subsequent
1503     calls to KVM_RUN will immediately exit with a KVM_SYSTEM_EVENT_WAKEUP
1504     event and inadvertently waste CPU cycles.
1505
1506     Additionally, if userspace takes action to suppress a wakeup event,
1507     it is strongly recommended that it also restores the vCPU to its
1508     original state when the vCPU is made RUNNABLE again. For example,
1509     if userspace masked a pending interrupt to suppress the wakeup,
1510     the interrupt should be unmasked before returning control to the
1511     guest.
1512
1513For riscv:
1514^^^^^^^^^^
1515
1516The only states that are valid are KVM_MP_STATE_STOPPED and
1517KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1518
15194.39 KVM_SET_MP_STATE
1520---------------------
1521
1522:Capability: KVM_CAP_MP_STATE
1523:Architectures: x86, s390, arm64, riscv
1524:Type: vcpu ioctl
1525:Parameters: struct kvm_mp_state (in)
1526:Returns: 0 on success; -1 on error
1527
1528Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1529arguments.
1530
1531On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1532in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1533these architectures.
1534
1535For arm64/riscv:
1536^^^^^^^^^^^^^^^^
1537
1538The only states that are valid are KVM_MP_STATE_STOPPED and
1539KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1540
15414.40 KVM_SET_IDENTITY_MAP_ADDR
1542------------------------------
1543
1544:Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1545:Architectures: x86
1546:Type: vm ioctl
1547:Parameters: unsigned long identity (in)
1548:Returns: 0 on success, -1 on error
1549
1550This ioctl defines the physical address of a one-page region in the guest
1551physical address space.  The region must be within the first 4GB of the
1552guest physical address space and must not conflict with any memory slot
1553or any mmio address.  The guest may malfunction if it accesses this memory
1554region.
1555
1556Setting the address to 0 will result in resetting the address to its default
1557(0xfffbc000).
1558
1559This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1560because of a quirk in the virtualization implementation (see the internals
1561documentation when it pops into existence).
1562
1563Fails if any VCPU has already been created.
1564
15654.41 KVM_SET_BOOT_CPU_ID
1566------------------------
1567
1568:Capability: KVM_CAP_SET_BOOT_CPU_ID
1569:Architectures: x86
1570:Type: vm ioctl
1571:Parameters: unsigned long vcpu_id
1572:Returns: 0 on success, -1 on error
1573
1574Define which vcpu is the Bootstrap Processor (BSP).  Values are the same
1575as the vcpu id in KVM_CREATE_VCPU.  If this ioctl is not called, the default
1576is vcpu 0. This ioctl has to be called before vcpu creation,
1577otherwise it will return EBUSY error.
1578
1579
15804.42 KVM_GET_XSAVE
1581------------------
1582
1583:Capability: KVM_CAP_XSAVE
1584:Architectures: x86
1585:Type: vcpu ioctl
1586:Parameters: struct kvm_xsave (out)
1587:Returns: 0 on success, -1 on error
1588
1589
1590::
1591
1592  struct kvm_xsave {
1593	__u32 region[1024];
1594	__u32 extra[0];
1595  };
1596
1597This ioctl would copy current vcpu's xsave struct to the userspace.
1598
1599
16004.43 KVM_SET_XSAVE
1601------------------
1602
1603:Capability: KVM_CAP_XSAVE and KVM_CAP_XSAVE2
1604:Architectures: x86
1605:Type: vcpu ioctl
1606:Parameters: struct kvm_xsave (in)
1607:Returns: 0 on success, -1 on error
1608
1609::
1610
1611
1612  struct kvm_xsave {
1613	__u32 region[1024];
1614	__u32 extra[0];
1615  };
1616
1617This ioctl would copy userspace's xsave struct to the kernel. It copies
1618as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2),
1619when invoked on the vm file descriptor. The size value returned by
1620KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
1621Currently, it is only greater than 4096 if a dynamic feature has been
1622enabled with ``arch_prctl()``, but this may change in the future.
1623
1624The offsets of the state save areas in struct kvm_xsave follow the
1625contents of CPUID leaf 0xD on the host.
1626
1627
16284.44 KVM_GET_XCRS
1629-----------------
1630
1631:Capability: KVM_CAP_XCRS
1632:Architectures: x86
1633:Type: vcpu ioctl
1634:Parameters: struct kvm_xcrs (out)
1635:Returns: 0 on success, -1 on error
1636
1637::
1638
1639  struct kvm_xcr {
1640	__u32 xcr;
1641	__u32 reserved;
1642	__u64 value;
1643  };
1644
1645  struct kvm_xcrs {
1646	__u32 nr_xcrs;
1647	__u32 flags;
1648	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1649	__u64 padding[16];
1650  };
1651
1652This ioctl would copy current vcpu's xcrs to the userspace.
1653
1654
16554.45 KVM_SET_XCRS
1656-----------------
1657
1658:Capability: KVM_CAP_XCRS
1659:Architectures: x86
1660:Type: vcpu ioctl
1661:Parameters: struct kvm_xcrs (in)
1662:Returns: 0 on success, -1 on error
1663
1664::
1665
1666  struct kvm_xcr {
1667	__u32 xcr;
1668	__u32 reserved;
1669	__u64 value;
1670  };
1671
1672  struct kvm_xcrs {
1673	__u32 nr_xcrs;
1674	__u32 flags;
1675	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1676	__u64 padding[16];
1677  };
1678
1679This ioctl would set vcpu's xcr to the value userspace specified.
1680
1681
16824.46 KVM_GET_SUPPORTED_CPUID
1683----------------------------
1684
1685:Capability: KVM_CAP_EXT_CPUID
1686:Architectures: x86
1687:Type: system ioctl
1688:Parameters: struct kvm_cpuid2 (in/out)
1689:Returns: 0 on success, -1 on error
1690
1691::
1692
1693  struct kvm_cpuid2 {
1694	__u32 nent;
1695	__u32 padding;
1696	struct kvm_cpuid_entry2 entries[0];
1697  };
1698
1699  #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
1700  #define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1) /* deprecated */
1701  #define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2) /* deprecated */
1702
1703  struct kvm_cpuid_entry2 {
1704	__u32 function;
1705	__u32 index;
1706	__u32 flags;
1707	__u32 eax;
1708	__u32 ebx;
1709	__u32 ecx;
1710	__u32 edx;
1711	__u32 padding[3];
1712  };
1713
1714This ioctl returns x86 cpuid features which are supported by both the
1715hardware and kvm in its default configuration.  Userspace can use the
1716information returned by this ioctl to construct cpuid information (for
1717KVM_SET_CPUID2) that is consistent with hardware, kernel, and
1718userspace capabilities, and with user requirements (for example, the
1719user may wish to constrain cpuid to emulate older hardware, or for
1720feature consistency across a cluster).
1721
1722Dynamically-enabled feature bits need to be requested with
1723``arch_prctl()`` before calling this ioctl. Feature bits that have not
1724been requested are excluded from the result.
1725
1726Note that certain capabilities, such as KVM_CAP_X86_DISABLE_EXITS, may
1727expose cpuid features (e.g. MONITOR) which are not supported by kvm in
1728its default configuration. If userspace enables such capabilities, it
1729is responsible for modifying the results of this ioctl appropriately.
1730
1731Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1732with the 'nent' field indicating the number of entries in the variable-size
1733array 'entries'.  If the number of entries is too low to describe the cpu
1734capabilities, an error (E2BIG) is returned.  If the number is too high,
1735the 'nent' field is adjusted and an error (ENOMEM) is returned.  If the
1736number is just right, the 'nent' field is adjusted to the number of valid
1737entries in the 'entries' array, which is then filled.
1738
1739The entries returned are the host cpuid as returned by the cpuid instruction,
1740with unknown or unsupported features masked out.  Some features (for example,
1741x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1742emulate them efficiently. The fields in each entry are defined as follows:
1743
1744  function:
1745         the eax value used to obtain the entry
1746
1747  index:
1748         the ecx value used to obtain the entry (for entries that are
1749         affected by ecx)
1750
1751  flags:
1752     an OR of zero or more of the following:
1753
1754        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1755           if the index field is valid
1756
1757   eax, ebx, ecx, edx:
1758         the values returned by the cpuid instruction for
1759         this function/index combination
1760
1761The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1762as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1763support.  Instead it is reported via::
1764
1765  ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1766
1767if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1768feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1769
1770
17714.47 KVM_PPC_GET_PVINFO
1772-----------------------
1773
1774:Capability: KVM_CAP_PPC_GET_PVINFO
1775:Architectures: ppc
1776:Type: vm ioctl
1777:Parameters: struct kvm_ppc_pvinfo (out)
1778:Returns: 0 on success, !0 on error
1779
1780::
1781
1782  struct kvm_ppc_pvinfo {
1783	__u32 flags;
1784	__u32 hcall[4];
1785	__u8  pad[108];
1786  };
1787
1788This ioctl fetches PV specific information that need to be passed to the guest
1789using the device tree or other means from vm context.
1790
1791The hcall array defines 4 instructions that make up a hypercall.
1792
1793If any additional field gets added to this structure later on, a bit for that
1794additional piece of information will be set in the flags bitmap.
1795
1796The flags bitmap is defined as::
1797
1798   /* the host supports the ePAPR idle hcall
1799   #define KVM_PPC_PVINFO_FLAGS_EV_IDLE   (1<<0)
1800
18014.52 KVM_SET_GSI_ROUTING
1802------------------------
1803
1804:Capability: KVM_CAP_IRQ_ROUTING
1805:Architectures: x86 s390 arm64
1806:Type: vm ioctl
1807:Parameters: struct kvm_irq_routing (in)
1808:Returns: 0 on success, -1 on error
1809
1810Sets the GSI routing table entries, overwriting any previously set entries.
1811
1812On arm64, GSI routing has the following limitation:
1813
1814- GSI routing does not apply to KVM_IRQ_LINE but only to KVM_IRQFD.
1815
1816::
1817
1818  struct kvm_irq_routing {
1819	__u32 nr;
1820	__u32 flags;
1821	struct kvm_irq_routing_entry entries[0];
1822  };
1823
1824No flags are specified so far, the corresponding field must be set to zero.
1825
1826::
1827
1828  struct kvm_irq_routing_entry {
1829	__u32 gsi;
1830	__u32 type;
1831	__u32 flags;
1832	__u32 pad;
1833	union {
1834		struct kvm_irq_routing_irqchip irqchip;
1835		struct kvm_irq_routing_msi msi;
1836		struct kvm_irq_routing_s390_adapter adapter;
1837		struct kvm_irq_routing_hv_sint hv_sint;
1838		struct kvm_irq_routing_xen_evtchn xen_evtchn;
1839		__u32 pad[8];
1840	} u;
1841  };
1842
1843  /* gsi routing entry types */
1844  #define KVM_IRQ_ROUTING_IRQCHIP 1
1845  #define KVM_IRQ_ROUTING_MSI 2
1846  #define KVM_IRQ_ROUTING_S390_ADAPTER 3
1847  #define KVM_IRQ_ROUTING_HV_SINT 4
1848  #define KVM_IRQ_ROUTING_XEN_EVTCHN 5
1849
1850flags:
1851
1852- KVM_MSI_VALID_DEVID: used along with KVM_IRQ_ROUTING_MSI routing entry
1853  type, specifies that the devid field contains a valid value.  The per-VM
1854  KVM_CAP_MSI_DEVID capability advertises the requirement to provide
1855  the device ID.  If this capability is not available, userspace should
1856  never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
1857- zero otherwise
1858
1859::
1860
1861  struct kvm_irq_routing_irqchip {
1862	__u32 irqchip;
1863	__u32 pin;
1864  };
1865
1866  struct kvm_irq_routing_msi {
1867	__u32 address_lo;
1868	__u32 address_hi;
1869	__u32 data;
1870	union {
1871		__u32 pad;
1872		__u32 devid;
1873	};
1874  };
1875
1876If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
1877for the device that wrote the MSI message.  For PCI, this is usually a
1878BFD identifier in the lower 16 bits.
1879
1880On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
1881feature of KVM_CAP_X2APIC_API capability is enabled.  If it is enabled,
1882address_hi bits 31-8 provide bits 31-8 of the destination id.  Bits 7-0 of
1883address_hi must be zero.
1884
1885::
1886
1887  struct kvm_irq_routing_s390_adapter {
1888	__u64 ind_addr;
1889	__u64 summary_addr;
1890	__u64 ind_offset;
1891	__u32 summary_offset;
1892	__u32 adapter_id;
1893  };
1894
1895  struct kvm_irq_routing_hv_sint {
1896	__u32 vcpu;
1897	__u32 sint;
1898  };
1899
1900  struct kvm_irq_routing_xen_evtchn {
1901	__u32 port;
1902	__u32 vcpu;
1903	__u32 priority;
1904  };
1905
1906
1907When KVM_CAP_XEN_HVM includes the KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL bit
1908in its indication of supported features, routing to Xen event channels
1909is supported. Although the priority field is present, only the value
1910KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL is supported, which means delivery by
19112 level event channels. FIFO event channel support may be added in
1912the future.
1913
1914
19154.55 KVM_SET_TSC_KHZ
1916--------------------
1917
1918:Capability: KVM_CAP_TSC_CONTROL / KVM_CAP_VM_TSC_CONTROL
1919:Architectures: x86
1920:Type: vcpu ioctl / vm ioctl
1921:Parameters: virtual tsc_khz
1922:Returns: 0 on success, -1 on error
1923
1924Specifies the tsc frequency for the virtual machine. The unit of the
1925frequency is KHz.
1926
1927If the KVM_CAP_VM_TSC_CONTROL capability is advertised, this can also
1928be used as a vm ioctl to set the initial tsc frequency of subsequently
1929created vCPUs.
1930
19314.56 KVM_GET_TSC_KHZ
1932--------------------
1933
1934:Capability: KVM_CAP_GET_TSC_KHZ / KVM_CAP_VM_TSC_CONTROL
1935:Architectures: x86
1936:Type: vcpu ioctl / vm ioctl
1937:Parameters: none
1938:Returns: virtual tsc-khz on success, negative value on error
1939
1940Returns the tsc frequency of the guest. The unit of the return value is
1941KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1942error.
1943
1944
19454.57 KVM_GET_LAPIC
1946------------------
1947
1948:Capability: KVM_CAP_IRQCHIP
1949:Architectures: x86
1950:Type: vcpu ioctl
1951:Parameters: struct kvm_lapic_state (out)
1952:Returns: 0 on success, -1 on error
1953
1954::
1955
1956  #define KVM_APIC_REG_SIZE 0x400
1957  struct kvm_lapic_state {
1958	char regs[KVM_APIC_REG_SIZE];
1959  };
1960
1961Reads the Local APIC registers and copies them into the input argument.  The
1962data format and layout are the same as documented in the architecture manual.
1963
1964If KVM_X2APIC_API_USE_32BIT_IDS feature of KVM_CAP_X2APIC_API is
1965enabled, then the format of APIC_ID register depends on the APIC mode
1966(reported by MSR_IA32_APICBASE) of its VCPU.  x2APIC stores APIC ID in
1967the APIC_ID register (bytes 32-35).  xAPIC only allows an 8-bit APIC ID
1968which is stored in bits 31-24 of the APIC register, or equivalently in
1969byte 35 of struct kvm_lapic_state's regs field.  KVM_GET_LAPIC must then
1970be called after MSR_IA32_APICBASE has been set with KVM_SET_MSR.
1971
1972If KVM_X2APIC_API_USE_32BIT_IDS feature is disabled, struct kvm_lapic_state
1973always uses xAPIC format.
1974
1975
19764.58 KVM_SET_LAPIC
1977------------------
1978
1979:Capability: KVM_CAP_IRQCHIP
1980:Architectures: x86
1981:Type: vcpu ioctl
1982:Parameters: struct kvm_lapic_state (in)
1983:Returns: 0 on success, -1 on error
1984
1985::
1986
1987  #define KVM_APIC_REG_SIZE 0x400
1988  struct kvm_lapic_state {
1989	char regs[KVM_APIC_REG_SIZE];
1990  };
1991
1992Copies the input argument into the Local APIC registers.  The data format
1993and layout are the same as documented in the architecture manual.
1994
1995The format of the APIC ID register (bytes 32-35 of struct kvm_lapic_state's
1996regs field) depends on the state of the KVM_CAP_X2APIC_API capability.
1997See the note in KVM_GET_LAPIC.
1998
1999
20004.59 KVM_IOEVENTFD
2001------------------
2002
2003:Capability: KVM_CAP_IOEVENTFD
2004:Architectures: all
2005:Type: vm ioctl
2006:Parameters: struct kvm_ioeventfd (in)
2007:Returns: 0 on success, !0 on error
2008
2009This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
2010within the guest.  A guest write in the registered address will signal the
2011provided event instead of triggering an exit.
2012
2013::
2014
2015  struct kvm_ioeventfd {
2016	__u64 datamatch;
2017	__u64 addr;        /* legal pio/mmio address */
2018	__u32 len;         /* 0, 1, 2, 4, or 8 bytes    */
2019	__s32 fd;
2020	__u32 flags;
2021	__u8  pad[36];
2022  };
2023
2024For the special case of virtio-ccw devices on s390, the ioevent is matched
2025to a subchannel/virtqueue tuple instead.
2026
2027The following flags are defined::
2028
2029  #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
2030  #define KVM_IOEVENTFD_FLAG_PIO       (1 << kvm_ioeventfd_flag_nr_pio)
2031  #define KVM_IOEVENTFD_FLAG_DEASSIGN  (1 << kvm_ioeventfd_flag_nr_deassign)
2032  #define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
2033	(1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
2034
2035If datamatch flag is set, the event will be signaled only if the written value
2036to the registered address is equal to datamatch in struct kvm_ioeventfd.
2037
2038For virtio-ccw devices, addr contains the subchannel id and datamatch the
2039virtqueue index.
2040
2041With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
2042the kernel will ignore the length of guest write and may get a faster vmexit.
2043The speedup may only apply to specific architectures, but the ioeventfd will
2044work anyway.
2045
20464.60 KVM_DIRTY_TLB
2047------------------
2048
2049:Capability: KVM_CAP_SW_TLB
2050:Architectures: ppc
2051:Type: vcpu ioctl
2052:Parameters: struct kvm_dirty_tlb (in)
2053:Returns: 0 on success, -1 on error
2054
2055::
2056
2057  struct kvm_dirty_tlb {
2058	__u64 bitmap;
2059	__u32 num_dirty;
2060  };
2061
2062This must be called whenever userspace has changed an entry in the shared
2063TLB, prior to calling KVM_RUN on the associated vcpu.
2064
2065The "bitmap" field is the userspace address of an array.  This array
2066consists of a number of bits, equal to the total number of TLB entries as
2067determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
2068nearest multiple of 64.
2069
2070Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
2071array.
2072
2073The array is little-endian: the bit 0 is the least significant bit of the
2074first byte, bit 8 is the least significant bit of the second byte, etc.
2075This avoids any complications with differing word sizes.
2076
2077The "num_dirty" field is a performance hint for KVM to determine whether it
2078should skip processing the bitmap and just invalidate everything.  It must
2079be set to the number of set bits in the bitmap.
2080
2081
20824.62 KVM_CREATE_SPAPR_TCE
2083-------------------------
2084
2085:Capability: KVM_CAP_SPAPR_TCE
2086:Architectures: powerpc
2087:Type: vm ioctl
2088:Parameters: struct kvm_create_spapr_tce (in)
2089:Returns: file descriptor for manipulating the created TCE table
2090
2091This creates a virtual TCE (translation control entry) table, which
2092is an IOMMU for PAPR-style virtual I/O.  It is used to translate
2093logical addresses used in virtual I/O into guest physical addresses,
2094and provides a scatter/gather capability for PAPR virtual I/O.
2095
2096::
2097
2098  /* for KVM_CAP_SPAPR_TCE */
2099  struct kvm_create_spapr_tce {
2100	__u64 liobn;
2101	__u32 window_size;
2102  };
2103
2104The liobn field gives the logical IO bus number for which to create a
2105TCE table.  The window_size field specifies the size of the DMA window
2106which this TCE table will translate - the table will contain one 64
2107bit TCE entry for every 4kiB of the DMA window.
2108
2109When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
2110table has been created using this ioctl(), the kernel will handle it
2111in real mode, updating the TCE table.  H_PUT_TCE calls for other
2112liobns will cause a vm exit and must be handled by userspace.
2113
2114The return value is a file descriptor which can be passed to mmap(2)
2115to map the created TCE table into userspace.  This lets userspace read
2116the entries written by kernel-handled H_PUT_TCE calls, and also lets
2117userspace update the TCE table directly which is useful in some
2118circumstances.
2119
2120
21214.63 KVM_ALLOCATE_RMA
2122---------------------
2123
2124:Capability: KVM_CAP_PPC_RMA
2125:Architectures: powerpc
2126:Type: vm ioctl
2127:Parameters: struct kvm_allocate_rma (out)
2128:Returns: file descriptor for mapping the allocated RMA
2129
2130This allocates a Real Mode Area (RMA) from the pool allocated at boot
2131time by the kernel.  An RMA is a physically-contiguous, aligned region
2132of memory used on older POWER processors to provide the memory which
2133will be accessed by real-mode (MMU off) accesses in a KVM guest.
2134POWER processors support a set of sizes for the RMA that usually
2135includes 64MB, 128MB, 256MB and some larger powers of two.
2136
2137::
2138
2139  /* for KVM_ALLOCATE_RMA */
2140  struct kvm_allocate_rma {
2141	__u64 rma_size;
2142  };
2143
2144The return value is a file descriptor which can be passed to mmap(2)
2145to map the allocated RMA into userspace.  The mapped area can then be
2146passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
2147RMA for a virtual machine.  The size of the RMA in bytes (which is
2148fixed at host kernel boot time) is returned in the rma_size field of
2149the argument structure.
2150
2151The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
2152is supported; 2 if the processor requires all virtual machines to have
2153an RMA, or 1 if the processor can use an RMA but doesn't require it,
2154because it supports the Virtual RMA (VRMA) facility.
2155
2156
21574.64 KVM_NMI
2158------------
2159
2160:Capability: KVM_CAP_USER_NMI
2161:Architectures: x86
2162:Type: vcpu ioctl
2163:Parameters: none
2164:Returns: 0 on success, -1 on error
2165
2166Queues an NMI on the thread's vcpu.  Note this is well defined only
2167when KVM_CREATE_IRQCHIP has not been called, since this is an interface
2168between the virtual cpu core and virtual local APIC.  After KVM_CREATE_IRQCHIP
2169has been called, this interface is completely emulated within the kernel.
2170
2171To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
2172following algorithm:
2173
2174  - pause the vcpu
2175  - read the local APIC's state (KVM_GET_LAPIC)
2176  - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
2177  - if so, issue KVM_NMI
2178  - resume the vcpu
2179
2180Some guests configure the LINT1 NMI input to cause a panic, aiding in
2181debugging.
2182
2183
21844.65 KVM_S390_UCAS_MAP
2185----------------------
2186
2187:Capability: KVM_CAP_S390_UCONTROL
2188:Architectures: s390
2189:Type: vcpu ioctl
2190:Parameters: struct kvm_s390_ucas_mapping (in)
2191:Returns: 0 in case of success
2192
2193The parameter is defined like this::
2194
2195	struct kvm_s390_ucas_mapping {
2196		__u64 user_addr;
2197		__u64 vcpu_addr;
2198		__u64 length;
2199	};
2200
2201This ioctl maps the memory at "user_addr" with the length "length" to
2202the vcpu's address space starting at "vcpu_addr". All parameters need to
2203be aligned by 1 megabyte.
2204
2205
22064.66 KVM_S390_UCAS_UNMAP
2207------------------------
2208
2209:Capability: KVM_CAP_S390_UCONTROL
2210:Architectures: s390
2211:Type: vcpu ioctl
2212:Parameters: struct kvm_s390_ucas_mapping (in)
2213:Returns: 0 in case of success
2214
2215The parameter is defined like this::
2216
2217	struct kvm_s390_ucas_mapping {
2218		__u64 user_addr;
2219		__u64 vcpu_addr;
2220		__u64 length;
2221	};
2222
2223This ioctl unmaps the memory in the vcpu's address space starting at
2224"vcpu_addr" with the length "length". The field "user_addr" is ignored.
2225All parameters need to be aligned by 1 megabyte.
2226
2227
22284.67 KVM_S390_VCPU_FAULT
2229------------------------
2230
2231:Capability: KVM_CAP_S390_UCONTROL
2232:Architectures: s390
2233:Type: vcpu ioctl
2234:Parameters: vcpu absolute address (in)
2235:Returns: 0 in case of success
2236
2237This call creates a page table entry on the virtual cpu's address space
2238(for user controlled virtual machines) or the virtual machine's address
2239space (for regular virtual machines). This only works for minor faults,
2240thus it's recommended to access subject memory page via the user page
2241table upfront. This is useful to handle validity intercepts for user
2242controlled virtual machines to fault in the virtual cpu's lowcore pages
2243prior to calling the KVM_RUN ioctl.
2244
2245
22464.68 KVM_SET_ONE_REG
2247--------------------
2248
2249:Capability: KVM_CAP_ONE_REG
2250:Architectures: all
2251:Type: vcpu ioctl
2252:Parameters: struct kvm_one_reg (in)
2253:Returns: 0 on success, negative value on failure
2254
2255Errors:
2256
2257  ======   ============================================================
2258  ENOENT   no such register
2259  EINVAL   invalid register ID, or no such register or used with VMs in
2260           protected virtualization mode on s390
2261  EPERM    (arm64) register access not allowed before vcpu finalization
2262  ======   ============================================================
2263
2264(These error codes are indicative only: do not rely on a specific error
2265code being returned in a specific situation.)
2266
2267::
2268
2269  struct kvm_one_reg {
2270       __u64 id;
2271       __u64 addr;
2272 };
2273
2274Using this ioctl, a single vcpu register can be set to a specific value
2275defined by user space with the passed in struct kvm_one_reg, where id
2276refers to the register identifier as described below and addr is a pointer
2277to a variable with the respective size. There can be architecture agnostic
2278and architecture specific registers. Each have their own range of operation
2279and their own constants and width. To keep track of the implemented
2280registers, find a list below:
2281
2282  ======= =============================== ============
2283  Arch              Register              Width (bits)
2284  ======= =============================== ============
2285  PPC     KVM_REG_PPC_HIOR                64
2286  PPC     KVM_REG_PPC_IAC1                64
2287  PPC     KVM_REG_PPC_IAC2                64
2288  PPC     KVM_REG_PPC_IAC3                64
2289  PPC     KVM_REG_PPC_IAC4                64
2290  PPC     KVM_REG_PPC_DAC1                64
2291  PPC     KVM_REG_PPC_DAC2                64
2292  PPC     KVM_REG_PPC_DABR                64
2293  PPC     KVM_REG_PPC_DSCR                64
2294  PPC     KVM_REG_PPC_PURR                64
2295  PPC     KVM_REG_PPC_SPURR               64
2296  PPC     KVM_REG_PPC_DAR                 64
2297  PPC     KVM_REG_PPC_DSISR               32
2298  PPC     KVM_REG_PPC_AMR                 64
2299  PPC     KVM_REG_PPC_UAMOR               64
2300  PPC     KVM_REG_PPC_MMCR0               64
2301  PPC     KVM_REG_PPC_MMCR1               64
2302  PPC     KVM_REG_PPC_MMCRA               64
2303  PPC     KVM_REG_PPC_MMCR2               64
2304  PPC     KVM_REG_PPC_MMCRS               64
2305  PPC     KVM_REG_PPC_MMCR3               64
2306  PPC     KVM_REG_PPC_SIAR                64
2307  PPC     KVM_REG_PPC_SDAR                64
2308  PPC     KVM_REG_PPC_SIER                64
2309  PPC     KVM_REG_PPC_SIER2               64
2310  PPC     KVM_REG_PPC_SIER3               64
2311  PPC     KVM_REG_PPC_PMC1                32
2312  PPC     KVM_REG_PPC_PMC2                32
2313  PPC     KVM_REG_PPC_PMC3                32
2314  PPC     KVM_REG_PPC_PMC4                32
2315  PPC     KVM_REG_PPC_PMC5                32
2316  PPC     KVM_REG_PPC_PMC6                32
2317  PPC     KVM_REG_PPC_PMC7                32
2318  PPC     KVM_REG_PPC_PMC8                32
2319  PPC     KVM_REG_PPC_FPR0                64
2320  ...
2321  PPC     KVM_REG_PPC_FPR31               64
2322  PPC     KVM_REG_PPC_VR0                 128
2323  ...
2324  PPC     KVM_REG_PPC_VR31                128
2325  PPC     KVM_REG_PPC_VSR0                128
2326  ...
2327  PPC     KVM_REG_PPC_VSR31               128
2328  PPC     KVM_REG_PPC_FPSCR               64
2329  PPC     KVM_REG_PPC_VSCR                32
2330  PPC     KVM_REG_PPC_VPA_ADDR            64
2331  PPC     KVM_REG_PPC_VPA_SLB             128
2332  PPC     KVM_REG_PPC_VPA_DTL             128
2333  PPC     KVM_REG_PPC_EPCR                32
2334  PPC     KVM_REG_PPC_EPR                 32
2335  PPC     KVM_REG_PPC_TCR                 32
2336  PPC     KVM_REG_PPC_TSR                 32
2337  PPC     KVM_REG_PPC_OR_TSR              32
2338  PPC     KVM_REG_PPC_CLEAR_TSR           32
2339  PPC     KVM_REG_PPC_MAS0                32
2340  PPC     KVM_REG_PPC_MAS1                32
2341  PPC     KVM_REG_PPC_MAS2                64
2342  PPC     KVM_REG_PPC_MAS7_3              64
2343  PPC     KVM_REG_PPC_MAS4                32
2344  PPC     KVM_REG_PPC_MAS6                32
2345  PPC     KVM_REG_PPC_MMUCFG              32
2346  PPC     KVM_REG_PPC_TLB0CFG             32
2347  PPC     KVM_REG_PPC_TLB1CFG             32
2348  PPC     KVM_REG_PPC_TLB2CFG             32
2349  PPC     KVM_REG_PPC_TLB3CFG             32
2350  PPC     KVM_REG_PPC_TLB0PS              32
2351  PPC     KVM_REG_PPC_TLB1PS              32
2352  PPC     KVM_REG_PPC_TLB2PS              32
2353  PPC     KVM_REG_PPC_TLB3PS              32
2354  PPC     KVM_REG_PPC_EPTCFG              32
2355  PPC     KVM_REG_PPC_ICP_STATE           64
2356  PPC     KVM_REG_PPC_VP_STATE            128
2357  PPC     KVM_REG_PPC_TB_OFFSET           64
2358  PPC     KVM_REG_PPC_SPMC1               32
2359  PPC     KVM_REG_PPC_SPMC2               32
2360  PPC     KVM_REG_PPC_IAMR                64
2361  PPC     KVM_REG_PPC_TFHAR               64
2362  PPC     KVM_REG_PPC_TFIAR               64
2363  PPC     KVM_REG_PPC_TEXASR              64
2364  PPC     KVM_REG_PPC_FSCR                64
2365  PPC     KVM_REG_PPC_PSPB                32
2366  PPC     KVM_REG_PPC_EBBHR               64
2367  PPC     KVM_REG_PPC_EBBRR               64
2368  PPC     KVM_REG_PPC_BESCR               64
2369  PPC     KVM_REG_PPC_TAR                 64
2370  PPC     KVM_REG_PPC_DPDES               64
2371  PPC     KVM_REG_PPC_DAWR                64
2372  PPC     KVM_REG_PPC_DAWRX               64
2373  PPC     KVM_REG_PPC_CIABR               64
2374  PPC     KVM_REG_PPC_IC                  64
2375  PPC     KVM_REG_PPC_VTB                 64
2376  PPC     KVM_REG_PPC_CSIGR               64
2377  PPC     KVM_REG_PPC_TACR                64
2378  PPC     KVM_REG_PPC_TCSCR               64
2379  PPC     KVM_REG_PPC_PID                 64
2380  PPC     KVM_REG_PPC_ACOP                64
2381  PPC     KVM_REG_PPC_VRSAVE              32
2382  PPC     KVM_REG_PPC_LPCR                32
2383  PPC     KVM_REG_PPC_LPCR_64             64
2384  PPC     KVM_REG_PPC_PPR                 64
2385  PPC     KVM_REG_PPC_ARCH_COMPAT         32
2386  PPC     KVM_REG_PPC_DABRX               32
2387  PPC     KVM_REG_PPC_WORT                64
2388  PPC	  KVM_REG_PPC_SPRG9               64
2389  PPC	  KVM_REG_PPC_DBSR                32
2390  PPC     KVM_REG_PPC_TIDR                64
2391  PPC     KVM_REG_PPC_PSSCR               64
2392  PPC     KVM_REG_PPC_DEC_EXPIRY          64
2393  PPC     KVM_REG_PPC_PTCR                64
2394  PPC     KVM_REG_PPC_DAWR1               64
2395  PPC     KVM_REG_PPC_DAWRX1              64
2396  PPC     KVM_REG_PPC_TM_GPR0             64
2397  ...
2398  PPC     KVM_REG_PPC_TM_GPR31            64
2399  PPC     KVM_REG_PPC_TM_VSR0             128
2400  ...
2401  PPC     KVM_REG_PPC_TM_VSR63            128
2402  PPC     KVM_REG_PPC_TM_CR               64
2403  PPC     KVM_REG_PPC_TM_LR               64
2404  PPC     KVM_REG_PPC_TM_CTR              64
2405  PPC     KVM_REG_PPC_TM_FPSCR            64
2406  PPC     KVM_REG_PPC_TM_AMR              64
2407  PPC     KVM_REG_PPC_TM_PPR              64
2408  PPC     KVM_REG_PPC_TM_VRSAVE           64
2409  PPC     KVM_REG_PPC_TM_VSCR             32
2410  PPC     KVM_REG_PPC_TM_DSCR             64
2411  PPC     KVM_REG_PPC_TM_TAR              64
2412  PPC     KVM_REG_PPC_TM_XER              64
2413
2414  MIPS    KVM_REG_MIPS_R0                 64
2415  ...
2416  MIPS    KVM_REG_MIPS_R31                64
2417  MIPS    KVM_REG_MIPS_HI                 64
2418  MIPS    KVM_REG_MIPS_LO                 64
2419  MIPS    KVM_REG_MIPS_PC                 64
2420  MIPS    KVM_REG_MIPS_CP0_INDEX          32
2421  MIPS    KVM_REG_MIPS_CP0_ENTRYLO0       64
2422  MIPS    KVM_REG_MIPS_CP0_ENTRYLO1       64
2423  MIPS    KVM_REG_MIPS_CP0_CONTEXT        64
2424  MIPS    KVM_REG_MIPS_CP0_CONTEXTCONFIG  32
2425  MIPS    KVM_REG_MIPS_CP0_USERLOCAL      64
2426  MIPS    KVM_REG_MIPS_CP0_XCONTEXTCONFIG 64
2427  MIPS    KVM_REG_MIPS_CP0_PAGEMASK       32
2428  MIPS    KVM_REG_MIPS_CP0_PAGEGRAIN      32
2429  MIPS    KVM_REG_MIPS_CP0_SEGCTL0        64
2430  MIPS    KVM_REG_MIPS_CP0_SEGCTL1        64
2431  MIPS    KVM_REG_MIPS_CP0_SEGCTL2        64
2432  MIPS    KVM_REG_MIPS_CP0_PWBASE         64
2433  MIPS    KVM_REG_MIPS_CP0_PWFIELD        64
2434  MIPS    KVM_REG_MIPS_CP0_PWSIZE         64
2435  MIPS    KVM_REG_MIPS_CP0_WIRED          32
2436  MIPS    KVM_REG_MIPS_CP0_PWCTL          32
2437  MIPS    KVM_REG_MIPS_CP0_HWRENA         32
2438  MIPS    KVM_REG_MIPS_CP0_BADVADDR       64
2439  MIPS    KVM_REG_MIPS_CP0_BADINSTR       32
2440  MIPS    KVM_REG_MIPS_CP0_BADINSTRP      32
2441  MIPS    KVM_REG_MIPS_CP0_COUNT          32
2442  MIPS    KVM_REG_MIPS_CP0_ENTRYHI        64
2443  MIPS    KVM_REG_MIPS_CP0_COMPARE        32
2444  MIPS    KVM_REG_MIPS_CP0_STATUS         32
2445  MIPS    KVM_REG_MIPS_CP0_INTCTL         32
2446  MIPS    KVM_REG_MIPS_CP0_CAUSE          32
2447  MIPS    KVM_REG_MIPS_CP0_EPC            64
2448  MIPS    KVM_REG_MIPS_CP0_PRID           32
2449  MIPS    KVM_REG_MIPS_CP0_EBASE          64
2450  MIPS    KVM_REG_MIPS_CP0_CONFIG         32
2451  MIPS    KVM_REG_MIPS_CP0_CONFIG1        32
2452  MIPS    KVM_REG_MIPS_CP0_CONFIG2        32
2453  MIPS    KVM_REG_MIPS_CP0_CONFIG3        32
2454  MIPS    KVM_REG_MIPS_CP0_CONFIG4        32
2455  MIPS    KVM_REG_MIPS_CP0_CONFIG5        32
2456  MIPS    KVM_REG_MIPS_CP0_CONFIG7        32
2457  MIPS    KVM_REG_MIPS_CP0_XCONTEXT       64
2458  MIPS    KVM_REG_MIPS_CP0_ERROREPC       64
2459  MIPS    KVM_REG_MIPS_CP0_KSCRATCH1      64
2460  MIPS    KVM_REG_MIPS_CP0_KSCRATCH2      64
2461  MIPS    KVM_REG_MIPS_CP0_KSCRATCH3      64
2462  MIPS    KVM_REG_MIPS_CP0_KSCRATCH4      64
2463  MIPS    KVM_REG_MIPS_CP0_KSCRATCH5      64
2464  MIPS    KVM_REG_MIPS_CP0_KSCRATCH6      64
2465  MIPS    KVM_REG_MIPS_CP0_MAAR(0..63)    64
2466  MIPS    KVM_REG_MIPS_COUNT_CTL          64
2467  MIPS    KVM_REG_MIPS_COUNT_RESUME       64
2468  MIPS    KVM_REG_MIPS_COUNT_HZ           64
2469  MIPS    KVM_REG_MIPS_FPR_32(0..31)      32
2470  MIPS    KVM_REG_MIPS_FPR_64(0..31)      64
2471  MIPS    KVM_REG_MIPS_VEC_128(0..31)     128
2472  MIPS    KVM_REG_MIPS_FCR_IR             32
2473  MIPS    KVM_REG_MIPS_FCR_CSR            32
2474  MIPS    KVM_REG_MIPS_MSA_IR             32
2475  MIPS    KVM_REG_MIPS_MSA_CSR            32
2476  ======= =============================== ============
2477
2478ARM registers are mapped using the lower 32 bits.  The upper 16 of that
2479is the register group type, or coprocessor number:
2480
2481ARM core registers have the following id bit patterns::
2482
2483  0x4020 0000 0010 <index into the kvm_regs struct:16>
2484
2485ARM 32-bit CP15 registers have the following id bit patterns::
2486
2487  0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2488
2489ARM 64-bit CP15 registers have the following id bit patterns::
2490
2491  0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2492
2493ARM CCSIDR registers are demultiplexed by CSSELR value::
2494
2495  0x4020 0000 0011 00 <csselr:8>
2496
2497ARM 32-bit VFP control registers have the following id bit patterns::
2498
2499  0x4020 0000 0012 1 <regno:12>
2500
2501ARM 64-bit FP registers have the following id bit patterns::
2502
2503  0x4030 0000 0012 0 <regno:12>
2504
2505ARM firmware pseudo-registers have the following bit pattern::
2506
2507  0x4030 0000 0014 <regno:16>
2508
2509
2510arm64 registers are mapped using the lower 32 bits. The upper 16 of
2511that is the register group type, or coprocessor number:
2512
2513arm64 core/FP-SIMD registers have the following id bit patterns. Note
2514that the size of the access is variable, as the kvm_regs structure
2515contains elements ranging from 32 to 128 bits. The index is a 32bit
2516value in the kvm_regs structure seen as a 32bit array::
2517
2518  0x60x0 0000 0010 <index into the kvm_regs struct:16>
2519
2520Specifically:
2521
2522======================= ========= ===== =======================================
2523    Encoding            Register  Bits  kvm_regs member
2524======================= ========= ===== =======================================
2525  0x6030 0000 0010 0000 X0          64  regs.regs[0]
2526  0x6030 0000 0010 0002 X1          64  regs.regs[1]
2527  ...
2528  0x6030 0000 0010 003c X30         64  regs.regs[30]
2529  0x6030 0000 0010 003e SP          64  regs.sp
2530  0x6030 0000 0010 0040 PC          64  regs.pc
2531  0x6030 0000 0010 0042 PSTATE      64  regs.pstate
2532  0x6030 0000 0010 0044 SP_EL1      64  sp_el1
2533  0x6030 0000 0010 0046 ELR_EL1     64  elr_el1
2534  0x6030 0000 0010 0048 SPSR_EL1    64  spsr[KVM_SPSR_EL1] (alias SPSR_SVC)
2535  0x6030 0000 0010 004a SPSR_ABT    64  spsr[KVM_SPSR_ABT]
2536  0x6030 0000 0010 004c SPSR_UND    64  spsr[KVM_SPSR_UND]
2537  0x6030 0000 0010 004e SPSR_IRQ    64  spsr[KVM_SPSR_IRQ]
2538  0x6060 0000 0010 0050 SPSR_FIQ    64  spsr[KVM_SPSR_FIQ]
2539  0x6040 0000 0010 0054 V0         128  fp_regs.vregs[0]    [1]_
2540  0x6040 0000 0010 0058 V1         128  fp_regs.vregs[1]    [1]_
2541  ...
2542  0x6040 0000 0010 00d0 V31        128  fp_regs.vregs[31]   [1]_
2543  0x6020 0000 0010 00d4 FPSR        32  fp_regs.fpsr
2544  0x6020 0000 0010 00d5 FPCR        32  fp_regs.fpcr
2545======================= ========= ===== =======================================
2546
2547.. [1] These encodings are not accepted for SVE-enabled vcpus.  See
2548       KVM_ARM_VCPU_INIT.
2549
2550       The equivalent register content can be accessed via bits [127:0] of
2551       the corresponding SVE Zn registers instead for vcpus that have SVE
2552       enabled (see below).
2553
2554arm64 CCSIDR registers are demultiplexed by CSSELR value::
2555
2556  0x6020 0000 0011 00 <csselr:8>
2557
2558arm64 system registers have the following id bit patterns::
2559
2560  0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2561
2562.. warning::
2563
2564     Two system register IDs do not follow the specified pattern.  These
2565     are KVM_REG_ARM_TIMER_CVAL and KVM_REG_ARM_TIMER_CNT, which map to
2566     system registers CNTV_CVAL_EL0 and CNTVCT_EL0 respectively.  These
2567     two had their values accidentally swapped, which means TIMER_CVAL is
2568     derived from the register encoding for CNTVCT_EL0 and TIMER_CNT is
2569     derived from the register encoding for CNTV_CVAL_EL0.  As this is
2570     API, it must remain this way.
2571
2572arm64 firmware pseudo-registers have the following bit pattern::
2573
2574  0x6030 0000 0014 <regno:16>
2575
2576arm64 SVE registers have the following bit patterns::
2577
2578  0x6080 0000 0015 00 <n:5> <slice:5>   Zn bits[2048*slice + 2047 : 2048*slice]
2579  0x6050 0000 0015 04 <n:4> <slice:5>   Pn bits[256*slice + 255 : 256*slice]
2580  0x6050 0000 0015 060 <slice:5>        FFR bits[256*slice + 255 : 256*slice]
2581  0x6060 0000 0015 ffff                 KVM_REG_ARM64_SVE_VLS pseudo-register
2582
2583Access to register IDs where 2048 * slice >= 128 * max_vq will fail with
2584ENOENT.  max_vq is the vcpu's maximum supported vector length in 128-bit
2585quadwords: see [2]_ below.
2586
2587These registers are only accessible on vcpus for which SVE is enabled.
2588See KVM_ARM_VCPU_INIT for details.
2589
2590In addition, except for KVM_REG_ARM64_SVE_VLS, these registers are not
2591accessible until the vcpu's SVE configuration has been finalized
2592using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).  See KVM_ARM_VCPU_INIT
2593and KVM_ARM_VCPU_FINALIZE for more information about this procedure.
2594
2595KVM_REG_ARM64_SVE_VLS is a pseudo-register that allows the set of vector
2596lengths supported by the vcpu to be discovered and configured by
2597userspace.  When transferred to or from user memory via KVM_GET_ONE_REG
2598or KVM_SET_ONE_REG, the value of this register is of type
2599__u64[KVM_ARM64_SVE_VLS_WORDS], and encodes the set of vector lengths as
2600follows::
2601
2602  __u64 vector_lengths[KVM_ARM64_SVE_VLS_WORDS];
2603
2604  if (vq >= SVE_VQ_MIN && vq <= SVE_VQ_MAX &&
2605      ((vector_lengths[(vq - KVM_ARM64_SVE_VQ_MIN) / 64] >>
2606		((vq - KVM_ARM64_SVE_VQ_MIN) % 64)) & 1))
2607	/* Vector length vq * 16 bytes supported */
2608  else
2609	/* Vector length vq * 16 bytes not supported */
2610
2611.. [2] The maximum value vq for which the above condition is true is
2612       max_vq.  This is the maximum vector length available to the guest on
2613       this vcpu, and determines which register slices are visible through
2614       this ioctl interface.
2615
2616(See Documentation/arm64/sve.rst for an explanation of the "vq"
2617nomenclature.)
2618
2619KVM_REG_ARM64_SVE_VLS is only accessible after KVM_ARM_VCPU_INIT.
2620KVM_ARM_VCPU_INIT initialises it to the best set of vector lengths that
2621the host supports.
2622
2623Userspace may subsequently modify it if desired until the vcpu's SVE
2624configuration is finalized using KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE).
2625
2626Apart from simply removing all vector lengths from the host set that
2627exceed some value, support for arbitrarily chosen sets of vector lengths
2628is hardware-dependent and may not be available.  Attempting to configure
2629an invalid set of vector lengths via KVM_SET_ONE_REG will fail with
2630EINVAL.
2631
2632After the vcpu's SVE configuration is finalized, further attempts to
2633write this register will fail with EPERM.
2634
2635arm64 bitmap feature firmware pseudo-registers have the following bit pattern::
2636
2637  0x6030 0000 0016 <regno:16>
2638
2639The bitmap feature firmware registers exposes the hypercall services that
2640are available for userspace to configure. The set bits corresponds to the
2641services that are available for the guests to access. By default, KVM
2642sets all the supported bits during VM initialization. The userspace can
2643discover the available services via KVM_GET_ONE_REG, and write back the
2644bitmap corresponding to the features that it wishes guests to see via
2645KVM_SET_ONE_REG.
2646
2647Note: These registers are immutable once any of the vCPUs of the VM has
2648run at least once. A KVM_SET_ONE_REG in such a scenario will return
2649a -EBUSY to userspace.
2650
2651(See Documentation/virt/kvm/arm/hypercalls.rst for more details.)
2652
2653
2654MIPS registers are mapped using the lower 32 bits.  The upper 16 of that is
2655the register group type:
2656
2657MIPS core registers (see above) have the following id bit patterns::
2658
2659  0x7030 0000 0000 <reg:16>
2660
2661MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2662patterns depending on whether they're 32-bit or 64-bit registers::
2663
2664  0x7020 0000 0001 00 <reg:5> <sel:3>   (32-bit)
2665  0x7030 0000 0001 00 <reg:5> <sel:3>   (64-bit)
2666
2667Note: KVM_REG_MIPS_CP0_ENTRYLO0 and KVM_REG_MIPS_CP0_ENTRYLO1 are the MIPS64
2668versions of the EntryLo registers regardless of the word size of the host
2669hardware, host kernel, guest, and whether XPA is present in the guest, i.e.
2670with the RI and XI bits (if they exist) in bits 63 and 62 respectively, and
2671the PFNX field starting at bit 30.
2672
2673MIPS MAARs (see KVM_REG_MIPS_CP0_MAAR(*) above) have the following id bit
2674patterns::
2675
2676  0x7030 0000 0001 01 <reg:8>
2677
2678MIPS KVM control registers (see above) have the following id bit patterns::
2679
2680  0x7030 0000 0002 <reg:16>
2681
2682MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2683id bit patterns depending on the size of the register being accessed. They are
2684always accessed according to the current guest FPU mode (Status.FR and
2685Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2686if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2687registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2688overlap the FPU registers::
2689
2690  0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2691  0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2692  0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2693
2694MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2695following id bit patterns::
2696
2697  0x7020 0000 0003 01 <0:3> <reg:5>
2698
2699MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2700following id bit patterns::
2701
2702  0x7020 0000 0003 02 <0:3> <reg:5>
2703
2704RISC-V registers are mapped using the lower 32 bits. The upper 8 bits of
2705that is the register group type.
2706
2707RISC-V config registers are meant for configuring a Guest VCPU and it has
2708the following id bit patterns::
2709
2710  0x8020 0000 01 <index into the kvm_riscv_config struct:24> (32bit Host)
2711  0x8030 0000 01 <index into the kvm_riscv_config struct:24> (64bit Host)
2712
2713Following are the RISC-V config registers:
2714
2715======================= ========= =============================================
2716    Encoding            Register  Description
2717======================= ========= =============================================
2718  0x80x0 0000 0100 0000 isa       ISA feature bitmap of Guest VCPU
2719======================= ========= =============================================
2720
2721The isa config register can be read anytime but can only be written before
2722a Guest VCPU runs. It will have ISA feature bits matching underlying host
2723set by default.
2724
2725RISC-V core registers represent the general excution state of a Guest VCPU
2726and it has the following id bit patterns::
2727
2728  0x8020 0000 02 <index into the kvm_riscv_core struct:24> (32bit Host)
2729  0x8030 0000 02 <index into the kvm_riscv_core struct:24> (64bit Host)
2730
2731Following are the RISC-V core registers:
2732
2733======================= ========= =============================================
2734    Encoding            Register  Description
2735======================= ========= =============================================
2736  0x80x0 0000 0200 0000 regs.pc   Program counter
2737  0x80x0 0000 0200 0001 regs.ra   Return address
2738  0x80x0 0000 0200 0002 regs.sp   Stack pointer
2739  0x80x0 0000 0200 0003 regs.gp   Global pointer
2740  0x80x0 0000 0200 0004 regs.tp   Task pointer
2741  0x80x0 0000 0200 0005 regs.t0   Caller saved register 0
2742  0x80x0 0000 0200 0006 regs.t1   Caller saved register 1
2743  0x80x0 0000 0200 0007 regs.t2   Caller saved register 2
2744  0x80x0 0000 0200 0008 regs.s0   Callee saved register 0
2745  0x80x0 0000 0200 0009 regs.s1   Callee saved register 1
2746  0x80x0 0000 0200 000a regs.a0   Function argument (or return value) 0
2747  0x80x0 0000 0200 000b regs.a1   Function argument (or return value) 1
2748  0x80x0 0000 0200 000c regs.a2   Function argument 2
2749  0x80x0 0000 0200 000d regs.a3   Function argument 3
2750  0x80x0 0000 0200 000e regs.a4   Function argument 4
2751  0x80x0 0000 0200 000f regs.a5   Function argument 5
2752  0x80x0 0000 0200 0010 regs.a6   Function argument 6
2753  0x80x0 0000 0200 0011 regs.a7   Function argument 7
2754  0x80x0 0000 0200 0012 regs.s2   Callee saved register 2
2755  0x80x0 0000 0200 0013 regs.s3   Callee saved register 3
2756  0x80x0 0000 0200 0014 regs.s4   Callee saved register 4
2757  0x80x0 0000 0200 0015 regs.s5   Callee saved register 5
2758  0x80x0 0000 0200 0016 regs.s6   Callee saved register 6
2759  0x80x0 0000 0200 0017 regs.s7   Callee saved register 7
2760  0x80x0 0000 0200 0018 regs.s8   Callee saved register 8
2761  0x80x0 0000 0200 0019 regs.s9   Callee saved register 9
2762  0x80x0 0000 0200 001a regs.s10  Callee saved register 10
2763  0x80x0 0000 0200 001b regs.s11  Callee saved register 11
2764  0x80x0 0000 0200 001c regs.t3   Caller saved register 3
2765  0x80x0 0000 0200 001d regs.t4   Caller saved register 4
2766  0x80x0 0000 0200 001e regs.t5   Caller saved register 5
2767  0x80x0 0000 0200 001f regs.t6   Caller saved register 6
2768  0x80x0 0000 0200 0020 mode      Privilege mode (1 = S-mode or 0 = U-mode)
2769======================= ========= =============================================
2770
2771RISC-V csr registers represent the supervisor mode control/status registers
2772of a Guest VCPU and it has the following id bit patterns::
2773
2774  0x8020 0000 03 <index into the kvm_riscv_csr struct:24> (32bit Host)
2775  0x8030 0000 03 <index into the kvm_riscv_csr struct:24> (64bit Host)
2776
2777Following are the RISC-V csr registers:
2778
2779======================= ========= =============================================
2780    Encoding            Register  Description
2781======================= ========= =============================================
2782  0x80x0 0000 0300 0000 sstatus   Supervisor status
2783  0x80x0 0000 0300 0001 sie       Supervisor interrupt enable
2784  0x80x0 0000 0300 0002 stvec     Supervisor trap vector base
2785  0x80x0 0000 0300 0003 sscratch  Supervisor scratch register
2786  0x80x0 0000 0300 0004 sepc      Supervisor exception program counter
2787  0x80x0 0000 0300 0005 scause    Supervisor trap cause
2788  0x80x0 0000 0300 0006 stval     Supervisor bad address or instruction
2789  0x80x0 0000 0300 0007 sip       Supervisor interrupt pending
2790  0x80x0 0000 0300 0008 satp      Supervisor address translation and protection
2791======================= ========= =============================================
2792
2793RISC-V timer registers represent the timer state of a Guest VCPU and it has
2794the following id bit patterns::
2795
2796  0x8030 0000 04 <index into the kvm_riscv_timer struct:24>
2797
2798Following are the RISC-V timer registers:
2799
2800======================= ========= =============================================
2801    Encoding            Register  Description
2802======================= ========= =============================================
2803  0x8030 0000 0400 0000 frequency Time base frequency (read-only)
2804  0x8030 0000 0400 0001 time      Time value visible to Guest
2805  0x8030 0000 0400 0002 compare   Time compare programmed by Guest
2806  0x8030 0000 0400 0003 state     Time compare state (1 = ON or 0 = OFF)
2807======================= ========= =============================================
2808
2809RISC-V F-extension registers represent the single precision floating point
2810state of a Guest VCPU and it has the following id bit patterns::
2811
2812  0x8020 0000 05 <index into the __riscv_f_ext_state struct:24>
2813
2814Following are the RISC-V F-extension registers:
2815
2816======================= ========= =============================================
2817    Encoding            Register  Description
2818======================= ========= =============================================
2819  0x8020 0000 0500 0000 f[0]      Floating point register 0
2820  ...
2821  0x8020 0000 0500 001f f[31]     Floating point register 31
2822  0x8020 0000 0500 0020 fcsr      Floating point control and status register
2823======================= ========= =============================================
2824
2825RISC-V D-extension registers represent the double precision floating point
2826state of a Guest VCPU and it has the following id bit patterns::
2827
2828  0x8020 0000 06 <index into the __riscv_d_ext_state struct:24> (fcsr)
2829  0x8030 0000 06 <index into the __riscv_d_ext_state struct:24> (non-fcsr)
2830
2831Following are the RISC-V D-extension registers:
2832
2833======================= ========= =============================================
2834    Encoding            Register  Description
2835======================= ========= =============================================
2836  0x8030 0000 0600 0000 f[0]      Floating point register 0
2837  ...
2838  0x8030 0000 0600 001f f[31]     Floating point register 31
2839  0x8020 0000 0600 0020 fcsr      Floating point control and status register
2840======================= ========= =============================================
2841
2842
28434.69 KVM_GET_ONE_REG
2844--------------------
2845
2846:Capability: KVM_CAP_ONE_REG
2847:Architectures: all
2848:Type: vcpu ioctl
2849:Parameters: struct kvm_one_reg (in and out)
2850:Returns: 0 on success, negative value on failure
2851
2852Errors include:
2853
2854  ======== ============================================================
2855  ENOENT   no such register
2856  EINVAL   invalid register ID, or no such register or used with VMs in
2857           protected virtualization mode on s390
2858  EPERM    (arm64) register access not allowed before vcpu finalization
2859  ======== ============================================================
2860
2861(These error codes are indicative only: do not rely on a specific error
2862code being returned in a specific situation.)
2863
2864This ioctl allows to receive the value of a single register implemented
2865in a vcpu. The register to read is indicated by the "id" field of the
2866kvm_one_reg struct passed in. On success, the register value can be found
2867at the memory location pointed to by "addr".
2868
2869The list of registers accessible using this interface is identical to the
2870list in 4.68.
2871
2872
28734.70 KVM_KVMCLOCK_CTRL
2874----------------------
2875
2876:Capability: KVM_CAP_KVMCLOCK_CTRL
2877:Architectures: Any that implement pvclocks (currently x86 only)
2878:Type: vcpu ioctl
2879:Parameters: None
2880:Returns: 0 on success, -1 on error
2881
2882This ioctl sets a flag accessible to the guest indicating that the specified
2883vCPU has been paused by the host userspace.
2884
2885The host will set a flag in the pvclock structure that is checked from the
2886soft lockup watchdog.  The flag is part of the pvclock structure that is
2887shared between guest and host, specifically the second bit of the flags
2888field of the pvclock_vcpu_time_info structure.  It will be set exclusively by
2889the host and read/cleared exclusively by the guest.  The guest operation of
2890checking and clearing the flag must be an atomic operation so
2891load-link/store-conditional, or equivalent must be used.  There are two cases
2892where the guest will clear the flag: when the soft lockup watchdog timer resets
2893itself or when a soft lockup is detected.  This ioctl can be called any time
2894after pausing the vcpu, but before it is resumed.
2895
2896
28974.71 KVM_SIGNAL_MSI
2898-------------------
2899
2900:Capability: KVM_CAP_SIGNAL_MSI
2901:Architectures: x86 arm64
2902:Type: vm ioctl
2903:Parameters: struct kvm_msi (in)
2904:Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2905
2906Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2907MSI messages.
2908
2909::
2910
2911  struct kvm_msi {
2912	__u32 address_lo;
2913	__u32 address_hi;
2914	__u32 data;
2915	__u32 flags;
2916	__u32 devid;
2917	__u8  pad[12];
2918  };
2919
2920flags:
2921  KVM_MSI_VALID_DEVID: devid contains a valid value.  The per-VM
2922  KVM_CAP_MSI_DEVID capability advertises the requirement to provide
2923  the device ID.  If this capability is not available, userspace
2924  should never set the KVM_MSI_VALID_DEVID flag as the ioctl might fail.
2925
2926If KVM_MSI_VALID_DEVID is set, devid contains a unique device identifier
2927for the device that wrote the MSI message.  For PCI, this is usually a
2928BFD identifier in the lower 16 bits.
2929
2930On x86, address_hi is ignored unless the KVM_X2APIC_API_USE_32BIT_IDS
2931feature of KVM_CAP_X2APIC_API capability is enabled.  If it is enabled,
2932address_hi bits 31-8 provide bits 31-8 of the destination id.  Bits 7-0 of
2933address_hi must be zero.
2934
2935
29364.71 KVM_CREATE_PIT2
2937--------------------
2938
2939:Capability: KVM_CAP_PIT2
2940:Architectures: x86
2941:Type: vm ioctl
2942:Parameters: struct kvm_pit_config (in)
2943:Returns: 0 on success, -1 on error
2944
2945Creates an in-kernel device model for the i8254 PIT. This call is only valid
2946after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2947parameters have to be passed::
2948
2949  struct kvm_pit_config {
2950	__u32 flags;
2951	__u32 pad[15];
2952  };
2953
2954Valid flags are::
2955
2956  #define KVM_PIT_SPEAKER_DUMMY     1 /* emulate speaker port stub */
2957
2958PIT timer interrupts may use a per-VM kernel thread for injection. If it
2959exists, this thread will have a name of the following pattern::
2960
2961  kvm-pit/<owner-process-pid>
2962
2963When running a guest with elevated priorities, the scheduling parameters of
2964this thread may have to be adjusted accordingly.
2965
2966This IOCTL replaces the obsolete KVM_CREATE_PIT.
2967
2968
29694.72 KVM_GET_PIT2
2970-----------------
2971
2972:Capability: KVM_CAP_PIT_STATE2
2973:Architectures: x86
2974:Type: vm ioctl
2975:Parameters: struct kvm_pit_state2 (out)
2976:Returns: 0 on success, -1 on error
2977
2978Retrieves the state of the in-kernel PIT model. Only valid after
2979KVM_CREATE_PIT2. The state is returned in the following structure::
2980
2981  struct kvm_pit_state2 {
2982	struct kvm_pit_channel_state channels[3];
2983	__u32 flags;
2984	__u32 reserved[9];
2985  };
2986
2987Valid flags are::
2988
2989  /* disable PIT in HPET legacy mode */
2990  #define KVM_PIT_FLAGS_HPET_LEGACY     0x00000001
2991  /* speaker port data bit enabled */
2992  #define KVM_PIT_FLAGS_SPEAKER_DATA_ON 0x00000002
2993
2994This IOCTL replaces the obsolete KVM_GET_PIT.
2995
2996
29974.73 KVM_SET_PIT2
2998-----------------
2999
3000:Capability: KVM_CAP_PIT_STATE2
3001:Architectures: x86
3002:Type: vm ioctl
3003:Parameters: struct kvm_pit_state2 (in)
3004:Returns: 0 on success, -1 on error
3005
3006Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
3007See KVM_GET_PIT2 for details on struct kvm_pit_state2.
3008
3009This IOCTL replaces the obsolete KVM_SET_PIT.
3010
3011
30124.74 KVM_PPC_GET_SMMU_INFO
3013--------------------------
3014
3015:Capability: KVM_CAP_PPC_GET_SMMU_INFO
3016:Architectures: powerpc
3017:Type: vm ioctl
3018:Parameters: None
3019:Returns: 0 on success, -1 on error
3020
3021This populates and returns a structure describing the features of
3022the "Server" class MMU emulation supported by KVM.
3023This can in turn be used by userspace to generate the appropriate
3024device-tree properties for the guest operating system.
3025
3026The structure contains some global information, followed by an
3027array of supported segment page sizes::
3028
3029      struct kvm_ppc_smmu_info {
3030	     __u64 flags;
3031	     __u32 slb_size;
3032	     __u32 pad;
3033	     struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
3034      };
3035
3036The supported flags are:
3037
3038    - KVM_PPC_PAGE_SIZES_REAL:
3039        When that flag is set, guest page sizes must "fit" the backing
3040        store page sizes. When not set, any page size in the list can
3041        be used regardless of how they are backed by userspace.
3042
3043    - KVM_PPC_1T_SEGMENTS
3044        The emulated MMU supports 1T segments in addition to the
3045        standard 256M ones.
3046
3047    - KVM_PPC_NO_HASH
3048	This flag indicates that HPT guests are not supported by KVM,
3049	thus all guests must use radix MMU mode.
3050
3051The "slb_size" field indicates how many SLB entries are supported
3052
3053The "sps" array contains 8 entries indicating the supported base
3054page sizes for a segment in increasing order. Each entry is defined
3055as follow::
3056
3057   struct kvm_ppc_one_seg_page_size {
3058	__u32 page_shift;	/* Base page shift of segment (or 0) */
3059	__u32 slb_enc;		/* SLB encoding for BookS */
3060	struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
3061   };
3062
3063An entry with a "page_shift" of 0 is unused. Because the array is
3064organized in increasing order, a lookup can stop when encoutering
3065such an entry.
3066
3067The "slb_enc" field provides the encoding to use in the SLB for the
3068page size. The bits are in positions such as the value can directly
3069be OR'ed into the "vsid" argument of the slbmte instruction.
3070
3071The "enc" array is a list which for each of those segment base page
3072size provides the list of supported actual page sizes (which can be
3073only larger or equal to the base page size), along with the
3074corresponding encoding in the hash PTE. Similarly, the array is
30758 entries sorted by increasing sizes and an entry with a "0" shift
3076is an empty entry and a terminator::
3077
3078   struct kvm_ppc_one_page_size {
3079	__u32 page_shift;	/* Page shift (or 0) */
3080	__u32 pte_enc;		/* Encoding in the HPTE (>>12) */
3081   };
3082
3083The "pte_enc" field provides a value that can OR'ed into the hash
3084PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
3085into the hash PTE second double word).
3086
30874.75 KVM_IRQFD
3088--------------
3089
3090:Capability: KVM_CAP_IRQFD
3091:Architectures: x86 s390 arm64
3092:Type: vm ioctl
3093:Parameters: struct kvm_irqfd (in)
3094:Returns: 0 on success, -1 on error
3095
3096Allows setting an eventfd to directly trigger a guest interrupt.
3097kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
3098kvm_irqfd.gsi specifies the irqchip pin toggled by this event.  When
3099an event is triggered on the eventfd, an interrupt is injected into
3100the guest using the specified gsi pin.  The irqfd is removed using
3101the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
3102and kvm_irqfd.gsi.
3103
3104With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
3105mechanism allowing emulation of level-triggered, irqfd-based
3106interrupts.  When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
3107additional eventfd in the kvm_irqfd.resamplefd field.  When operating
3108in resample mode, posting of an interrupt through kvm_irq.fd asserts
3109the specified gsi in the irqchip.  When the irqchip is resampled, such
3110as from an EOI, the gsi is de-asserted and the user is notified via
3111kvm_irqfd.resamplefd.  It is the user's responsibility to re-queue
3112the interrupt if the device making use of it still requires service.
3113Note that closing the resamplefd is not sufficient to disable the
3114irqfd.  The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
3115and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
3116
3117On arm64, gsi routing being supported, the following can happen:
3118
3119- in case no routing entry is associated to this gsi, injection fails
3120- in case the gsi is associated to an irqchip routing entry,
3121  irqchip.pin + 32 corresponds to the injected SPI ID.
3122- in case the gsi is associated to an MSI routing entry, the MSI
3123  message and device ID are translated into an LPI (support restricted
3124  to GICv3 ITS in-kernel emulation).
3125
31264.76 KVM_PPC_ALLOCATE_HTAB
3127--------------------------
3128
3129:Capability: KVM_CAP_PPC_ALLOC_HTAB
3130:Architectures: powerpc
3131:Type: vm ioctl
3132:Parameters: Pointer to u32 containing hash table order (in/out)
3133:Returns: 0 on success, -1 on error
3134
3135This requests the host kernel to allocate an MMU hash table for a
3136guest using the PAPR paravirtualization interface.  This only does
3137anything if the kernel is configured to use the Book 3S HV style of
3138virtualization.  Otherwise the capability doesn't exist and the ioctl
3139returns an ENOTTY error.  The rest of this description assumes Book 3S
3140HV.
3141
3142There must be no vcpus running when this ioctl is called; if there
3143are, it will do nothing and return an EBUSY error.
3144
3145The parameter is a pointer to a 32-bit unsigned integer variable
3146containing the order (log base 2) of the desired size of the hash
3147table, which must be between 18 and 46.  On successful return from the
3148ioctl, the value will not be changed by the kernel.
3149
3150If no hash table has been allocated when any vcpu is asked to run
3151(with the KVM_RUN ioctl), the host kernel will allocate a
3152default-sized hash table (16 MB).
3153
3154If this ioctl is called when a hash table has already been allocated,
3155with a different order from the existing hash table, the existing hash
3156table will be freed and a new one allocated.  If this is ioctl is
3157called when a hash table has already been allocated of the same order
3158as specified, the kernel will clear out the existing hash table (zero
3159all HPTEs).  In either case, if the guest is using the virtualized
3160real-mode area (VRMA) facility, the kernel will re-create the VMRA
3161HPTEs on the next KVM_RUN of any vcpu.
3162
31634.77 KVM_S390_INTERRUPT
3164-----------------------
3165
3166:Capability: basic
3167:Architectures: s390
3168:Type: vm ioctl, vcpu ioctl
3169:Parameters: struct kvm_s390_interrupt (in)
3170:Returns: 0 on success, -1 on error
3171
3172Allows to inject an interrupt to the guest. Interrupts can be floating
3173(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
3174
3175Interrupt parameters are passed via kvm_s390_interrupt::
3176
3177  struct kvm_s390_interrupt {
3178	__u32 type;
3179	__u32 parm;
3180	__u64 parm64;
3181  };
3182
3183type can be one of the following:
3184
3185KVM_S390_SIGP_STOP (vcpu)
3186    - sigp stop; optional flags in parm
3187KVM_S390_PROGRAM_INT (vcpu)
3188    - program check; code in parm
3189KVM_S390_SIGP_SET_PREFIX (vcpu)
3190    - sigp set prefix; prefix address in parm
3191KVM_S390_RESTART (vcpu)
3192    - restart
3193KVM_S390_INT_CLOCK_COMP (vcpu)
3194    - clock comparator interrupt
3195KVM_S390_INT_CPU_TIMER (vcpu)
3196    - CPU timer interrupt
3197KVM_S390_INT_VIRTIO (vm)
3198    - virtio external interrupt; external interrupt
3199      parameters in parm and parm64
3200KVM_S390_INT_SERVICE (vm)
3201    - sclp external interrupt; sclp parameter in parm
3202KVM_S390_INT_EMERGENCY (vcpu)
3203    - sigp emergency; source cpu in parm
3204KVM_S390_INT_EXTERNAL_CALL (vcpu)
3205    - sigp external call; source cpu in parm
3206KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm)
3207    - compound value to indicate an
3208      I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
3209      I/O interruption parameters in parm (subchannel) and parm64 (intparm,
3210      interruption subclass)
3211KVM_S390_MCHK (vm, vcpu)
3212    - machine check interrupt; cr 14 bits in parm, machine check interrupt
3213      code in parm64 (note that machine checks needing further payload are not
3214      supported by this ioctl)
3215
3216This is an asynchronous vcpu ioctl and can be invoked from any thread.
3217
32184.78 KVM_PPC_GET_HTAB_FD
3219------------------------
3220
3221:Capability: KVM_CAP_PPC_HTAB_FD
3222:Architectures: powerpc
3223:Type: vm ioctl
3224:Parameters: Pointer to struct kvm_get_htab_fd (in)
3225:Returns: file descriptor number (>= 0) on success, -1 on error
3226
3227This returns a file descriptor that can be used either to read out the
3228entries in the guest's hashed page table (HPT), or to write entries to
3229initialize the HPT.  The returned fd can only be written to if the
3230KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
3231can only be read if that bit is clear.  The argument struct looks like
3232this::
3233
3234  /* For KVM_PPC_GET_HTAB_FD */
3235  struct kvm_get_htab_fd {
3236	__u64	flags;
3237	__u64	start_index;
3238	__u64	reserved[2];
3239  };
3240
3241  /* Values for kvm_get_htab_fd.flags */
3242  #define KVM_GET_HTAB_BOLTED_ONLY	((__u64)0x1)
3243  #define KVM_GET_HTAB_WRITE		((__u64)0x2)
3244
3245The 'start_index' field gives the index in the HPT of the entry at
3246which to start reading.  It is ignored when writing.
3247
3248Reads on the fd will initially supply information about all
3249"interesting" HPT entries.  Interesting entries are those with the
3250bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
3251all entries.  When the end of the HPT is reached, the read() will
3252return.  If read() is called again on the fd, it will start again from
3253the beginning of the HPT, but will only return HPT entries that have
3254changed since they were last read.
3255
3256Data read or written is structured as a header (8 bytes) followed by a
3257series of valid HPT entries (16 bytes) each.  The header indicates how
3258many valid HPT entries there are and how many invalid entries follow
3259the valid entries.  The invalid entries are not represented explicitly
3260in the stream.  The header format is::
3261
3262  struct kvm_get_htab_header {
3263	__u32	index;
3264	__u16	n_valid;
3265	__u16	n_invalid;
3266  };
3267
3268Writes to the fd create HPT entries starting at the index given in the
3269header; first 'n_valid' valid entries with contents from the data
3270written, then 'n_invalid' invalid entries, invalidating any previously
3271valid entries found.
3272
32734.79 KVM_CREATE_DEVICE
3274----------------------
3275
3276:Capability: KVM_CAP_DEVICE_CTRL
3277:Architectures: all
3278:Type: vm ioctl
3279:Parameters: struct kvm_create_device (in/out)
3280:Returns: 0 on success, -1 on error
3281
3282Errors:
3283
3284  ======  =======================================================
3285  ENODEV  The device type is unknown or unsupported
3286  EEXIST  Device already created, and this type of device may not
3287          be instantiated multiple times
3288  ======  =======================================================
3289
3290  Other error conditions may be defined by individual device types or
3291  have their standard meanings.
3292
3293Creates an emulated device in the kernel.  The file descriptor returned
3294in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
3295
3296If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
3297device type is supported (not necessarily whether it can be created
3298in the current vm).
3299
3300Individual devices should not define flags.  Attributes should be used
3301for specifying any behavior that is not implied by the device type
3302number.
3303
3304::
3305
3306  struct kvm_create_device {
3307	__u32	type;	/* in: KVM_DEV_TYPE_xxx */
3308	__u32	fd;	/* out: device handle */
3309	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
3310  };
3311
33124.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
3313--------------------------------------------
3314
3315:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3316             KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3317             KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device (no set)
3318:Architectures: x86, arm64, s390
3319:Type: device ioctl, vm ioctl, vcpu ioctl
3320:Parameters: struct kvm_device_attr
3321:Returns: 0 on success, -1 on error
3322
3323Errors:
3324
3325  =====   =============================================================
3326  ENXIO   The group or attribute is unknown/unsupported for this device
3327          or hardware support is missing.
3328  EPERM   The attribute cannot (currently) be accessed this way
3329          (e.g. read-only attribute, or attribute that only makes
3330          sense when the device is in a different state)
3331  =====   =============================================================
3332
3333  Other error conditions may be defined by individual device types.
3334
3335Gets/sets a specified piece of device configuration and/or state.  The
3336semantics are device-specific.  See individual device documentation in
3337the "devices" directory.  As with ONE_REG, the size of the data
3338transferred is defined by the particular attribute.
3339
3340::
3341
3342  struct kvm_device_attr {
3343	__u32	flags;		/* no flags currently defined */
3344	__u32	group;		/* device-defined */
3345	__u64	attr;		/* group-defined */
3346	__u64	addr;		/* userspace address of attr data */
3347  };
3348
33494.81 KVM_HAS_DEVICE_ATTR
3350------------------------
3351
3352:Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device,
3353             KVM_CAP_VCPU_ATTRIBUTES for vcpu device
3354             KVM_CAP_SYS_ATTRIBUTES for system (/dev/kvm) device
3355:Type: device ioctl, vm ioctl, vcpu ioctl
3356:Parameters: struct kvm_device_attr
3357:Returns: 0 on success, -1 on error
3358
3359Errors:
3360
3361  =====   =============================================================
3362  ENXIO   The group or attribute is unknown/unsupported for this device
3363          or hardware support is missing.
3364  =====   =============================================================
3365
3366Tests whether a device supports a particular attribute.  A successful
3367return indicates the attribute is implemented.  It does not necessarily
3368indicate that the attribute can be read or written in the device's
3369current state.  "addr" is ignored.
3370
33714.82 KVM_ARM_VCPU_INIT
3372----------------------
3373
3374:Capability: basic
3375:Architectures: arm64
3376:Type: vcpu ioctl
3377:Parameters: struct kvm_vcpu_init (in)
3378:Returns: 0 on success; -1 on error
3379
3380Errors:
3381
3382  ======     =================================================================
3383  EINVAL     the target is unknown, or the combination of features is invalid.
3384  ENOENT     a features bit specified is unknown.
3385  ======     =================================================================
3386
3387This tells KVM what type of CPU to present to the guest, and what
3388optional features it should have.  This will cause a reset of the cpu
3389registers to their initial values.  If this is not called, KVM_RUN will
3390return ENOEXEC for that vcpu.
3391
3392The initial values are defined as:
3393	- Processor state:
3394		* AArch64: EL1h, D, A, I and F bits set. All other bits
3395		  are cleared.
3396		* AArch32: SVC, A, I and F bits set. All other bits are
3397		  cleared.
3398	- General Purpose registers, including PC and SP: set to 0
3399	- FPSIMD/NEON registers: set to 0
3400	- SVE registers: set to 0
3401	- System registers: Reset to their architecturally defined
3402	  values as for a warm reset to EL1 (resp. SVC)
3403
3404Note that because some registers reflect machine topology, all vcpus
3405should be created before this ioctl is invoked.
3406
3407Userspace can call this function multiple times for a given vcpu, including
3408after the vcpu has been run. This will reset the vcpu to its initial
3409state. All calls to this function after the initial call must use the same
3410target and same set of feature flags, otherwise EINVAL will be returned.
3411
3412Possible features:
3413
3414	- KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
3415	  Depends on KVM_CAP_ARM_PSCI.  If not set, the CPU will be powered on
3416	  and execute guest code when KVM_RUN is called.
3417	- KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
3418	  Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
3419	- KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 (or a future revision
3420          backward compatible with v0.2) for the CPU.
3421	  Depends on KVM_CAP_ARM_PSCI_0_2.
3422	- KVM_ARM_VCPU_PMU_V3: Emulate PMUv3 for the CPU.
3423	  Depends on KVM_CAP_ARM_PMU_V3.
3424
3425	- KVM_ARM_VCPU_PTRAUTH_ADDRESS: Enables Address Pointer authentication
3426	  for arm64 only.
3427	  Depends on KVM_CAP_ARM_PTRAUTH_ADDRESS.
3428	  If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3429	  both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3430	  KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3431	  requested.
3432
3433	- KVM_ARM_VCPU_PTRAUTH_GENERIC: Enables Generic Pointer authentication
3434	  for arm64 only.
3435	  Depends on KVM_CAP_ARM_PTRAUTH_GENERIC.
3436	  If KVM_CAP_ARM_PTRAUTH_ADDRESS and KVM_CAP_ARM_PTRAUTH_GENERIC are
3437	  both present, then both KVM_ARM_VCPU_PTRAUTH_ADDRESS and
3438	  KVM_ARM_VCPU_PTRAUTH_GENERIC must be requested or neither must be
3439	  requested.
3440
3441	- KVM_ARM_VCPU_SVE: Enables SVE for the CPU (arm64 only).
3442	  Depends on KVM_CAP_ARM_SVE.
3443	  Requires KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3444
3445	   * After KVM_ARM_VCPU_INIT:
3446
3447	      - KVM_REG_ARM64_SVE_VLS may be read using KVM_GET_ONE_REG: the
3448	        initial value of this pseudo-register indicates the best set of
3449	        vector lengths possible for a vcpu on this host.
3450
3451	   * Before KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3452
3453	      - KVM_RUN and KVM_GET_REG_LIST are not available;
3454
3455	      - KVM_GET_ONE_REG and KVM_SET_ONE_REG cannot be used to access
3456	        the scalable archietctural SVE registers
3457	        KVM_REG_ARM64_SVE_ZREG(), KVM_REG_ARM64_SVE_PREG() or
3458	        KVM_REG_ARM64_SVE_FFR;
3459
3460	      - KVM_REG_ARM64_SVE_VLS may optionally be written using
3461	        KVM_SET_ONE_REG, to modify the set of vector lengths available
3462	        for the vcpu.
3463
3464	   * After KVM_ARM_VCPU_FINALIZE(KVM_ARM_VCPU_SVE):
3465
3466	      - the KVM_REG_ARM64_SVE_VLS pseudo-register is immutable, and can
3467	        no longer be written using KVM_SET_ONE_REG.
3468
34694.83 KVM_ARM_PREFERRED_TARGET
3470-----------------------------
3471
3472:Capability: basic
3473:Architectures: arm64
3474:Type: vm ioctl
3475:Parameters: struct kvm_vcpu_init (out)
3476:Returns: 0 on success; -1 on error
3477
3478Errors:
3479
3480  ======     ==========================================
3481  ENODEV     no preferred target available for the host
3482  ======     ==========================================
3483
3484This queries KVM for preferred CPU target type which can be emulated
3485by KVM on underlying host.
3486
3487The ioctl returns struct kvm_vcpu_init instance containing information
3488about preferred CPU target type and recommended features for it.  The
3489kvm_vcpu_init->features bitmap returned will have feature bits set if
3490the preferred target recommends setting these features, but this is
3491not mandatory.
3492
3493The information returned by this ioctl can be used to prepare an instance
3494of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
3495VCPU matching underlying host.
3496
3497
34984.84 KVM_GET_REG_LIST
3499---------------------
3500
3501:Capability: basic
3502:Architectures: arm64, mips
3503:Type: vcpu ioctl
3504:Parameters: struct kvm_reg_list (in/out)
3505:Returns: 0 on success; -1 on error
3506
3507Errors:
3508
3509  =====      ==============================================================
3510  E2BIG      the reg index list is too big to fit in the array specified by
3511             the user (the number required will be written into n).
3512  =====      ==============================================================
3513
3514::
3515
3516  struct kvm_reg_list {
3517	__u64 n; /* number of registers in reg[] */
3518	__u64 reg[0];
3519  };
3520
3521This ioctl returns the guest registers that are supported for the
3522KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
3523
3524
35254.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
3526-----------------------------------------
3527
3528:Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
3529:Architectures: arm64
3530:Type: vm ioctl
3531:Parameters: struct kvm_arm_device_address (in)
3532:Returns: 0 on success, -1 on error
3533
3534Errors:
3535
3536  ======  ============================================
3537  ENODEV  The device id is unknown
3538  ENXIO   Device not supported on current system
3539  EEXIST  Address already set
3540  E2BIG   Address outside guest physical address space
3541  EBUSY   Address overlaps with other device range
3542  ======  ============================================
3543
3544::
3545
3546  struct kvm_arm_device_addr {
3547	__u64 id;
3548	__u64 addr;
3549  };
3550
3551Specify a device address in the guest's physical address space where guests
3552can access emulated or directly exposed devices, which the host kernel needs
3553to know about. The id field is an architecture specific identifier for a
3554specific device.
3555
3556arm64 divides the id field into two parts, a device id and an
3557address type id specific to the individual device::
3558
3559  bits:  | 63        ...       32 | 31    ...    16 | 15    ...    0 |
3560  field: |        0x00000000      |     device id   |  addr type id  |
3561
3562arm64 currently only require this when using the in-kernel GIC
3563support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
3564as the device id.  When setting the base address for the guest's
3565mapping of the VGIC virtual CPU and distributor interface, the ioctl
3566must be called after calling KVM_CREATE_IRQCHIP, but before calling
3567KVM_RUN on any of the VCPUs.  Calling this ioctl twice for any of the
3568base addresses will return -EEXIST.
3569
3570Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
3571should be used instead.
3572
3573
35744.86 KVM_PPC_RTAS_DEFINE_TOKEN
3575------------------------------
3576
3577:Capability: KVM_CAP_PPC_RTAS
3578:Architectures: ppc
3579:Type: vm ioctl
3580:Parameters: struct kvm_rtas_token_args
3581:Returns: 0 on success, -1 on error
3582
3583Defines a token value for a RTAS (Run Time Abstraction Services)
3584service in order to allow it to be handled in the kernel.  The
3585argument struct gives the name of the service, which must be the name
3586of a service that has a kernel-side implementation.  If the token
3587value is non-zero, it will be associated with that service, and
3588subsequent RTAS calls by the guest specifying that token will be
3589handled by the kernel.  If the token value is 0, then any token
3590associated with the service will be forgotten, and subsequent RTAS
3591calls by the guest for that service will be passed to userspace to be
3592handled.
3593
35944.87 KVM_SET_GUEST_DEBUG
3595------------------------
3596
3597:Capability: KVM_CAP_SET_GUEST_DEBUG
3598:Architectures: x86, s390, ppc, arm64
3599:Type: vcpu ioctl
3600:Parameters: struct kvm_guest_debug (in)
3601:Returns: 0 on success; -1 on error
3602
3603::
3604
3605  struct kvm_guest_debug {
3606       __u32 control;
3607       __u32 pad;
3608       struct kvm_guest_debug_arch arch;
3609  };
3610
3611Set up the processor specific debug registers and configure vcpu for
3612handling guest debug events. There are two parts to the structure, the
3613first a control bitfield indicates the type of debug events to handle
3614when running. Common control bits are:
3615
3616  - KVM_GUESTDBG_ENABLE:        guest debugging is enabled
3617  - KVM_GUESTDBG_SINGLESTEP:    the next run should single-step
3618
3619The top 16 bits of the control field are architecture specific control
3620flags which can include the following:
3621
3622  - KVM_GUESTDBG_USE_SW_BP:     using software breakpoints [x86, arm64]
3623  - KVM_GUESTDBG_USE_HW_BP:     using hardware breakpoints [x86, s390]
3624  - KVM_GUESTDBG_USE_HW:        using hardware debug events [arm64]
3625  - KVM_GUESTDBG_INJECT_DB:     inject DB type exception [x86]
3626  - KVM_GUESTDBG_INJECT_BP:     inject BP type exception [x86]
3627  - KVM_GUESTDBG_EXIT_PENDING:  trigger an immediate guest exit [s390]
3628  - KVM_GUESTDBG_BLOCKIRQ:      avoid injecting interrupts/NMI/SMI [x86]
3629
3630For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
3631are enabled in memory so we need to ensure breakpoint exceptions are
3632correctly trapped and the KVM run loop exits at the breakpoint and not
3633running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
3634we need to ensure the guest vCPUs architecture specific registers are
3635updated to the correct (supplied) values.
3636
3637The second part of the structure is architecture specific and
3638typically contains a set of debug registers.
3639
3640For arm64 the number of debug registers is implementation defined and
3641can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
3642KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
3643indicating the number of supported registers.
3644
3645For ppc, the KVM_CAP_PPC_GUEST_DEBUG_SSTEP capability indicates whether
3646the single-step debug event (KVM_GUESTDBG_SINGLESTEP) is supported.
3647
3648Also when supported, KVM_CAP_SET_GUEST_DEBUG2 capability indicates the
3649supported KVM_GUESTDBG_* bits in the control field.
3650
3651When debug events exit the main run loop with the reason
3652KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
3653structure containing architecture specific debug information.
3654
36554.88 KVM_GET_EMULATED_CPUID
3656---------------------------
3657
3658:Capability: KVM_CAP_EXT_EMUL_CPUID
3659:Architectures: x86
3660:Type: system ioctl
3661:Parameters: struct kvm_cpuid2 (in/out)
3662:Returns: 0 on success, -1 on error
3663
3664::
3665
3666  struct kvm_cpuid2 {
3667	__u32 nent;
3668	__u32 flags;
3669	struct kvm_cpuid_entry2 entries[0];
3670  };
3671
3672The member 'flags' is used for passing flags from userspace.
3673
3674::
3675
3676  #define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
3677  #define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1) /* deprecated */
3678  #define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2) /* deprecated */
3679
3680  struct kvm_cpuid_entry2 {
3681	__u32 function;
3682	__u32 index;
3683	__u32 flags;
3684	__u32 eax;
3685	__u32 ebx;
3686	__u32 ecx;
3687	__u32 edx;
3688	__u32 padding[3];
3689  };
3690
3691This ioctl returns x86 cpuid features which are emulated by
3692kvm.Userspace can use the information returned by this ioctl to query
3693which features are emulated by kvm instead of being present natively.
3694
3695Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
3696structure with the 'nent' field indicating the number of entries in
3697the variable-size array 'entries'. If the number of entries is too low
3698to describe the cpu capabilities, an error (E2BIG) is returned. If the
3699number is too high, the 'nent' field is adjusted and an error (ENOMEM)
3700is returned. If the number is just right, the 'nent' field is adjusted
3701to the number of valid entries in the 'entries' array, which is then
3702filled.
3703
3704The entries returned are the set CPUID bits of the respective features
3705which kvm emulates, as returned by the CPUID instruction, with unknown
3706or unsupported feature bits cleared.
3707
3708Features like x2apic, for example, may not be present in the host cpu
3709but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
3710emulated efficiently and thus not included here.
3711
3712The fields in each entry are defined as follows:
3713
3714  function:
3715	 the eax value used to obtain the entry
3716  index:
3717	 the ecx value used to obtain the entry (for entries that are
3718         affected by ecx)
3719  flags:
3720    an OR of zero or more of the following:
3721
3722        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
3723           if the index field is valid
3724
3725   eax, ebx, ecx, edx:
3726
3727         the values returned by the cpuid instruction for
3728         this function/index combination
3729
37304.89 KVM_S390_MEM_OP
3731--------------------
3732
3733:Capability: KVM_CAP_S390_MEM_OP, KVM_CAP_S390_PROTECTED, KVM_CAP_S390_MEM_OP_EXTENSION
3734:Architectures: s390
3735:Type: vm ioctl, vcpu ioctl
3736:Parameters: struct kvm_s390_mem_op (in)
3737:Returns: = 0 on success,
3738          < 0 on generic error (e.g. -EFAULT or -ENOMEM),
3739          > 0 if an exception occurred while walking the page tables
3740
3741Read or write data from/to the VM's memory.
3742The KVM_CAP_S390_MEM_OP_EXTENSION capability specifies what functionality is
3743supported.
3744
3745Parameters are specified via the following structure::
3746
3747  struct kvm_s390_mem_op {
3748	__u64 gaddr;		/* the guest address */
3749	__u64 flags;		/* flags */
3750	__u32 size;		/* amount of bytes */
3751	__u32 op;		/* type of operation */
3752	__u64 buf;		/* buffer in userspace */
3753	union {
3754		struct {
3755			__u8 ar;	/* the access register number */
3756			__u8 key;	/* access key, ignored if flag unset */
3757		};
3758		__u32 sida_offset; /* offset into the sida */
3759		__u8 reserved[32]; /* ignored */
3760	};
3761  };
3762
3763The start address of the memory region has to be specified in the "gaddr"
3764field, and the length of the region in the "size" field (which must not
3765be 0). The maximum value for "size" can be obtained by checking the
3766KVM_CAP_S390_MEM_OP capability. "buf" is the buffer supplied by the
3767userspace application where the read data should be written to for
3768a read access, or where the data that should be written is stored for
3769a write access.  The "reserved" field is meant for future extensions.
3770Reserved and unused values are ignored. Future extension that add members must
3771introduce new flags.
3772
3773The type of operation is specified in the "op" field. Flags modifying
3774their behavior can be set in the "flags" field. Undefined flag bits must
3775be set to 0.
3776
3777Possible operations are:
3778  * ``KVM_S390_MEMOP_LOGICAL_READ``
3779  * ``KVM_S390_MEMOP_LOGICAL_WRITE``
3780  * ``KVM_S390_MEMOP_ABSOLUTE_READ``
3781  * ``KVM_S390_MEMOP_ABSOLUTE_WRITE``
3782  * ``KVM_S390_MEMOP_SIDA_READ``
3783  * ``KVM_S390_MEMOP_SIDA_WRITE``
3784
3785Logical read/write:
3786^^^^^^^^^^^^^^^^^^^
3787
3788Access logical memory, i.e. translate the given guest address to an absolute
3789address given the state of the VCPU and use the absolute address as target of
3790the access. "ar" designates the access register number to be used; the valid
3791range is 0..15.
3792Logical accesses are permitted for the VCPU ioctl only.
3793Logical accesses are permitted for non-protected guests only.
3794
3795Supported flags:
3796  * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3797  * ``KVM_S390_MEMOP_F_INJECT_EXCEPTION``
3798  * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3799
3800The KVM_S390_MEMOP_F_CHECK_ONLY flag can be set to check whether the
3801corresponding memory access would cause an access exception; however,
3802no actual access to the data in memory at the destination is performed.
3803In this case, "buf" is unused and can be NULL.
3804
3805In case an access exception occurred during the access (or would occur
3806in case of KVM_S390_MEMOP_F_CHECK_ONLY), the ioctl returns a positive
3807error number indicating the type of exception. This exception is also
3808raised directly at the corresponding VCPU if the flag
3809KVM_S390_MEMOP_F_INJECT_EXCEPTION is set.
3810On protection exceptions, unless specified otherwise, the injected
3811translation-exception identifier (TEID) indicates suppression.
3812
3813If the KVM_S390_MEMOP_F_SKEY_PROTECTION flag is set, storage key
3814protection is also in effect and may cause exceptions if accesses are
3815prohibited given the access key designated by "key"; the valid range is 0..15.
3816KVM_S390_MEMOP_F_SKEY_PROTECTION is available if KVM_CAP_S390_MEM_OP_EXTENSION
3817is > 0.
3818Since the accessed memory may span multiple pages and those pages might have
3819different storage keys, it is possible that a protection exception occurs
3820after memory has been modified. In this case, if the exception is injected,
3821the TEID does not indicate suppression.
3822
3823Absolute read/write:
3824^^^^^^^^^^^^^^^^^^^^
3825
3826Access absolute memory. This operation is intended to be used with the
3827KVM_S390_MEMOP_F_SKEY_PROTECTION flag, to allow accessing memory and performing
3828the checks required for storage key protection as one operation (as opposed to
3829user space getting the storage keys, performing the checks, and accessing
3830memory thereafter, which could lead to a delay between check and access).
3831Absolute accesses are permitted for the VM ioctl if KVM_CAP_S390_MEM_OP_EXTENSION
3832is > 0.
3833Currently absolute accesses are not permitted for VCPU ioctls.
3834Absolute accesses are permitted for non-protected guests only.
3835
3836Supported flags:
3837  * ``KVM_S390_MEMOP_F_CHECK_ONLY``
3838  * ``KVM_S390_MEMOP_F_SKEY_PROTECTION``
3839
3840The semantics of the flags are as for logical accesses.
3841
3842SIDA read/write:
3843^^^^^^^^^^^^^^^^
3844
3845Access the secure instruction data area which contains memory operands necessary
3846for instruction emulation for protected guests.
3847SIDA accesses are available if the KVM_CAP_S390_PROTECTED capability is available.
3848SIDA accesses are permitted for the VCPU ioctl only.
3849SIDA accesses are permitted for protected guests only.
3850
3851No flags are supported.
3852
38534.90 KVM_S390_GET_SKEYS
3854-----------------------
3855
3856:Capability: KVM_CAP_S390_SKEYS
3857:Architectures: s390
3858:Type: vm ioctl
3859:Parameters: struct kvm_s390_skeys
3860:Returns: 0 on success, KVM_S390_GET_SKEYS_NONE if guest is not using storage
3861          keys, negative value on error
3862
3863This ioctl is used to get guest storage key values on the s390
3864architecture. The ioctl takes parameters via the kvm_s390_skeys struct::
3865
3866  struct kvm_s390_skeys {
3867	__u64 start_gfn;
3868	__u64 count;
3869	__u64 skeydata_addr;
3870	__u32 flags;
3871	__u32 reserved[9];
3872  };
3873
3874The start_gfn field is the number of the first guest frame whose storage keys
3875you want to get.
3876
3877The count field is the number of consecutive frames (starting from start_gfn)
3878whose storage keys to get. The count field must be at least 1 and the maximum
3879allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3880will cause the ioctl to return -EINVAL.
3881
3882The skeydata_addr field is the address to a buffer large enough to hold count
3883bytes. This buffer will be filled with storage key data by the ioctl.
3884
38854.91 KVM_S390_SET_SKEYS
3886-----------------------
3887
3888:Capability: KVM_CAP_S390_SKEYS
3889:Architectures: s390
3890:Type: vm ioctl
3891:Parameters: struct kvm_s390_skeys
3892:Returns: 0 on success, negative value on error
3893
3894This ioctl is used to set guest storage key values on the s390
3895architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
3896See section on KVM_S390_GET_SKEYS for struct definition.
3897
3898The start_gfn field is the number of the first guest frame whose storage keys
3899you want to set.
3900
3901The count field is the number of consecutive frames (starting from start_gfn)
3902whose storage keys to get. The count field must be at least 1 and the maximum
3903allowed value is defined as KVM_S390_SKEYS_MAX. Values outside this range
3904will cause the ioctl to return -EINVAL.
3905
3906The skeydata_addr field is the address to a buffer containing count bytes of
3907storage keys. Each byte in the buffer will be set as the storage key for a
3908single frame starting at start_gfn for count frames.
3909
3910Note: If any architecturally invalid key value is found in the given data then
3911the ioctl will return -EINVAL.
3912
39134.92 KVM_S390_IRQ
3914-----------------
3915
3916:Capability: KVM_CAP_S390_INJECT_IRQ
3917:Architectures: s390
3918:Type: vcpu ioctl
3919:Parameters: struct kvm_s390_irq (in)
3920:Returns: 0 on success, -1 on error
3921
3922Errors:
3923
3924
3925  ======  =================================================================
3926  EINVAL  interrupt type is invalid
3927          type is KVM_S390_SIGP_STOP and flag parameter is invalid value,
3928          type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
3929          than the maximum of VCPUs
3930  EBUSY   type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped,
3931          type is KVM_S390_SIGP_STOP and a stop irq is already pending,
3932          type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
3933          is already pending
3934  ======  =================================================================
3935
3936Allows to inject an interrupt to the guest.
3937
3938Using struct kvm_s390_irq as a parameter allows
3939to inject additional payload which is not
3940possible via KVM_S390_INTERRUPT.
3941
3942Interrupt parameters are passed via kvm_s390_irq::
3943
3944  struct kvm_s390_irq {
3945	__u64 type;
3946	union {
3947		struct kvm_s390_io_info io;
3948		struct kvm_s390_ext_info ext;
3949		struct kvm_s390_pgm_info pgm;
3950		struct kvm_s390_emerg_info emerg;
3951		struct kvm_s390_extcall_info extcall;
3952		struct kvm_s390_prefix_info prefix;
3953		struct kvm_s390_stop_info stop;
3954		struct kvm_s390_mchk_info mchk;
3955		char reserved[64];
3956	} u;
3957  };
3958
3959type can be one of the following:
3960
3961- KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
3962- KVM_S390_PROGRAM_INT - program check; parameters in .pgm
3963- KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
3964- KVM_S390_RESTART - restart; no parameters
3965- KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
3966- KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
3967- KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
3968- KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
3969- KVM_S390_MCHK - machine check interrupt; parameters in .mchk
3970
3971This is an asynchronous vcpu ioctl and can be invoked from any thread.
3972
39734.94 KVM_S390_GET_IRQ_STATE
3974---------------------------
3975
3976:Capability: KVM_CAP_S390_IRQ_STATE
3977:Architectures: s390
3978:Type: vcpu ioctl
3979:Parameters: struct kvm_s390_irq_state (out)
3980:Returns: >= number of bytes copied into buffer,
3981          -EINVAL if buffer size is 0,
3982          -ENOBUFS if buffer size is too small to fit all pending interrupts,
3983          -EFAULT if the buffer address was invalid
3984
3985This ioctl allows userspace to retrieve the complete state of all currently
3986pending interrupts in a single buffer. Use cases include migration
3987and introspection. The parameter structure contains the address of a
3988userspace buffer and its length::
3989
3990  struct kvm_s390_irq_state {
3991	__u64 buf;
3992	__u32 flags;        /* will stay unused for compatibility reasons */
3993	__u32 len;
3994	__u32 reserved[4];  /* will stay unused for compatibility reasons */
3995  };
3996
3997Userspace passes in the above struct and for each pending interrupt a
3998struct kvm_s390_irq is copied to the provided buffer.
3999
4000The structure contains a flags and a reserved field for future extensions. As
4001the kernel never checked for flags == 0 and QEMU never pre-zeroed flags and
4002reserved, these fields can not be used in the future without breaking
4003compatibility.
4004
4005If -ENOBUFS is returned the buffer provided was too small and userspace
4006may retry with a bigger buffer.
4007
40084.95 KVM_S390_SET_IRQ_STATE
4009---------------------------
4010
4011:Capability: KVM_CAP_S390_IRQ_STATE
4012:Architectures: s390
4013:Type: vcpu ioctl
4014:Parameters: struct kvm_s390_irq_state (in)
4015:Returns: 0 on success,
4016          -EFAULT if the buffer address was invalid,
4017          -EINVAL for an invalid buffer length (see below),
4018          -EBUSY if there were already interrupts pending,
4019          errors occurring when actually injecting the
4020          interrupt. See KVM_S390_IRQ.
4021
4022This ioctl allows userspace to set the complete state of all cpu-local
4023interrupts currently pending for the vcpu. It is intended for restoring
4024interrupt state after a migration. The input parameter is a userspace buffer
4025containing a struct kvm_s390_irq_state::
4026
4027  struct kvm_s390_irq_state {
4028	__u64 buf;
4029	__u32 flags;        /* will stay unused for compatibility reasons */
4030	__u32 len;
4031	__u32 reserved[4];  /* will stay unused for compatibility reasons */
4032  };
4033
4034The restrictions for flags and reserved apply as well.
4035(see KVM_S390_GET_IRQ_STATE)
4036
4037The userspace memory referenced by buf contains a struct kvm_s390_irq
4038for each interrupt to be injected into the guest.
4039If one of the interrupts could not be injected for some reason the
4040ioctl aborts.
4041
4042len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
4043and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
4044which is the maximum number of possibly pending cpu-local interrupts.
4045
40464.96 KVM_SMI
4047------------
4048
4049:Capability: KVM_CAP_X86_SMM
4050:Architectures: x86
4051:Type: vcpu ioctl
4052:Parameters: none
4053:Returns: 0 on success, -1 on error
4054
4055Queues an SMI on the thread's vcpu.
4056
40574.97 KVM_X86_SET_MSR_FILTER
4058----------------------------
4059
4060:Capability: KVM_CAP_X86_MSR_FILTER
4061:Architectures: x86
4062:Type: vm ioctl
4063:Parameters: struct kvm_msr_filter
4064:Returns: 0 on success, < 0 on error
4065
4066::
4067
4068  struct kvm_msr_filter_range {
4069  #define KVM_MSR_FILTER_READ  (1 << 0)
4070  #define KVM_MSR_FILTER_WRITE (1 << 1)
4071	__u32 flags;
4072	__u32 nmsrs; /* number of msrs in bitmap */
4073	__u32 base;  /* MSR index the bitmap starts at */
4074	__u8 *bitmap; /* a 1 bit allows the operations in flags, 0 denies */
4075  };
4076
4077  #define KVM_MSR_FILTER_MAX_RANGES 16
4078  struct kvm_msr_filter {
4079  #define KVM_MSR_FILTER_DEFAULT_ALLOW (0 << 0)
4080  #define KVM_MSR_FILTER_DEFAULT_DENY  (1 << 0)
4081	__u32 flags;
4082	struct kvm_msr_filter_range ranges[KVM_MSR_FILTER_MAX_RANGES];
4083  };
4084
4085flags values for ``struct kvm_msr_filter_range``:
4086
4087``KVM_MSR_FILTER_READ``
4088
4089  Filter read accesses to MSRs using the given bitmap. A 0 in the bitmap
4090  indicates that read accesses should be denied, while a 1 indicates that
4091  a read for a particular MSR should be allowed regardless of the default
4092  filter action.
4093
4094``KVM_MSR_FILTER_WRITE``
4095
4096  Filter write accesses to MSRs using the given bitmap. A 0 in the bitmap
4097  indicates that write accesses should be denied, while a 1 indicates that
4098  a write for a particular MSR should be allowed regardless of the default
4099  filter action.
4100
4101flags values for ``struct kvm_msr_filter``:
4102
4103``KVM_MSR_FILTER_DEFAULT_ALLOW``
4104
4105  If no filter range matches an MSR index that is getting accessed, KVM will
4106  allow accesses to all MSRs by default.
4107
4108``KVM_MSR_FILTER_DEFAULT_DENY``
4109
4110  If no filter range matches an MSR index that is getting accessed, KVM will
4111  deny accesses to all MSRs by default.
4112
4113This ioctl allows userspace to define up to 16 bitmaps of MSR ranges to deny
4114guest MSR accesses that would normally be allowed by KVM.  If an MSR is not
4115covered by a specific range, the "default" filtering behavior applies.  Each
4116bitmap range covers MSRs from [base .. base+nmsrs).
4117
4118If an MSR access is denied by userspace, the resulting KVM behavior depends on
4119whether or not KVM_CAP_X86_USER_SPACE_MSR's KVM_MSR_EXIT_REASON_FILTER is
4120enabled.  If KVM_MSR_EXIT_REASON_FILTER is enabled, KVM will exit to userspace
4121on denied accesses, i.e. userspace effectively intercepts the MSR access.  If
4122KVM_MSR_EXIT_REASON_FILTER is not enabled, KVM will inject a #GP into the guest
4123on denied accesses.
4124
4125If an MSR access is allowed by userspace, KVM will emulate and/or virtualize
4126the access in accordance with the vCPU model.  Note, KVM may still ultimately
4127inject a #GP if an access is allowed by userspace, e.g. if KVM doesn't support
4128the MSR, or to follow architectural behavior for the MSR.
4129
4130By default, KVM operates in KVM_MSR_FILTER_DEFAULT_ALLOW mode with no MSR range
4131filters.
4132
4133Calling this ioctl with an empty set of ranges (all nmsrs == 0) disables MSR
4134filtering. In that mode, ``KVM_MSR_FILTER_DEFAULT_DENY`` is invalid and causes
4135an error.
4136
4137.. warning::
4138   MSR accesses as part of nested VM-Enter/VM-Exit are not filtered.
4139   This includes both writes to individual VMCS fields and reads/writes
4140   through the MSR lists pointed to by the VMCS.
4141
4142   x2APIC MSR accesses cannot be filtered (KVM silently ignores filters that
4143   cover any x2APIC MSRs).
4144
4145Note, invoking this ioctl while a vCPU is running is inherently racy.  However,
4146KVM does guarantee that vCPUs will see either the previous filter or the new
4147filter, e.g. MSRs with identical settings in both the old and new filter will
4148have deterministic behavior.
4149
4150Similarly, if userspace wishes to intercept on denied accesses,
4151KVM_MSR_EXIT_REASON_FILTER must be enabled before activating any filters, and
4152left enabled until after all filters are deactivated.  Failure to do so may
4153result in KVM injecting a #GP instead of exiting to userspace.
4154
41554.98 KVM_CREATE_SPAPR_TCE_64
4156----------------------------
4157
4158:Capability: KVM_CAP_SPAPR_TCE_64
4159:Architectures: powerpc
4160:Type: vm ioctl
4161:Parameters: struct kvm_create_spapr_tce_64 (in)
4162:Returns: file descriptor for manipulating the created TCE table
4163
4164This is an extension for KVM_CAP_SPAPR_TCE which only supports 32bit
4165windows, described in 4.62 KVM_CREATE_SPAPR_TCE
4166
4167This capability uses extended struct in ioctl interface::
4168
4169  /* for KVM_CAP_SPAPR_TCE_64 */
4170  struct kvm_create_spapr_tce_64 {
4171	__u64 liobn;
4172	__u32 page_shift;
4173	__u32 flags;
4174	__u64 offset;	/* in pages */
4175	__u64 size; 	/* in pages */
4176  };
4177
4178The aim of extension is to support an additional bigger DMA window with
4179a variable page size.
4180KVM_CREATE_SPAPR_TCE_64 receives a 64bit window size, an IOMMU page shift and
4181a bus offset of the corresponding DMA window, @size and @offset are numbers
4182of IOMMU pages.
4183
4184@flags are not used at the moment.
4185
4186The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
4187
41884.99 KVM_REINJECT_CONTROL
4189-------------------------
4190
4191:Capability: KVM_CAP_REINJECT_CONTROL
4192:Architectures: x86
4193:Type: vm ioctl
4194:Parameters: struct kvm_reinject_control (in)
4195:Returns: 0 on success,
4196         -EFAULT if struct kvm_reinject_control cannot be read,
4197         -ENXIO if KVM_CREATE_PIT or KVM_CREATE_PIT2 didn't succeed earlier.
4198
4199i8254 (PIT) has two modes, reinject and !reinject.  The default is reinject,
4200where KVM queues elapsed i8254 ticks and monitors completion of interrupt from
4201vector(s) that i8254 injects.  Reinject mode dequeues a tick and injects its
4202interrupt whenever there isn't a pending interrupt from i8254.
4203!reinject mode injects an interrupt as soon as a tick arrives.
4204
4205::
4206
4207  struct kvm_reinject_control {
4208	__u8 pit_reinject;
4209	__u8 reserved[31];
4210  };
4211
4212pit_reinject = 0 (!reinject mode) is recommended, unless running an old
4213operating system that uses the PIT for timing (e.g. Linux 2.4.x).
4214
42154.100 KVM_PPC_CONFIGURE_V3_MMU
4216------------------------------
4217
4218:Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
4219:Architectures: ppc
4220:Type: vm ioctl
4221:Parameters: struct kvm_ppc_mmuv3_cfg (in)
4222:Returns: 0 on success,
4223         -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
4224         -EINVAL if the configuration is invalid
4225
4226This ioctl controls whether the guest will use radix or HPT (hashed
4227page table) translation, and sets the pointer to the process table for
4228the guest.
4229
4230::
4231
4232  struct kvm_ppc_mmuv3_cfg {
4233	__u64	flags;
4234	__u64	process_table;
4235  };
4236
4237There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
4238KVM_PPC_MMUV3_GTSE.  KVM_PPC_MMUV3_RADIX, if set, configures the guest
4239to use radix tree translation, and if clear, to use HPT translation.
4240KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
4241to be able to use the global TLB and SLB invalidation instructions;
4242if clear, the guest may not use these instructions.
4243
4244The process_table field specifies the address and size of the guest
4245process table, which is in the guest's space.  This field is formatted
4246as the second doubleword of the partition table entry, as defined in
4247the Power ISA V3.00, Book III section 5.7.6.1.
4248
42494.101 KVM_PPC_GET_RMMU_INFO
4250---------------------------
4251
4252:Capability: KVM_CAP_PPC_RADIX_MMU
4253:Architectures: ppc
4254:Type: vm ioctl
4255:Parameters: struct kvm_ppc_rmmu_info (out)
4256:Returns: 0 on success,
4257	 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
4258	 -EINVAL if no useful information can be returned
4259
4260This ioctl returns a structure containing two things: (a) a list
4261containing supported radix tree geometries, and (b) a list that maps
4262page sizes to put in the "AP" (actual page size) field for the tlbie
4263(TLB invalidate entry) instruction.
4264
4265::
4266
4267  struct kvm_ppc_rmmu_info {
4268	struct kvm_ppc_radix_geom {
4269		__u8	page_shift;
4270		__u8	level_bits[4];
4271		__u8	pad[3];
4272	}	geometries[8];
4273	__u32	ap_encodings[8];
4274  };
4275
4276The geometries[] field gives up to 8 supported geometries for the
4277radix page table, in terms of the log base 2 of the smallest page
4278size, and the number of bits indexed at each level of the tree, from
4279the PTE level up to the PGD level in that order.  Any unused entries
4280will have 0 in the page_shift field.
4281
4282The ap_encodings gives the supported page sizes and their AP field
4283encodings, encoded with the AP value in the top 3 bits and the log
4284base 2 of the page size in the bottom 6 bits.
4285
42864.102 KVM_PPC_RESIZE_HPT_PREPARE
4287--------------------------------
4288
4289:Capability: KVM_CAP_SPAPR_RESIZE_HPT
4290:Architectures: powerpc
4291:Type: vm ioctl
4292:Parameters: struct kvm_ppc_resize_hpt (in)
4293:Returns: 0 on successful completion,
4294	 >0 if a new HPT is being prepared, the value is an estimated
4295         number of milliseconds until preparation is complete,
4296         -EFAULT if struct kvm_reinject_control cannot be read,
4297	 -EINVAL if the supplied shift or flags are invalid,
4298	 -ENOMEM if unable to allocate the new HPT,
4299
4300Used to implement the PAPR extension for runtime resizing of a guest's
4301Hashed Page Table (HPT).  Specifically this starts, stops or monitors
4302the preparation of a new potential HPT for the guest, essentially
4303implementing the H_RESIZE_HPT_PREPARE hypercall.
4304
4305::
4306
4307  struct kvm_ppc_resize_hpt {
4308	__u64 flags;
4309	__u32 shift;
4310	__u32 pad;
4311  };
4312
4313If called with shift > 0 when there is no pending HPT for the guest,
4314this begins preparation of a new pending HPT of size 2^(shift) bytes.
4315It then returns a positive integer with the estimated number of
4316milliseconds until preparation is complete.
4317
4318If called when there is a pending HPT whose size does not match that
4319requested in the parameters, discards the existing pending HPT and
4320creates a new one as above.
4321
4322If called when there is a pending HPT of the size requested, will:
4323
4324  * If preparation of the pending HPT is already complete, return 0
4325  * If preparation of the pending HPT has failed, return an error
4326    code, then discard the pending HPT.
4327  * If preparation of the pending HPT is still in progress, return an
4328    estimated number of milliseconds until preparation is complete.
4329
4330If called with shift == 0, discards any currently pending HPT and
4331returns 0 (i.e. cancels any in-progress preparation).
4332
4333flags is reserved for future expansion, currently setting any bits in
4334flags will result in an -EINVAL.
4335
4336Normally this will be called repeatedly with the same parameters until
4337it returns <= 0.  The first call will initiate preparation, subsequent
4338ones will monitor preparation until it completes or fails.
4339
43404.103 KVM_PPC_RESIZE_HPT_COMMIT
4341-------------------------------
4342
4343:Capability: KVM_CAP_SPAPR_RESIZE_HPT
4344:Architectures: powerpc
4345:Type: vm ioctl
4346:Parameters: struct kvm_ppc_resize_hpt (in)
4347:Returns: 0 on successful completion,
4348         -EFAULT if struct kvm_reinject_control cannot be read,
4349	 -EINVAL if the supplied shift or flags are invalid,
4350	 -ENXIO is there is no pending HPT, or the pending HPT doesn't
4351         have the requested size,
4352	 -EBUSY if the pending HPT is not fully prepared,
4353	 -ENOSPC if there was a hash collision when moving existing
4354         HPT entries to the new HPT,
4355	 -EIO on other error conditions
4356
4357Used to implement the PAPR extension for runtime resizing of a guest's
4358Hashed Page Table (HPT).  Specifically this requests that the guest be
4359transferred to working with the new HPT, essentially implementing the
4360H_RESIZE_HPT_COMMIT hypercall.
4361
4362::
4363
4364  struct kvm_ppc_resize_hpt {
4365	__u64 flags;
4366	__u32 shift;
4367	__u32 pad;
4368  };
4369
4370This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
4371returned 0 with the same parameters.  In other cases
4372KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
4373-EBUSY, though others may be possible if the preparation was started,
4374but failed).
4375
4376This will have undefined effects on the guest if it has not already
4377placed itself in a quiescent state where no vcpu will make MMU enabled
4378memory accesses.
4379
4380On succsful completion, the pending HPT will become the guest's active
4381HPT and the previous HPT will be discarded.
4382
4383On failure, the guest will still be operating on its previous HPT.
4384
43854.104 KVM_X86_GET_MCE_CAP_SUPPORTED
4386-----------------------------------
4387
4388:Capability: KVM_CAP_MCE
4389:Architectures: x86
4390:Type: system ioctl
4391:Parameters: u64 mce_cap (out)
4392:Returns: 0 on success, -1 on error
4393
4394Returns supported MCE capabilities. The u64 mce_cap parameter
4395has the same format as the MSR_IA32_MCG_CAP register. Supported
4396capabilities will have the corresponding bits set.
4397
43984.105 KVM_X86_SETUP_MCE
4399-----------------------
4400
4401:Capability: KVM_CAP_MCE
4402:Architectures: x86
4403:Type: vcpu ioctl
4404:Parameters: u64 mcg_cap (in)
4405:Returns: 0 on success,
4406         -EFAULT if u64 mcg_cap cannot be read,
4407         -EINVAL if the requested number of banks is invalid,
4408         -EINVAL if requested MCE capability is not supported.
4409
4410Initializes MCE support for use. The u64 mcg_cap parameter
4411has the same format as the MSR_IA32_MCG_CAP register and
4412specifies which capabilities should be enabled. The maximum
4413supported number of error-reporting banks can be retrieved when
4414checking for KVM_CAP_MCE. The supported capabilities can be
4415retrieved with KVM_X86_GET_MCE_CAP_SUPPORTED.
4416
44174.106 KVM_X86_SET_MCE
4418---------------------
4419
4420:Capability: KVM_CAP_MCE
4421:Architectures: x86
4422:Type: vcpu ioctl
4423:Parameters: struct kvm_x86_mce (in)
4424:Returns: 0 on success,
4425         -EFAULT if struct kvm_x86_mce cannot be read,
4426         -EINVAL if the bank number is invalid,
4427         -EINVAL if VAL bit is not set in status field.
4428
4429Inject a machine check error (MCE) into the guest. The input
4430parameter is::
4431
4432  struct kvm_x86_mce {
4433	__u64 status;
4434	__u64 addr;
4435	__u64 misc;
4436	__u64 mcg_status;
4437	__u8 bank;
4438	__u8 pad1[7];
4439	__u64 pad2[3];
4440  };
4441
4442If the MCE being reported is an uncorrected error, KVM will
4443inject it as an MCE exception into the guest. If the guest
4444MCG_STATUS register reports that an MCE is in progress, KVM
4445causes an KVM_EXIT_SHUTDOWN vmexit.
4446
4447Otherwise, if the MCE is a corrected error, KVM will just
4448store it in the corresponding bank (provided this bank is
4449not holding a previously reported uncorrected error).
4450
44514.107 KVM_S390_GET_CMMA_BITS
4452----------------------------
4453
4454:Capability: KVM_CAP_S390_CMMA_MIGRATION
4455:Architectures: s390
4456:Type: vm ioctl
4457:Parameters: struct kvm_s390_cmma_log (in, out)
4458:Returns: 0 on success, a negative value on error
4459
4460This ioctl is used to get the values of the CMMA bits on the s390
4461architecture. It is meant to be used in two scenarios:
4462
4463- During live migration to save the CMMA values. Live migration needs
4464  to be enabled via the KVM_REQ_START_MIGRATION VM property.
4465- To non-destructively peek at the CMMA values, with the flag
4466  KVM_S390_CMMA_PEEK set.
4467
4468The ioctl takes parameters via the kvm_s390_cmma_log struct. The desired
4469values are written to a buffer whose location is indicated via the "values"
4470member in the kvm_s390_cmma_log struct.  The values in the input struct are
4471also updated as needed.
4472
4473Each CMMA value takes up one byte.
4474
4475::
4476
4477  struct kvm_s390_cmma_log {
4478	__u64 start_gfn;
4479	__u32 count;
4480	__u32 flags;
4481	union {
4482		__u64 remaining;
4483		__u64 mask;
4484	};
4485	__u64 values;
4486  };
4487
4488start_gfn is the number of the first guest frame whose CMMA values are
4489to be retrieved,
4490
4491count is the length of the buffer in bytes,
4492
4493values points to the buffer where the result will be written to.
4494
4495If count is greater than KVM_S390_SKEYS_MAX, then it is considered to be
4496KVM_S390_SKEYS_MAX. KVM_S390_SKEYS_MAX is re-used for consistency with
4497other ioctls.
4498
4499The result is written in the buffer pointed to by the field values, and
4500the values of the input parameter are updated as follows.
4501
4502Depending on the flags, different actions are performed. The only
4503supported flag so far is KVM_S390_CMMA_PEEK.
4504
4505The default behaviour if KVM_S390_CMMA_PEEK is not set is:
4506start_gfn will indicate the first page frame whose CMMA bits were dirty.
4507It is not necessarily the same as the one passed as input, as clean pages
4508are skipped.
4509
4510count will indicate the number of bytes actually written in the buffer.
4511It can (and very often will) be smaller than the input value, since the
4512buffer is only filled until 16 bytes of clean values are found (which
4513are then not copied in the buffer). Since a CMMA migration block needs
4514the base address and the length, for a total of 16 bytes, we will send
4515back some clean data if there is some dirty data afterwards, as long as
4516the size of the clean data does not exceed the size of the header. This
4517allows to minimize the amount of data to be saved or transferred over
4518the network at the expense of more roundtrips to userspace. The next
4519invocation of the ioctl will skip over all the clean values, saving
4520potentially more than just the 16 bytes we found.
4521
4522If KVM_S390_CMMA_PEEK is set:
4523the existing storage attributes are read even when not in migration
4524mode, and no other action is performed;
4525
4526the output start_gfn will be equal to the input start_gfn,
4527
4528the output count will be equal to the input count, except if the end of
4529memory has been reached.
4530
4531In both cases:
4532the field "remaining" will indicate the total number of dirty CMMA values
4533still remaining, or 0 if KVM_S390_CMMA_PEEK is set and migration mode is
4534not enabled.
4535
4536mask is unused.
4537
4538values points to the userspace buffer where the result will be stored.
4539
4540This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4541complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4542KVM_S390_CMMA_PEEK is not set but migration mode was not enabled, with
4543-EFAULT if the userspace address is invalid or if no page table is
4544present for the addresses (e.g. when using hugepages).
4545
45464.108 KVM_S390_SET_CMMA_BITS
4547----------------------------
4548
4549:Capability: KVM_CAP_S390_CMMA_MIGRATION
4550:Architectures: s390
4551:Type: vm ioctl
4552:Parameters: struct kvm_s390_cmma_log (in)
4553:Returns: 0 on success, a negative value on error
4554
4555This ioctl is used to set the values of the CMMA bits on the s390
4556architecture. It is meant to be used during live migration to restore
4557the CMMA values, but there are no restrictions on its use.
4558The ioctl takes parameters via the kvm_s390_cmma_values struct.
4559Each CMMA value takes up one byte.
4560
4561::
4562
4563  struct kvm_s390_cmma_log {
4564	__u64 start_gfn;
4565	__u32 count;
4566	__u32 flags;
4567	union {
4568		__u64 remaining;
4569		__u64 mask;
4570 	};
4571	__u64 values;
4572  };
4573
4574start_gfn indicates the starting guest frame number,
4575
4576count indicates how many values are to be considered in the buffer,
4577
4578flags is not used and must be 0.
4579
4580mask indicates which PGSTE bits are to be considered.
4581
4582remaining is not used.
4583
4584values points to the buffer in userspace where to store the values.
4585
4586This ioctl can fail with -ENOMEM if not enough memory can be allocated to
4587complete the task, with -ENXIO if CMMA is not enabled, with -EINVAL if
4588the count field is too large (e.g. more than KVM_S390_CMMA_SIZE_MAX) or
4589if the flags field was not 0, with -EFAULT if the userspace address is
4590invalid, if invalid pages are written to (e.g. after the end of memory)
4591or if no page table is present for the addresses (e.g. when using
4592hugepages).
4593
45944.109 KVM_PPC_GET_CPU_CHAR
4595--------------------------
4596
4597:Capability: KVM_CAP_PPC_GET_CPU_CHAR
4598:Architectures: powerpc
4599:Type: vm ioctl
4600:Parameters: struct kvm_ppc_cpu_char (out)
4601:Returns: 0 on successful completion,
4602	 -EFAULT if struct kvm_ppc_cpu_char cannot be written
4603
4604This ioctl gives userspace information about certain characteristics
4605of the CPU relating to speculative execution of instructions and
4606possible information leakage resulting from speculative execution (see
4607CVE-2017-5715, CVE-2017-5753 and CVE-2017-5754).  The information is
4608returned in struct kvm_ppc_cpu_char, which looks like this::
4609
4610  struct kvm_ppc_cpu_char {
4611	__u64	character;		/* characteristics of the CPU */
4612	__u64	behaviour;		/* recommended software behaviour */
4613	__u64	character_mask;		/* valid bits in character */
4614	__u64	behaviour_mask;		/* valid bits in behaviour */
4615  };
4616
4617For extensibility, the character_mask and behaviour_mask fields
4618indicate which bits of character and behaviour have been filled in by
4619the kernel.  If the set of defined bits is extended in future then
4620userspace will be able to tell whether it is running on a kernel that
4621knows about the new bits.
4622
4623The character field describes attributes of the CPU which can help
4624with preventing inadvertent information disclosure - specifically,
4625whether there is an instruction to flash-invalidate the L1 data cache
4626(ori 30,30,0 or mtspr SPRN_TRIG2,rN), whether the L1 data cache is set
4627to a mode where entries can only be used by the thread that created
4628them, whether the bcctr[l] instruction prevents speculation, and
4629whether a speculation barrier instruction (ori 31,31,0) is provided.
4630
4631The behaviour field describes actions that software should take to
4632prevent inadvertent information disclosure, and thus describes which
4633vulnerabilities the hardware is subject to; specifically whether the
4634L1 data cache should be flushed when returning to user mode from the
4635kernel, and whether a speculation barrier should be placed between an
4636array bounds check and the array access.
4637
4638These fields use the same bit definitions as the new
4639H_GET_CPU_CHARACTERISTICS hypercall.
4640
46414.110 KVM_MEMORY_ENCRYPT_OP
4642---------------------------
4643
4644:Capability: basic
4645:Architectures: x86
4646:Type: vm
4647:Parameters: an opaque platform specific structure (in/out)
4648:Returns: 0 on success; -1 on error
4649
4650If the platform supports creating encrypted VMs then this ioctl can be used
4651for issuing platform-specific memory encryption commands to manage those
4652encrypted VMs.
4653
4654Currently, this ioctl is used for issuing Secure Encrypted Virtualization
4655(SEV) commands on AMD Processors. The SEV commands are defined in
4656Documentation/virt/kvm/x86/amd-memory-encryption.rst.
4657
46584.111 KVM_MEMORY_ENCRYPT_REG_REGION
4659-----------------------------------
4660
4661:Capability: basic
4662:Architectures: x86
4663:Type: system
4664:Parameters: struct kvm_enc_region (in)
4665:Returns: 0 on success; -1 on error
4666
4667This ioctl can be used to register a guest memory region which may
4668contain encrypted data (e.g. guest RAM, SMRAM etc).
4669
4670It is used in the SEV-enabled guest. When encryption is enabled, a guest
4671memory region may contain encrypted data. The SEV memory encryption
4672engine uses a tweak such that two identical plaintext pages, each at
4673different locations will have differing ciphertexts. So swapping or
4674moving ciphertext of those pages will not result in plaintext being
4675swapped. So relocating (or migrating) physical backing pages for the SEV
4676guest will require some additional steps.
4677
4678Note: The current SEV key management spec does not provide commands to
4679swap or migrate (move) ciphertext pages. Hence, for now we pin the guest
4680memory region registered with the ioctl.
4681
46824.112 KVM_MEMORY_ENCRYPT_UNREG_REGION
4683-------------------------------------
4684
4685:Capability: basic
4686:Architectures: x86
4687:Type: system
4688:Parameters: struct kvm_enc_region (in)
4689:Returns: 0 on success; -1 on error
4690
4691This ioctl can be used to unregister the guest memory region registered
4692with KVM_MEMORY_ENCRYPT_REG_REGION ioctl above.
4693
46944.113 KVM_HYPERV_EVENTFD
4695------------------------
4696
4697:Capability: KVM_CAP_HYPERV_EVENTFD
4698:Architectures: x86
4699:Type: vm ioctl
4700:Parameters: struct kvm_hyperv_eventfd (in)
4701
4702This ioctl (un)registers an eventfd to receive notifications from the guest on
4703the specified Hyper-V connection id through the SIGNAL_EVENT hypercall, without
4704causing a user exit.  SIGNAL_EVENT hypercall with non-zero event flag number
4705(bits 24-31) still triggers a KVM_EXIT_HYPERV_HCALL user exit.
4706
4707::
4708
4709  struct kvm_hyperv_eventfd {
4710	__u32 conn_id;
4711	__s32 fd;
4712	__u32 flags;
4713	__u32 padding[3];
4714  };
4715
4716The conn_id field should fit within 24 bits::
4717
4718  #define KVM_HYPERV_CONN_ID_MASK		0x00ffffff
4719
4720The acceptable values for the flags field are::
4721
4722  #define KVM_HYPERV_EVENTFD_DEASSIGN	(1 << 0)
4723
4724:Returns: 0 on success,
4725 	  -EINVAL if conn_id or flags is outside the allowed range,
4726	  -ENOENT on deassign if the conn_id isn't registered,
4727	  -EEXIST on assign if the conn_id is already registered
4728
47294.114 KVM_GET_NESTED_STATE
4730--------------------------
4731
4732:Capability: KVM_CAP_NESTED_STATE
4733:Architectures: x86
4734:Type: vcpu ioctl
4735:Parameters: struct kvm_nested_state (in/out)
4736:Returns: 0 on success, -1 on error
4737
4738Errors:
4739
4740  =====      =============================================================
4741  E2BIG      the total state size exceeds the value of 'size' specified by
4742             the user; the size required will be written into size.
4743  =====      =============================================================
4744
4745::
4746
4747  struct kvm_nested_state {
4748	__u16 flags;
4749	__u16 format;
4750	__u32 size;
4751
4752	union {
4753		struct kvm_vmx_nested_state_hdr vmx;
4754		struct kvm_svm_nested_state_hdr svm;
4755
4756		/* Pad the header to 128 bytes.  */
4757		__u8 pad[120];
4758	} hdr;
4759
4760	union {
4761		struct kvm_vmx_nested_state_data vmx[0];
4762		struct kvm_svm_nested_state_data svm[0];
4763	} data;
4764  };
4765
4766  #define KVM_STATE_NESTED_GUEST_MODE		0x00000001
4767  #define KVM_STATE_NESTED_RUN_PENDING		0x00000002
4768  #define KVM_STATE_NESTED_EVMCS		0x00000004
4769
4770  #define KVM_STATE_NESTED_FORMAT_VMX		0
4771  #define KVM_STATE_NESTED_FORMAT_SVM		1
4772
4773  #define KVM_STATE_NESTED_VMX_VMCS_SIZE	0x1000
4774
4775  #define KVM_STATE_NESTED_VMX_SMM_GUEST_MODE	0x00000001
4776  #define KVM_STATE_NESTED_VMX_SMM_VMXON	0x00000002
4777
4778  #define KVM_STATE_VMX_PREEMPTION_TIMER_DEADLINE 0x00000001
4779
4780  struct kvm_vmx_nested_state_hdr {
4781	__u64 vmxon_pa;
4782	__u64 vmcs12_pa;
4783
4784	struct {
4785		__u16 flags;
4786	} smm;
4787
4788	__u32 flags;
4789	__u64 preemption_timer_deadline;
4790  };
4791
4792  struct kvm_vmx_nested_state_data {
4793	__u8 vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4794	__u8 shadow_vmcs12[KVM_STATE_NESTED_VMX_VMCS_SIZE];
4795  };
4796
4797This ioctl copies the vcpu's nested virtualization state from the kernel to
4798userspace.
4799
4800The maximum size of the state can be retrieved by passing KVM_CAP_NESTED_STATE
4801to the KVM_CHECK_EXTENSION ioctl().
4802
48034.115 KVM_SET_NESTED_STATE
4804--------------------------
4805
4806:Capability: KVM_CAP_NESTED_STATE
4807:Architectures: x86
4808:Type: vcpu ioctl
4809:Parameters: struct kvm_nested_state (in)
4810:Returns: 0 on success, -1 on error
4811
4812This copies the vcpu's kvm_nested_state struct from userspace to the kernel.
4813For the definition of struct kvm_nested_state, see KVM_GET_NESTED_STATE.
4814
48154.116 KVM_(UN)REGISTER_COALESCED_MMIO
4816-------------------------------------
4817
4818:Capability: KVM_CAP_COALESCED_MMIO (for coalesced mmio)
4819	     KVM_CAP_COALESCED_PIO (for coalesced pio)
4820:Architectures: all
4821:Type: vm ioctl
4822:Parameters: struct kvm_coalesced_mmio_zone
4823:Returns: 0 on success, < 0 on error
4824
4825Coalesced I/O is a performance optimization that defers hardware
4826register write emulation so that userspace exits are avoided.  It is
4827typically used to reduce the overhead of emulating frequently accessed
4828hardware registers.
4829
4830When a hardware register is configured for coalesced I/O, write accesses
4831do not exit to userspace and their value is recorded in a ring buffer
4832that is shared between kernel and userspace.
4833
4834Coalesced I/O is used if one or more write accesses to a hardware
4835register can be deferred until a read or a write to another hardware
4836register on the same device.  This last access will cause a vmexit and
4837userspace will process accesses from the ring buffer before emulating
4838it. That will avoid exiting to userspace on repeated writes.
4839
4840Coalesced pio is based on coalesced mmio. There is little difference
4841between coalesced mmio and pio except that coalesced pio records accesses
4842to I/O ports.
4843
48444.117 KVM_CLEAR_DIRTY_LOG (vm ioctl)
4845------------------------------------
4846
4847:Capability: KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4848:Architectures: x86, arm64, mips
4849:Type: vm ioctl
4850:Parameters: struct kvm_clear_dirty_log (in)
4851:Returns: 0 on success, -1 on error
4852
4853::
4854
4855  /* for KVM_CLEAR_DIRTY_LOG */
4856  struct kvm_clear_dirty_log {
4857	__u32 slot;
4858	__u32 num_pages;
4859	__u64 first_page;
4860	union {
4861		void __user *dirty_bitmap; /* one bit per page */
4862		__u64 padding;
4863	};
4864  };
4865
4866The ioctl clears the dirty status of pages in a memory slot, according to
4867the bitmap that is passed in struct kvm_clear_dirty_log's dirty_bitmap
4868field.  Bit 0 of the bitmap corresponds to page "first_page" in the
4869memory slot, and num_pages is the size in bits of the input bitmap.
4870first_page must be a multiple of 64; num_pages must also be a multiple of
487164 unless first_page + num_pages is the size of the memory slot.  For each
4872bit that is set in the input bitmap, the corresponding page is marked "clean"
4873in KVM's dirty bitmap, and dirty tracking is re-enabled for that page
4874(for example via write-protection, or by clearing the dirty bit in
4875a page table entry).
4876
4877If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of slot field specifies
4878the address space for which you want to clear the dirty status.  See
4879KVM_SET_USER_MEMORY_REGION for details on the usage of slot field.
4880
4881This ioctl is mostly useful when KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
4882is enabled; for more information, see the description of the capability.
4883However, it can always be used as long as KVM_CHECK_EXTENSION confirms
4884that KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 is present.
4885
48864.118 KVM_GET_SUPPORTED_HV_CPUID
4887--------------------------------
4888
4889:Capability: KVM_CAP_HYPERV_CPUID (vcpu), KVM_CAP_SYS_HYPERV_CPUID (system)
4890:Architectures: x86
4891:Type: system ioctl, vcpu ioctl
4892:Parameters: struct kvm_cpuid2 (in/out)
4893:Returns: 0 on success, -1 on error
4894
4895::
4896
4897  struct kvm_cpuid2 {
4898	__u32 nent;
4899	__u32 padding;
4900	struct kvm_cpuid_entry2 entries[0];
4901  };
4902
4903  struct kvm_cpuid_entry2 {
4904	__u32 function;
4905	__u32 index;
4906	__u32 flags;
4907	__u32 eax;
4908	__u32 ebx;
4909	__u32 ecx;
4910	__u32 edx;
4911	__u32 padding[3];
4912  };
4913
4914This ioctl returns x86 cpuid features leaves related to Hyper-V emulation in
4915KVM.  Userspace can use the information returned by this ioctl to construct
4916cpuid information presented to guests consuming Hyper-V enlightenments (e.g.
4917Windows or Hyper-V guests).
4918
4919CPUID feature leaves returned by this ioctl are defined by Hyper-V Top Level
4920Functional Specification (TLFS). These leaves can't be obtained with
4921KVM_GET_SUPPORTED_CPUID ioctl because some of them intersect with KVM feature
4922leaves (0x40000000, 0x40000001).
4923
4924Currently, the following list of CPUID leaves are returned:
4925
4926 - HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS
4927 - HYPERV_CPUID_INTERFACE
4928 - HYPERV_CPUID_VERSION
4929 - HYPERV_CPUID_FEATURES
4930 - HYPERV_CPUID_ENLIGHTMENT_INFO
4931 - HYPERV_CPUID_IMPLEMENT_LIMITS
4932 - HYPERV_CPUID_NESTED_FEATURES
4933 - HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS
4934 - HYPERV_CPUID_SYNDBG_INTERFACE
4935 - HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES
4936
4937Userspace invokes KVM_GET_SUPPORTED_HV_CPUID by passing a kvm_cpuid2 structure
4938with the 'nent' field indicating the number of entries in the variable-size
4939array 'entries'.  If the number of entries is too low to describe all Hyper-V
4940feature leaves, an error (E2BIG) is returned. If the number is more or equal
4941to the number of Hyper-V feature leaves, the 'nent' field is adjusted to the
4942number of valid entries in the 'entries' array, which is then filled.
4943
4944'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
4945userspace should not expect to get any particular value there.
4946
4947Note, vcpu version of KVM_GET_SUPPORTED_HV_CPUID is currently deprecated. Unlike
4948system ioctl which exposes all supported feature bits unconditionally, vcpu
4949version has the following quirks:
4950
4951- HYPERV_CPUID_NESTED_FEATURES leaf and HV_X64_ENLIGHTENED_VMCS_RECOMMENDED
4952  feature bit are only exposed when Enlightened VMCS was previously enabled
4953  on the corresponding vCPU (KVM_CAP_HYPERV_ENLIGHTENED_VMCS).
4954- HV_STIMER_DIRECT_MODE_AVAILABLE bit is only exposed with in-kernel LAPIC.
4955  (presumes KVM_CREATE_IRQCHIP has already been called).
4956
49574.119 KVM_ARM_VCPU_FINALIZE
4958---------------------------
4959
4960:Architectures: arm64
4961:Type: vcpu ioctl
4962:Parameters: int feature (in)
4963:Returns: 0 on success, -1 on error
4964
4965Errors:
4966
4967  ======     ==============================================================
4968  EPERM      feature not enabled, needs configuration, or already finalized
4969  EINVAL     feature unknown or not present
4970  ======     ==============================================================
4971
4972Recognised values for feature:
4973
4974  =====      ===========================================
4975  arm64      KVM_ARM_VCPU_SVE (requires KVM_CAP_ARM_SVE)
4976  =====      ===========================================
4977
4978Finalizes the configuration of the specified vcpu feature.
4979
4980The vcpu must already have been initialised, enabling the affected feature, by
4981means of a successful KVM_ARM_VCPU_INIT call with the appropriate flag set in
4982features[].
4983
4984For affected vcpu features, this is a mandatory step that must be performed
4985before the vcpu is fully usable.
4986
4987Between KVM_ARM_VCPU_INIT and KVM_ARM_VCPU_FINALIZE, the feature may be
4988configured by use of ioctls such as KVM_SET_ONE_REG.  The exact configuration
4989that should be performaned and how to do it are feature-dependent.
4990
4991Other calls that depend on a particular feature being finalized, such as
4992KVM_RUN, KVM_GET_REG_LIST, KVM_GET_ONE_REG and KVM_SET_ONE_REG, will fail with
4993-EPERM unless the feature has already been finalized by means of a
4994KVM_ARM_VCPU_FINALIZE call.
4995
4996See KVM_ARM_VCPU_INIT for details of vcpu features that require finalization
4997using this ioctl.
4998
49994.120 KVM_SET_PMU_EVENT_FILTER
5000------------------------------
5001
5002:Capability: KVM_CAP_PMU_EVENT_FILTER
5003:Architectures: x86
5004:Type: vm ioctl
5005:Parameters: struct kvm_pmu_event_filter (in)
5006:Returns: 0 on success, -1 on error
5007
5008::
5009
5010  struct kvm_pmu_event_filter {
5011	__u32 action;
5012	__u32 nevents;
5013	__u32 fixed_counter_bitmap;
5014	__u32 flags;
5015	__u32 pad[4];
5016	__u64 events[0];
5017  };
5018
5019This ioctl restricts the set of PMU events that the guest can program.
5020The argument holds a list of events which will be allowed or denied.
5021The eventsel+umask of each event the guest attempts to program is compared
5022against the events field to determine whether the guest should have access.
5023The events field only controls general purpose counters; fixed purpose
5024counters are controlled by the fixed_counter_bitmap.
5025
5026No flags are defined yet, the field must be zero.
5027
5028Valid values for 'action'::
5029
5030  #define KVM_PMU_EVENT_ALLOW 0
5031  #define KVM_PMU_EVENT_DENY 1
5032
50334.121 KVM_PPC_SVM_OFF
5034---------------------
5035
5036:Capability: basic
5037:Architectures: powerpc
5038:Type: vm ioctl
5039:Parameters: none
5040:Returns: 0 on successful completion,
5041
5042Errors:
5043
5044  ======     ================================================================
5045  EINVAL     if ultravisor failed to terminate the secure guest
5046  ENOMEM     if hypervisor failed to allocate new radix page tables for guest
5047  ======     ================================================================
5048
5049This ioctl is used to turn off the secure mode of the guest or transition
5050the guest from secure mode to normal mode. This is invoked when the guest
5051is reset. This has no effect if called for a normal guest.
5052
5053This ioctl issues an ultravisor call to terminate the secure guest,
5054unpins the VPA pages and releases all the device pages that are used to
5055track the secure pages by hypervisor.
5056
50574.122 KVM_S390_NORMAL_RESET
5058---------------------------
5059
5060:Capability: KVM_CAP_S390_VCPU_RESETS
5061:Architectures: s390
5062:Type: vcpu ioctl
5063:Parameters: none
5064:Returns: 0
5065
5066This ioctl resets VCPU registers and control structures according to
5067the cpu reset definition in the POP (Principles Of Operation).
5068
50694.123 KVM_S390_INITIAL_RESET
5070----------------------------
5071
5072:Capability: none
5073:Architectures: s390
5074:Type: vcpu ioctl
5075:Parameters: none
5076:Returns: 0
5077
5078This ioctl resets VCPU registers and control structures according to
5079the initial cpu reset definition in the POP. However, the cpu is not
5080put into ESA mode. This reset is a superset of the normal reset.
5081
50824.124 KVM_S390_CLEAR_RESET
5083--------------------------
5084
5085:Capability: KVM_CAP_S390_VCPU_RESETS
5086:Architectures: s390
5087:Type: vcpu ioctl
5088:Parameters: none
5089:Returns: 0
5090
5091This ioctl resets VCPU registers and control structures according to
5092the clear cpu reset definition in the POP. However, the cpu is not put
5093into ESA mode. This reset is a superset of the initial reset.
5094
5095
50964.125 KVM_S390_PV_COMMAND
5097-------------------------
5098
5099:Capability: KVM_CAP_S390_PROTECTED
5100:Architectures: s390
5101:Type: vm ioctl
5102:Parameters: struct kvm_pv_cmd
5103:Returns: 0 on success, < 0 on error
5104
5105::
5106
5107  struct kvm_pv_cmd {
5108	__u32 cmd;	/* Command to be executed */
5109	__u16 rc;	/* Ultravisor return code */
5110	__u16 rrc;	/* Ultravisor return reason code */
5111	__u64 data;	/* Data or address */
5112	__u32 flags;    /* flags for future extensions. Must be 0 for now */
5113	__u32 reserved[3];
5114  };
5115
5116**Ultravisor return codes**
5117The Ultravisor return (reason) codes are provided by the kernel if a
5118Ultravisor call has been executed to achieve the results expected by
5119the command. Therefore they are independent of the IOCTL return
5120code. If KVM changes `rc`, its value will always be greater than 0
5121hence setting it to 0 before issuing a PV command is advised to be
5122able to detect a change of `rc`.
5123
5124**cmd values:**
5125
5126KVM_PV_ENABLE
5127  Allocate memory and register the VM with the Ultravisor, thereby
5128  donating memory to the Ultravisor that will become inaccessible to
5129  KVM. All existing CPUs are converted to protected ones. After this
5130  command has succeeded, any CPU added via hotplug will become
5131  protected during its creation as well.
5132
5133  Errors:
5134
5135  =====      =============================
5136  EINTR      an unmasked signal is pending
5137  =====      =============================
5138
5139KVM_PV_DISABLE
5140  Deregister the VM from the Ultravisor and reclaim the memory that had
5141  been donated to the Ultravisor, making it usable by the kernel again.
5142  All registered VCPUs are converted back to non-protected ones. If a
5143  previous protected VM had been prepared for asynchonous teardown with
5144  KVM_PV_ASYNC_CLEANUP_PREPARE and not subsequently torn down with
5145  KVM_PV_ASYNC_CLEANUP_PERFORM, it will be torn down in this call
5146  together with the current protected VM.
5147
5148KVM_PV_VM_SET_SEC_PARMS
5149  Pass the image header from VM memory to the Ultravisor in
5150  preparation of image unpacking and verification.
5151
5152KVM_PV_VM_UNPACK
5153  Unpack (protect and decrypt) a page of the encrypted boot image.
5154
5155KVM_PV_VM_VERIFY
5156  Verify the integrity of the unpacked image. Only if this succeeds,
5157  KVM is allowed to start protected VCPUs.
5158
5159KVM_PV_INFO
5160  :Capability: KVM_CAP_S390_PROTECTED_DUMP
5161
5162  Presents an API that provides Ultravisor related data to userspace
5163  via subcommands. len_max is the size of the user space buffer,
5164  len_written is KVM's indication of how much bytes of that buffer
5165  were actually written to. len_written can be used to determine the
5166  valid fields if more response fields are added in the future.
5167
5168  ::
5169
5170     enum pv_cmd_info_id {
5171	KVM_PV_INFO_VM,
5172	KVM_PV_INFO_DUMP,
5173     };
5174
5175     struct kvm_s390_pv_info_header {
5176	__u32 id;
5177	__u32 len_max;
5178	__u32 len_written;
5179	__u32 reserved;
5180     };
5181
5182     struct kvm_s390_pv_info {
5183	struct kvm_s390_pv_info_header header;
5184	struct kvm_s390_pv_info_dump dump;
5185	struct kvm_s390_pv_info_vm vm;
5186     };
5187
5188**subcommands:**
5189
5190  KVM_PV_INFO_VM
5191    This subcommand provides basic Ultravisor information for PV
5192    hosts. These values are likely also exported as files in the sysfs
5193    firmware UV query interface but they are more easily available to
5194    programs in this API.
5195
5196    The installed calls and feature_indication members provide the
5197    installed UV calls and the UV's other feature indications.
5198
5199    The max_* members provide information about the maximum number of PV
5200    vcpus, PV guests and PV guest memory size.
5201
5202    ::
5203
5204      struct kvm_s390_pv_info_vm {
5205	__u64 inst_calls_list[4];
5206	__u64 max_cpus;
5207	__u64 max_guests;
5208	__u64 max_guest_addr;
5209	__u64 feature_indication;
5210      };
5211
5212
5213  KVM_PV_INFO_DUMP
5214    This subcommand provides information related to dumping PV guests.
5215
5216    ::
5217
5218      struct kvm_s390_pv_info_dump {
5219	__u64 dump_cpu_buffer_len;
5220	__u64 dump_config_mem_buffer_per_1m;
5221	__u64 dump_config_finalize_len;
5222      };
5223
5224KVM_PV_DUMP
5225  :Capability: KVM_CAP_S390_PROTECTED_DUMP
5226
5227  Presents an API that provides calls which facilitate dumping a
5228  protected VM.
5229
5230  ::
5231
5232    struct kvm_s390_pv_dmp {
5233      __u64 subcmd;
5234      __u64 buff_addr;
5235      __u64 buff_len;
5236      __u64 gaddr;		/* For dump storage state */
5237    };
5238
5239  **subcommands:**
5240
5241  KVM_PV_DUMP_INIT
5242    Initializes the dump process of a protected VM. If this call does
5243    not succeed all other subcommands will fail with -EINVAL. This
5244    subcommand will return -EINVAL if a dump process has not yet been
5245    completed.
5246
5247    Not all PV vms can be dumped, the owner needs to set `dump
5248    allowed` PCF bit 34 in the SE header to allow dumping.
5249
5250  KVM_PV_DUMP_CONFIG_STOR_STATE
5251     Stores `buff_len` bytes of tweak component values starting with
5252     the 1MB block specified by the absolute guest address
5253     (`gaddr`). `buff_len` needs to be `conf_dump_storage_state_len`
5254     aligned and at least >= the `conf_dump_storage_state_len` value
5255     provided by the dump uv_info data. buff_user might be written to
5256     even if an error rc is returned. For instance if we encounter a
5257     fault after writing the first page of data.
5258
5259  KVM_PV_DUMP_COMPLETE
5260    If the subcommand succeeds it completes the dump process and lets
5261    KVM_PV_DUMP_INIT be called again.
5262
5263    On success `conf_dump_finalize_len` bytes of completion data will be
5264    stored to the `buff_addr`. The completion data contains a key
5265    derivation seed, IV, tweak nonce and encryption keys as well as an
5266    authentication tag all of which are needed to decrypt the dump at a
5267    later time.
5268
5269KVM_PV_ASYNC_CLEANUP_PREPARE
5270  :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE
5271
5272  Prepare the current protected VM for asynchronous teardown. Most
5273  resources used by the current protected VM will be set aside for a
5274  subsequent asynchronous teardown. The current protected VM will then
5275  resume execution immediately as non-protected. There can be at most
5276  one protected VM prepared for asynchronous teardown at any time. If
5277  a protected VM had already been prepared for teardown without
5278  subsequently calling KVM_PV_ASYNC_CLEANUP_PERFORM, this call will
5279  fail. In that case, the userspace process should issue a normal
5280  KVM_PV_DISABLE. The resources set aside with this call will need to
5281  be cleaned up with a subsequent call to KVM_PV_ASYNC_CLEANUP_PERFORM
5282  or KVM_PV_DISABLE, otherwise they will be cleaned up when KVM
5283  terminates. KVM_PV_ASYNC_CLEANUP_PREPARE can be called again as soon
5284  as cleanup starts, i.e. before KVM_PV_ASYNC_CLEANUP_PERFORM finishes.
5285
5286KVM_PV_ASYNC_CLEANUP_PERFORM
5287  :Capability: KVM_CAP_S390_PROTECTED_ASYNC_DISABLE
5288
5289  Tear down the protected VM previously prepared for teardown with
5290  KVM_PV_ASYNC_CLEANUP_PREPARE. The resources that had been set aside
5291  will be freed during the execution of this command. This PV command
5292  should ideally be issued by userspace from a separate thread. If a
5293  fatal signal is received (or the process terminates naturally), the
5294  command will terminate immediately without completing, and the normal
5295  KVM shutdown procedure will take care of cleaning up all remaining
5296  protected VMs, including the ones whose teardown was interrupted by
5297  process termination.
5298
52994.126 KVM_XEN_HVM_SET_ATTR
5300--------------------------
5301
5302:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5303:Architectures: x86
5304:Type: vm ioctl
5305:Parameters: struct kvm_xen_hvm_attr
5306:Returns: 0 on success, < 0 on error
5307
5308::
5309
5310  struct kvm_xen_hvm_attr {
5311	__u16 type;
5312	__u16 pad[3];
5313	union {
5314		__u8 long_mode;
5315		__u8 vector;
5316		__u8 runstate_update_flag;
5317		struct {
5318			__u64 gfn;
5319		} shared_info;
5320		struct {
5321			__u32 send_port;
5322			__u32 type; /* EVTCHNSTAT_ipi / EVTCHNSTAT_interdomain */
5323			__u32 flags;
5324			union {
5325				struct {
5326					__u32 port;
5327					__u32 vcpu;
5328					__u32 priority;
5329				} port;
5330				struct {
5331					__u32 port; /* Zero for eventfd */
5332					__s32 fd;
5333				} eventfd;
5334				__u32 padding[4];
5335			} deliver;
5336		} evtchn;
5337		__u32 xen_version;
5338		__u64 pad[8];
5339	} u;
5340  };
5341
5342type values:
5343
5344KVM_XEN_ATTR_TYPE_LONG_MODE
5345  Sets the ABI mode of the VM to 32-bit or 64-bit (long mode). This
5346  determines the layout of the shared info pages exposed to the VM.
5347
5348KVM_XEN_ATTR_TYPE_SHARED_INFO
5349  Sets the guest physical frame number at which the Xen "shared info"
5350  page resides. Note that although Xen places vcpu_info for the first
5351  32 vCPUs in the shared_info page, KVM does not automatically do so
5352  and instead requires that KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO be used
5353  explicitly even when the vcpu_info for a given vCPU resides at the
5354  "default" location in the shared_info page. This is because KVM may
5355  not be aware of the Xen CPU id which is used as the index into the
5356  vcpu_info[] array, so may know the correct default location.
5357
5358  Note that the shared info page may be constantly written to by KVM;
5359  it contains the event channel bitmap used to deliver interrupts to
5360  a Xen guest, amongst other things. It is exempt from dirty tracking
5361  mechanisms — KVM will not explicitly mark the page as dirty each
5362  time an event channel interrupt is delivered to the guest! Thus,
5363  userspace should always assume that the designated GFN is dirty if
5364  any vCPU has been running or any event channel interrupts can be
5365  routed to the guest.
5366
5367  Setting the gfn to KVM_XEN_INVALID_GFN will disable the shared info
5368  page.
5369
5370KVM_XEN_ATTR_TYPE_UPCALL_VECTOR
5371  Sets the exception vector used to deliver Xen event channel upcalls.
5372  This is the HVM-wide vector injected directly by the hypervisor
5373  (not through the local APIC), typically configured by a guest via
5374  HVM_PARAM_CALLBACK_IRQ. This can be disabled again (e.g. for guest
5375  SHUTDOWN_soft_reset) by setting it to zero.
5376
5377KVM_XEN_ATTR_TYPE_EVTCHN
5378  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5379  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5380  an outbound port number for interception of EVTCHNOP_send requests
5381  from the guest. A given sending port number may be directed back to
5382  a specified vCPU (by APIC ID) / port / priority on the guest, or to
5383  trigger events on an eventfd. The vCPU and priority can be changed
5384  by setting KVM_XEN_EVTCHN_UPDATE in a subsequent call, but but other
5385  fields cannot change for a given sending port. A port mapping is
5386  removed by using KVM_XEN_EVTCHN_DEASSIGN in the flags field. Passing
5387  KVM_XEN_EVTCHN_RESET in the flags field removes all interception of
5388  outbound event channels. The values of the flags field are mutually
5389  exclusive and cannot be combined as a bitmask.
5390
5391KVM_XEN_ATTR_TYPE_XEN_VERSION
5392  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5393  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It configures
5394  the 32-bit version code returned to the guest when it invokes the
5395  XENVER_version call; typically (XEN_MAJOR << 16 | XEN_MINOR). PV
5396  Xen guests will often use this to as a dummy hypercall to trigger
5397  event channel delivery, so responding within the kernel without
5398  exiting to userspace is beneficial.
5399
5400KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG
5401  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5402  support for KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG. It enables the
5403  XEN_RUNSTATE_UPDATE flag which allows guest vCPUs to safely read
5404  other vCPUs' vcpu_runstate_info. Xen guests enable this feature via
5405  the VMASST_TYPE_runstate_update_flag of the HYPERVISOR_vm_assist
5406  hypercall.
5407
54084.127 KVM_XEN_HVM_GET_ATTR
5409--------------------------
5410
5411:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5412:Architectures: x86
5413:Type: vm ioctl
5414:Parameters: struct kvm_xen_hvm_attr
5415:Returns: 0 on success, < 0 on error
5416
5417Allows Xen VM attributes to be read. For the structure and types,
5418see KVM_XEN_HVM_SET_ATTR above. The KVM_XEN_ATTR_TYPE_EVTCHN
5419attribute cannot be read.
5420
54214.128 KVM_XEN_VCPU_SET_ATTR
5422---------------------------
5423
5424:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5425:Architectures: x86
5426:Type: vcpu ioctl
5427:Parameters: struct kvm_xen_vcpu_attr
5428:Returns: 0 on success, < 0 on error
5429
5430::
5431
5432  struct kvm_xen_vcpu_attr {
5433	__u16 type;
5434	__u16 pad[3];
5435	union {
5436		__u64 gpa;
5437		__u64 pad[4];
5438		struct {
5439			__u64 state;
5440			__u64 state_entry_time;
5441			__u64 time_running;
5442			__u64 time_runnable;
5443			__u64 time_blocked;
5444			__u64 time_offline;
5445		} runstate;
5446		__u32 vcpu_id;
5447		struct {
5448			__u32 port;
5449			__u32 priority;
5450			__u64 expires_ns;
5451		} timer;
5452		__u8 vector;
5453	} u;
5454  };
5455
5456type values:
5457
5458KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO
5459  Sets the guest physical address of the vcpu_info for a given vCPU.
5460  As with the shared_info page for the VM, the corresponding page may be
5461  dirtied at any time if event channel interrupt delivery is enabled, so
5462  userspace should always assume that the page is dirty without relying
5463  on dirty logging. Setting the gpa to KVM_XEN_INVALID_GPA will disable
5464  the vcpu_info.
5465
5466KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO
5467  Sets the guest physical address of an additional pvclock structure
5468  for a given vCPU. This is typically used for guest vsyscall support.
5469  Setting the gpa to KVM_XEN_INVALID_GPA will disable the structure.
5470
5471KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR
5472  Sets the guest physical address of the vcpu_runstate_info for a given
5473  vCPU. This is how a Xen guest tracks CPU state such as steal time.
5474  Setting the gpa to KVM_XEN_INVALID_GPA will disable the runstate area.
5475
5476KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT
5477  Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of
5478  the given vCPU from the .u.runstate.state member of the structure.
5479  KVM automatically accounts running and runnable time but blocked
5480  and offline states are only entered explicitly.
5481
5482KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA
5483  Sets all fields of the vCPU runstate data from the .u.runstate member
5484  of the structure, including the current runstate. The state_entry_time
5485  must equal the sum of the other four times.
5486
5487KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST
5488  This *adds* the contents of the .u.runstate members of the structure
5489  to the corresponding members of the given vCPU's runstate data, thus
5490  permitting atomic adjustments to the runstate times. The adjustment
5491  to the state_entry_time must equal the sum of the adjustments to the
5492  other four times. The state field must be set to -1, or to a valid
5493  runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked
5494  or RUNSTATE_offline) to set the current accounted state as of the
5495  adjusted state_entry_time.
5496
5497KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID
5498  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5499  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the Xen
5500  vCPU ID of the given vCPU, to allow timer-related VCPU operations to
5501  be intercepted by KVM.
5502
5503KVM_XEN_VCPU_ATTR_TYPE_TIMER
5504  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5505  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5506  event channel port/priority for the VIRQ_TIMER of the vCPU, as well
5507  as allowing a pending timer to be saved/restored. Setting the timer
5508  port to zero disables kernel handling of the singleshot timer.
5509
5510KVM_XEN_VCPU_ATTR_TYPE_UPCALL_VECTOR
5511  This attribute is available when the KVM_CAP_XEN_HVM ioctl indicates
5512  support for KVM_XEN_HVM_CONFIG_EVTCHN_SEND features. It sets the
5513  per-vCPU local APIC upcall vector, configured by a Xen guest with
5514  the HVMOP_set_evtchn_upcall_vector hypercall. This is typically
5515  used by Windows guests, and is distinct from the HVM-wide upcall
5516  vector configured with HVM_PARAM_CALLBACK_IRQ. It is disabled by
5517  setting the vector to zero.
5518
5519
55204.129 KVM_XEN_VCPU_GET_ATTR
5521---------------------------
5522
5523:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO
5524:Architectures: x86
5525:Type: vcpu ioctl
5526:Parameters: struct kvm_xen_vcpu_attr
5527:Returns: 0 on success, < 0 on error
5528
5529Allows Xen vCPU attributes to be read. For the structure and types,
5530see KVM_XEN_VCPU_SET_ATTR above.
5531
5532The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used
5533with the KVM_XEN_VCPU_GET_ATTR ioctl.
5534
55354.130 KVM_ARM_MTE_COPY_TAGS
5536---------------------------
5537
5538:Capability: KVM_CAP_ARM_MTE
5539:Architectures: arm64
5540:Type: vm ioctl
5541:Parameters: struct kvm_arm_copy_mte_tags
5542:Returns: number of bytes copied, < 0 on error (-EINVAL for incorrect
5543          arguments, -EFAULT if memory cannot be accessed).
5544
5545::
5546
5547  struct kvm_arm_copy_mte_tags {
5548	__u64 guest_ipa;
5549	__u64 length;
5550	void __user *addr;
5551	__u64 flags;
5552	__u64 reserved[2];
5553  };
5554
5555Copies Memory Tagging Extension (MTE) tags to/from guest tag memory. The
5556``guest_ipa`` and ``length`` fields must be ``PAGE_SIZE`` aligned. The ``addr``
5557field must point to a buffer which the tags will be copied to or from.
5558
5559``flags`` specifies the direction of copy, either ``KVM_ARM_TAGS_TO_GUEST`` or
5560``KVM_ARM_TAGS_FROM_GUEST``.
5561
5562The size of the buffer to store the tags is ``(length / 16)`` bytes
5563(granules in MTE are 16 bytes long). Each byte contains a single tag
5564value. This matches the format of ``PTRACE_PEEKMTETAGS`` and
5565``PTRACE_POKEMTETAGS``.
5566
5567If an error occurs before any data is copied then a negative error code is
5568returned. If some tags have been copied before an error occurs then the number
5569of bytes successfully copied is returned. If the call completes successfully
5570then ``length`` is returned.
5571
55724.131 KVM_GET_SREGS2
5573--------------------
5574
5575:Capability: KVM_CAP_SREGS2
5576:Architectures: x86
5577:Type: vcpu ioctl
5578:Parameters: struct kvm_sregs2 (out)
5579:Returns: 0 on success, -1 on error
5580
5581Reads special registers from the vcpu.
5582This ioctl (when supported) replaces the KVM_GET_SREGS.
5583
5584::
5585
5586        struct kvm_sregs2 {
5587                /* out (KVM_GET_SREGS2) / in (KVM_SET_SREGS2) */
5588                struct kvm_segment cs, ds, es, fs, gs, ss;
5589                struct kvm_segment tr, ldt;
5590                struct kvm_dtable gdt, idt;
5591                __u64 cr0, cr2, cr3, cr4, cr8;
5592                __u64 efer;
5593                __u64 apic_base;
5594                __u64 flags;
5595                __u64 pdptrs[4];
5596        };
5597
5598flags values for ``kvm_sregs2``:
5599
5600``KVM_SREGS2_FLAGS_PDPTRS_VALID``
5601
5602  Indicates thats the struct contain valid PDPTR values.
5603
5604
56054.132 KVM_SET_SREGS2
5606--------------------
5607
5608:Capability: KVM_CAP_SREGS2
5609:Architectures: x86
5610:Type: vcpu ioctl
5611:Parameters: struct kvm_sregs2 (in)
5612:Returns: 0 on success, -1 on error
5613
5614Writes special registers into the vcpu.
5615See KVM_GET_SREGS2 for the data structures.
5616This ioctl (when supported) replaces the KVM_SET_SREGS.
5617
56184.133 KVM_GET_STATS_FD
5619----------------------
5620
5621:Capability: KVM_CAP_STATS_BINARY_FD
5622:Architectures: all
5623:Type: vm ioctl, vcpu ioctl
5624:Parameters: none
5625:Returns: statistics file descriptor on success, < 0 on error
5626
5627Errors:
5628
5629  ======     ======================================================
5630  ENOMEM     if the fd could not be created due to lack of memory
5631  EMFILE     if the number of opened files exceeds the limit
5632  ======     ======================================================
5633
5634The returned file descriptor can be used to read VM/vCPU statistics data in
5635binary format. The data in the file descriptor consists of four blocks
5636organized as follows:
5637
5638+-------------+
5639|   Header    |
5640+-------------+
5641|  id string  |
5642+-------------+
5643| Descriptors |
5644+-------------+
5645| Stats Data  |
5646+-------------+
5647
5648Apart from the header starting at offset 0, please be aware that it is
5649not guaranteed that the four blocks are adjacent or in the above order;
5650the offsets of the id, descriptors and data blocks are found in the
5651header.  However, all four blocks are aligned to 64 bit offsets in the
5652file and they do not overlap.
5653
5654All blocks except the data block are immutable.  Userspace can read them
5655only one time after retrieving the file descriptor, and then use ``pread`` or
5656``lseek`` to read the statistics repeatedly.
5657
5658All data is in system endianness.
5659
5660The format of the header is as follows::
5661
5662	struct kvm_stats_header {
5663		__u32 flags;
5664		__u32 name_size;
5665		__u32 num_desc;
5666		__u32 id_offset;
5667		__u32 desc_offset;
5668		__u32 data_offset;
5669	};
5670
5671The ``flags`` field is not used at the moment. It is always read as 0.
5672
5673The ``name_size`` field is the size (in byte) of the statistics name string
5674(including trailing '\0') which is contained in the "id string" block and
5675appended at the end of every descriptor.
5676
5677The ``num_desc`` field is the number of descriptors that are included in the
5678descriptor block.  (The actual number of values in the data block may be
5679larger, since each descriptor may comprise more than one value).
5680
5681The ``id_offset`` field is the offset of the id string from the start of the
5682file indicated by the file descriptor. It is a multiple of 8.
5683
5684The ``desc_offset`` field is the offset of the Descriptors block from the start
5685of the file indicated by the file descriptor. It is a multiple of 8.
5686
5687The ``data_offset`` field is the offset of the Stats Data block from the start
5688of the file indicated by the file descriptor. It is a multiple of 8.
5689
5690The id string block contains a string which identifies the file descriptor on
5691which KVM_GET_STATS_FD was invoked.  The size of the block, including the
5692trailing ``'\0'``, is indicated by the ``name_size`` field in the header.
5693
5694The descriptors block is only needed to be read once for the lifetime of the
5695file descriptor contains a sequence of ``struct kvm_stats_desc``, each followed
5696by a string of size ``name_size``.
5697::
5698
5699	#define KVM_STATS_TYPE_SHIFT		0
5700	#define KVM_STATS_TYPE_MASK		(0xF << KVM_STATS_TYPE_SHIFT)
5701	#define KVM_STATS_TYPE_CUMULATIVE	(0x0 << KVM_STATS_TYPE_SHIFT)
5702	#define KVM_STATS_TYPE_INSTANT		(0x1 << KVM_STATS_TYPE_SHIFT)
5703	#define KVM_STATS_TYPE_PEAK		(0x2 << KVM_STATS_TYPE_SHIFT)
5704	#define KVM_STATS_TYPE_LINEAR_HIST	(0x3 << KVM_STATS_TYPE_SHIFT)
5705	#define KVM_STATS_TYPE_LOG_HIST		(0x4 << KVM_STATS_TYPE_SHIFT)
5706	#define KVM_STATS_TYPE_MAX		KVM_STATS_TYPE_LOG_HIST
5707
5708	#define KVM_STATS_UNIT_SHIFT		4
5709	#define KVM_STATS_UNIT_MASK		(0xF << KVM_STATS_UNIT_SHIFT)
5710	#define KVM_STATS_UNIT_NONE		(0x0 << KVM_STATS_UNIT_SHIFT)
5711	#define KVM_STATS_UNIT_BYTES		(0x1 << KVM_STATS_UNIT_SHIFT)
5712	#define KVM_STATS_UNIT_SECONDS		(0x2 << KVM_STATS_UNIT_SHIFT)
5713	#define KVM_STATS_UNIT_CYCLES		(0x3 << KVM_STATS_UNIT_SHIFT)
5714	#define KVM_STATS_UNIT_BOOLEAN		(0x4 << KVM_STATS_UNIT_SHIFT)
5715	#define KVM_STATS_UNIT_MAX		KVM_STATS_UNIT_BOOLEAN
5716
5717	#define KVM_STATS_BASE_SHIFT		8
5718	#define KVM_STATS_BASE_MASK		(0xF << KVM_STATS_BASE_SHIFT)
5719	#define KVM_STATS_BASE_POW10		(0x0 << KVM_STATS_BASE_SHIFT)
5720	#define KVM_STATS_BASE_POW2		(0x1 << KVM_STATS_BASE_SHIFT)
5721	#define KVM_STATS_BASE_MAX		KVM_STATS_BASE_POW2
5722
5723	struct kvm_stats_desc {
5724		__u32 flags;
5725		__s16 exponent;
5726		__u16 size;
5727		__u32 offset;
5728		__u32 bucket_size;
5729		char name[];
5730	};
5731
5732The ``flags`` field contains the type and unit of the statistics data described
5733by this descriptor. Its endianness is CPU native.
5734The following flags are supported:
5735
5736Bits 0-3 of ``flags`` encode the type:
5737
5738  * ``KVM_STATS_TYPE_CUMULATIVE``
5739    The statistics reports a cumulative count. The value of data can only be increased.
5740    Most of the counters used in KVM are of this type.
5741    The corresponding ``size`` field for this type is always 1.
5742    All cumulative statistics data are read/write.
5743  * ``KVM_STATS_TYPE_INSTANT``
5744    The statistics reports an instantaneous value. Its value can be increased or
5745    decreased. This type is usually used as a measurement of some resources,
5746    like the number of dirty pages, the number of large pages, etc.
5747    All instant statistics are read only.
5748    The corresponding ``size`` field for this type is always 1.
5749  * ``KVM_STATS_TYPE_PEAK``
5750    The statistics data reports a peak value, for example the maximum number
5751    of items in a hash table bucket, the longest time waited and so on.
5752    The value of data can only be increased.
5753    The corresponding ``size`` field for this type is always 1.
5754  * ``KVM_STATS_TYPE_LINEAR_HIST``
5755    The statistic is reported as a linear histogram. The number of
5756    buckets is specified by the ``size`` field. The size of buckets is specified
5757    by the ``hist_param`` field. The range of the Nth bucket (1 <= N < ``size``)
5758    is [``hist_param``*(N-1), ``hist_param``*N), while the range of the last
5759    bucket is [``hist_param``*(``size``-1), +INF). (+INF means positive infinity
5760    value.)
5761  * ``KVM_STATS_TYPE_LOG_HIST``
5762    The statistic is reported as a logarithmic histogram. The number of
5763    buckets is specified by the ``size`` field. The range of the first bucket is
5764    [0, 1), while the range of the last bucket is [pow(2, ``size``-2), +INF).
5765    Otherwise, The Nth bucket (1 < N < ``size``) covers
5766    [pow(2, N-2), pow(2, N-1)).
5767
5768Bits 4-7 of ``flags`` encode the unit:
5769
5770  * ``KVM_STATS_UNIT_NONE``
5771    There is no unit for the value of statistics data. This usually means that
5772    the value is a simple counter of an event.
5773  * ``KVM_STATS_UNIT_BYTES``
5774    It indicates that the statistics data is used to measure memory size, in the
5775    unit of Byte, KiByte, MiByte, GiByte, etc. The unit of the data is
5776    determined by the ``exponent`` field in the descriptor.
5777  * ``KVM_STATS_UNIT_SECONDS``
5778    It indicates that the statistics data is used to measure time or latency.
5779  * ``KVM_STATS_UNIT_CYCLES``
5780    It indicates that the statistics data is used to measure CPU clock cycles.
5781  * ``KVM_STATS_UNIT_BOOLEAN``
5782    It indicates that the statistic will always be either 0 or 1.  Boolean
5783    statistics of "peak" type will never go back from 1 to 0.  Boolean
5784    statistics can be linear histograms (with two buckets) but not logarithmic
5785    histograms.
5786
5787Note that, in the case of histograms, the unit applies to the bucket
5788ranges, while the bucket value indicates how many samples fell in the
5789bucket's range.
5790
5791Bits 8-11 of ``flags``, together with ``exponent``, encode the scale of the
5792unit:
5793
5794  * ``KVM_STATS_BASE_POW10``
5795    The scale is based on power of 10. It is used for measurement of time and
5796    CPU clock cycles.  For example, an exponent of -9 can be used with
5797    ``KVM_STATS_UNIT_SECONDS`` to express that the unit is nanoseconds.
5798  * ``KVM_STATS_BASE_POW2``
5799    The scale is based on power of 2. It is used for measurement of memory size.
5800    For example, an exponent of 20 can be used with ``KVM_STATS_UNIT_BYTES`` to
5801    express that the unit is MiB.
5802
5803The ``size`` field is the number of values of this statistics data. Its
5804value is usually 1 for most of simple statistics. 1 means it contains an
5805unsigned 64bit data.
5806
5807The ``offset`` field is the offset from the start of Data Block to the start of
5808the corresponding statistics data.
5809
5810The ``bucket_size`` field is used as a parameter for histogram statistics data.
5811It is only used by linear histogram statistics data, specifying the size of a
5812bucket in the unit expressed by bits 4-11 of ``flags`` together with ``exponent``.
5813
5814The ``name`` field is the name string of the statistics data. The name string
5815starts at the end of ``struct kvm_stats_desc``.  The maximum length including
5816the trailing ``'\0'``, is indicated by ``name_size`` in the header.
5817
5818The Stats Data block contains an array of 64-bit values in the same order
5819as the descriptors in Descriptors block.
5820
58214.134 KVM_GET_XSAVE2
5822--------------------
5823
5824:Capability: KVM_CAP_XSAVE2
5825:Architectures: x86
5826:Type: vcpu ioctl
5827:Parameters: struct kvm_xsave (out)
5828:Returns: 0 on success, -1 on error
5829
5830
5831::
5832
5833  struct kvm_xsave {
5834	__u32 region[1024];
5835	__u32 extra[0];
5836  };
5837
5838This ioctl would copy current vcpu's xsave struct to the userspace. It
5839copies as many bytes as are returned by KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2)
5840when invoked on the vm file descriptor. The size value returned by
5841KVM_CHECK_EXTENSION(KVM_CAP_XSAVE2) will always be at least 4096.
5842Currently, it is only greater than 4096 if a dynamic feature has been
5843enabled with ``arch_prctl()``, but this may change in the future.
5844
5845The offsets of the state save areas in struct kvm_xsave follow the contents
5846of CPUID leaf 0xD on the host.
5847
58484.135 KVM_XEN_HVM_EVTCHN_SEND
5849-----------------------------
5850
5851:Capability: KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_EVTCHN_SEND
5852:Architectures: x86
5853:Type: vm ioctl
5854:Parameters: struct kvm_irq_routing_xen_evtchn
5855:Returns: 0 on success, < 0 on error
5856
5857
5858::
5859
5860   struct kvm_irq_routing_xen_evtchn {
5861	__u32 port;
5862	__u32 vcpu;
5863	__u32 priority;
5864   };
5865
5866This ioctl injects an event channel interrupt directly to the guest vCPU.
5867
58684.136 KVM_S390_PV_CPU_COMMAND
5869-----------------------------
5870
5871:Capability: KVM_CAP_S390_PROTECTED_DUMP
5872:Architectures: s390
5873:Type: vcpu ioctl
5874:Parameters: none
5875:Returns: 0 on success, < 0 on error
5876
5877This ioctl closely mirrors `KVM_S390_PV_COMMAND` but handles requests
5878for vcpus. It re-uses the kvm_s390_pv_dmp struct and hence also shares
5879the command ids.
5880
5881**command:**
5882
5883KVM_PV_DUMP
5884  Presents an API that provides calls which facilitate dumping a vcpu
5885  of a protected VM.
5886
5887**subcommand:**
5888
5889KVM_PV_DUMP_CPU
5890  Provides encrypted dump data like register values.
5891  The length of the returned data is provided by uv_info.guest_cpu_stor_len.
5892
58934.137 KVM_S390_ZPCI_OP
5894----------------------
5895
5896:Capability: KVM_CAP_S390_ZPCI_OP
5897:Architectures: s390
5898:Type: vm ioctl
5899:Parameters: struct kvm_s390_zpci_op (in)
5900:Returns: 0 on success, <0 on error
5901
5902Used to manage hardware-assisted virtualization features for zPCI devices.
5903
5904Parameters are specified via the following structure::
5905
5906  struct kvm_s390_zpci_op {
5907	/* in */
5908	__u32 fh;		/* target device */
5909	__u8  op;		/* operation to perform */
5910	__u8  pad[3];
5911	union {
5912		/* for KVM_S390_ZPCIOP_REG_AEN */
5913		struct {
5914			__u64 ibv;	/* Guest addr of interrupt bit vector */
5915			__u64 sb;	/* Guest addr of summary bit */
5916			__u32 flags;
5917			__u32 noi;	/* Number of interrupts */
5918			__u8 isc;	/* Guest interrupt subclass */
5919			__u8 sbo;	/* Offset of guest summary bit vector */
5920			__u16 pad;
5921		} reg_aen;
5922		__u64 reserved[8];
5923	} u;
5924  };
5925
5926The type of operation is specified in the "op" field.
5927KVM_S390_ZPCIOP_REG_AEN is used to register the VM for adapter event
5928notification interpretation, which will allow firmware delivery of adapter
5929events directly to the vm, with KVM providing a backup delivery mechanism;
5930KVM_S390_ZPCIOP_DEREG_AEN is used to subsequently disable interpretation of
5931adapter event notifications.
5932
5933The target zPCI function must also be specified via the "fh" field.  For the
5934KVM_S390_ZPCIOP_REG_AEN operation, additional information to establish firmware
5935delivery must be provided via the "reg_aen" struct.
5936
5937The "pad" and "reserved" fields may be used for future extensions and should be
5938set to 0s by userspace.
5939
59405. The kvm_run structure
5941========================
5942
5943Application code obtains a pointer to the kvm_run structure by
5944mmap()ing a vcpu fd.  From that point, application code can control
5945execution by changing fields in kvm_run prior to calling the KVM_RUN
5946ioctl, and obtain information about the reason KVM_RUN returned by
5947looking up structure members.
5948
5949::
5950
5951  struct kvm_run {
5952	/* in */
5953	__u8 request_interrupt_window;
5954
5955Request that KVM_RUN return when it becomes possible to inject external
5956interrupts into the guest.  Useful in conjunction with KVM_INTERRUPT.
5957
5958::
5959
5960	__u8 immediate_exit;
5961
5962This field is polled once when KVM_RUN starts; if non-zero, KVM_RUN
5963exits immediately, returning -EINTR.  In the common scenario where a
5964signal is used to "kick" a VCPU out of KVM_RUN, this field can be used
5965to avoid usage of KVM_SET_SIGNAL_MASK, which has worse scalability.
5966Rather than blocking the signal outside KVM_RUN, userspace can set up
5967a signal handler that sets run->immediate_exit to a non-zero value.
5968
5969This field is ignored if KVM_CAP_IMMEDIATE_EXIT is not available.
5970
5971::
5972
5973	__u8 padding1[6];
5974
5975	/* out */
5976	__u32 exit_reason;
5977
5978When KVM_RUN has returned successfully (return value 0), this informs
5979application code why KVM_RUN has returned.  Allowable values for this
5980field are detailed below.
5981
5982::
5983
5984	__u8 ready_for_interrupt_injection;
5985
5986If request_interrupt_window has been specified, this field indicates
5987an interrupt can be injected now with KVM_INTERRUPT.
5988
5989::
5990
5991	__u8 if_flag;
5992
5993The value of the current interrupt flag.  Only valid if in-kernel
5994local APIC is not used.
5995
5996::
5997
5998	__u16 flags;
5999
6000More architecture-specific flags detailing state of the VCPU that may
6001affect the device's behavior. Current defined flags::
6002
6003  /* x86, set if the VCPU is in system management mode */
6004  #define KVM_RUN_X86_SMM     (1 << 0)
6005  /* x86, set if bus lock detected in VM */
6006  #define KVM_RUN_BUS_LOCK    (1 << 1)
6007  /* arm64, set for KVM_EXIT_DEBUG */
6008  #define KVM_DEBUG_ARCH_HSR_HIGH_VALID  (1 << 0)
6009
6010::
6011
6012	/* in (pre_kvm_run), out (post_kvm_run) */
6013	__u64 cr8;
6014
6015The value of the cr8 register.  Only valid if in-kernel local APIC is
6016not used.  Both input and output.
6017
6018::
6019
6020	__u64 apic_base;
6021
6022The value of the APIC BASE msr.  Only valid if in-kernel local
6023APIC is not used.  Both input and output.
6024
6025::
6026
6027	union {
6028		/* KVM_EXIT_UNKNOWN */
6029		struct {
6030			__u64 hardware_exit_reason;
6031		} hw;
6032
6033If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
6034reasons.  Further architecture-specific information is available in
6035hardware_exit_reason.
6036
6037::
6038
6039		/* KVM_EXIT_FAIL_ENTRY */
6040		struct {
6041			__u64 hardware_entry_failure_reason;
6042			__u32 cpu; /* if KVM_LAST_CPU */
6043		} fail_entry;
6044
6045If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
6046to unknown reasons.  Further architecture-specific information is
6047available in hardware_entry_failure_reason.
6048
6049::
6050
6051		/* KVM_EXIT_EXCEPTION */
6052		struct {
6053			__u32 exception;
6054			__u32 error_code;
6055		} ex;
6056
6057Unused.
6058
6059::
6060
6061		/* KVM_EXIT_IO */
6062		struct {
6063  #define KVM_EXIT_IO_IN  0
6064  #define KVM_EXIT_IO_OUT 1
6065			__u8 direction;
6066			__u8 size; /* bytes */
6067			__u16 port;
6068			__u32 count;
6069			__u64 data_offset; /* relative to kvm_run start */
6070		} io;
6071
6072If exit_reason is KVM_EXIT_IO, then the vcpu has
6073executed a port I/O instruction which could not be satisfied by kvm.
6074data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
6075where kvm expects application code to place the data for the next
6076KVM_RUN invocation (KVM_EXIT_IO_IN).  Data format is a packed array.
6077
6078::
6079
6080		/* KVM_EXIT_DEBUG */
6081		struct {
6082			struct kvm_debug_exit_arch arch;
6083		} debug;
6084
6085If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
6086for which architecture specific information is returned.
6087
6088::
6089
6090		/* KVM_EXIT_MMIO */
6091		struct {
6092			__u64 phys_addr;
6093			__u8  data[8];
6094			__u32 len;
6095			__u8  is_write;
6096		} mmio;
6097
6098If exit_reason is KVM_EXIT_MMIO, then the vcpu has
6099executed a memory-mapped I/O instruction which could not be satisfied
6100by kvm.  The 'data' member contains the written data if 'is_write' is
6101true, and should be filled by application code otherwise.
6102
6103The 'data' member contains, in its first 'len' bytes, the value as it would
6104appear if the VCPU performed a load or store of the appropriate width directly
6105to the byte array.
6106
6107.. note::
6108
6109      For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR, KVM_EXIT_XEN,
6110      KVM_EXIT_EPR, KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR the corresponding
6111      operations are complete (and guest state is consistent) only after userspace
6112      has re-entered the kernel with KVM_RUN.  The kernel side will first finish
6113      incomplete operations and then check for pending signals.
6114
6115      The pending state of the operation is not preserved in state which is
6116      visible to userspace, thus userspace should ensure that the operation is
6117      completed before performing a live migration.  Userspace can re-enter the
6118      guest with an unmasked signal pending or with the immediate_exit field set
6119      to complete pending operations without allowing any further instructions
6120      to be executed.
6121
6122::
6123
6124		/* KVM_EXIT_HYPERCALL */
6125		struct {
6126			__u64 nr;
6127			__u64 args[6];
6128			__u64 ret;
6129			__u32 longmode;
6130			__u32 pad;
6131		} hypercall;
6132
6133Unused.  This was once used for 'hypercall to userspace'.  To implement
6134such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
6135
6136.. note:: KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
6137
6138::
6139
6140		/* KVM_EXIT_TPR_ACCESS */
6141		struct {
6142			__u64 rip;
6143			__u32 is_write;
6144			__u32 pad;
6145		} tpr_access;
6146
6147To be documented (KVM_TPR_ACCESS_REPORTING).
6148
6149::
6150
6151		/* KVM_EXIT_S390_SIEIC */
6152		struct {
6153			__u8 icptcode;
6154			__u64 mask; /* psw upper half */
6155			__u64 addr; /* psw lower half */
6156			__u16 ipa;
6157			__u32 ipb;
6158		} s390_sieic;
6159
6160s390 specific.
6161
6162::
6163
6164		/* KVM_EXIT_S390_RESET */
6165  #define KVM_S390_RESET_POR       1
6166  #define KVM_S390_RESET_CLEAR     2
6167  #define KVM_S390_RESET_SUBSYSTEM 4
6168  #define KVM_S390_RESET_CPU_INIT  8
6169  #define KVM_S390_RESET_IPL       16
6170		__u64 s390_reset_flags;
6171
6172s390 specific.
6173
6174::
6175
6176		/* KVM_EXIT_S390_UCONTROL */
6177		struct {
6178			__u64 trans_exc_code;
6179			__u32 pgm_code;
6180		} s390_ucontrol;
6181
6182s390 specific. A page fault has occurred for a user controlled virtual
6183machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
6184resolved by the kernel.
6185The program code and the translation exception code that were placed
6186in the cpu's lowcore are presented here as defined by the z Architecture
6187Principles of Operation Book in the Chapter for Dynamic Address Translation
6188(DAT)
6189
6190::
6191
6192		/* KVM_EXIT_DCR */
6193		struct {
6194			__u32 dcrn;
6195			__u32 data;
6196			__u8  is_write;
6197		} dcr;
6198
6199Deprecated - was used for 440 KVM.
6200
6201::
6202
6203		/* KVM_EXIT_OSI */
6204		struct {
6205			__u64 gprs[32];
6206		} osi;
6207
6208MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
6209hypercalls and exit with this exit struct that contains all the guest gprs.
6210
6211If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
6212Userspace can now handle the hypercall and when it's done modify the gprs as
6213necessary. Upon guest entry all guest GPRs will then be replaced by the values
6214in this struct.
6215
6216::
6217
6218		/* KVM_EXIT_PAPR_HCALL */
6219		struct {
6220			__u64 nr;
6221			__u64 ret;
6222			__u64 args[9];
6223		} papr_hcall;
6224
6225This is used on 64-bit PowerPC when emulating a pSeries partition,
6226e.g. with the 'pseries' machine type in qemu.  It occurs when the
6227guest does a hypercall using the 'sc 1' instruction.  The 'nr' field
6228contains the hypercall number (from the guest R3), and 'args' contains
6229the arguments (from the guest R4 - R12).  Userspace should put the
6230return code in 'ret' and any extra returned values in args[].
6231The possible hypercalls are defined in the Power Architecture Platform
6232Requirements (PAPR) document available from www.power.org (free
6233developer registration required to access it).
6234
6235::
6236
6237		/* KVM_EXIT_S390_TSCH */
6238		struct {
6239			__u16 subchannel_id;
6240			__u16 subchannel_nr;
6241			__u32 io_int_parm;
6242			__u32 io_int_word;
6243			__u32 ipb;
6244			__u8 dequeued;
6245		} s390_tsch;
6246
6247s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
6248and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
6249interrupt for the target subchannel has been dequeued and subchannel_id,
6250subchannel_nr, io_int_parm and io_int_word contain the parameters for that
6251interrupt. ipb is needed for instruction parameter decoding.
6252
6253::
6254
6255		/* KVM_EXIT_EPR */
6256		struct {
6257			__u32 epr;
6258		} epr;
6259
6260On FSL BookE PowerPC chips, the interrupt controller has a fast patch
6261interrupt acknowledge path to the core. When the core successfully
6262delivers an interrupt, it automatically populates the EPR register with
6263the interrupt vector number and acknowledges the interrupt inside
6264the interrupt controller.
6265
6266In case the interrupt controller lives in user space, we need to do
6267the interrupt acknowledge cycle through it to fetch the next to be
6268delivered interrupt vector using this exit.
6269
6270It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
6271external interrupt has just been delivered into the guest. User space
6272should put the acknowledged interrupt vector into the 'epr' field.
6273
6274::
6275
6276		/* KVM_EXIT_SYSTEM_EVENT */
6277		struct {
6278  #define KVM_SYSTEM_EVENT_SHUTDOWN       1
6279  #define KVM_SYSTEM_EVENT_RESET          2
6280  #define KVM_SYSTEM_EVENT_CRASH          3
6281  #define KVM_SYSTEM_EVENT_WAKEUP         4
6282  #define KVM_SYSTEM_EVENT_SUSPEND        5
6283  #define KVM_SYSTEM_EVENT_SEV_TERM       6
6284			__u32 type;
6285                        __u32 ndata;
6286                        __u64 data[16];
6287		} system_event;
6288
6289If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
6290a system-level event using some architecture specific mechanism (hypercall
6291or some special instruction). In case of ARM64, this is triggered using
6292HVC instruction based PSCI call from the vcpu.
6293
6294The 'type' field describes the system-level event type.
6295Valid values for 'type' are:
6296
6297 - KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
6298   VM. Userspace is not obliged to honour this, and if it does honour
6299   this does not need to destroy the VM synchronously (ie it may call
6300   KVM_RUN again before shutdown finally occurs).
6301 - KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
6302   As with SHUTDOWN, userspace can choose to ignore the request, or
6303   to schedule the reset to occur in the future and may call KVM_RUN again.
6304 - KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
6305   has requested a crash condition maintenance. Userspace can choose
6306   to ignore the request, or to gather VM memory core dump and/or
6307   reset/shutdown of the VM.
6308 - KVM_SYSTEM_EVENT_SEV_TERM -- an AMD SEV guest requested termination.
6309   The guest physical address of the guest's GHCB is stored in `data[0]`.
6310 - KVM_SYSTEM_EVENT_WAKEUP -- the exiting vCPU is in a suspended state and
6311   KVM has recognized a wakeup event. Userspace may honor this event by
6312   marking the exiting vCPU as runnable, or deny it and call KVM_RUN again.
6313 - KVM_SYSTEM_EVENT_SUSPEND -- the guest has requested a suspension of
6314   the VM.
6315
6316If KVM_CAP_SYSTEM_EVENT_DATA is present, the 'data' field can contain
6317architecture specific information for the system-level event.  Only
6318the first `ndata` items (possibly zero) of the data array are valid.
6319
6320 - for arm64, data[0] is set to KVM_SYSTEM_EVENT_RESET_FLAG_PSCI_RESET2 if
6321   the guest issued a SYSTEM_RESET2 call according to v1.1 of the PSCI
6322   specification.
6323
6324 - for RISC-V, data[0] is set to the value of the second argument of the
6325   ``sbi_system_reset`` call.
6326
6327Previous versions of Linux defined a `flags` member in this struct.  The
6328field is now aliased to `data[0]`.  Userspace can assume that it is only
6329written if ndata is greater than 0.
6330
6331For arm/arm64:
6332--------------
6333
6334KVM_SYSTEM_EVENT_SUSPEND exits are enabled with the
6335KVM_CAP_ARM_SYSTEM_SUSPEND VM capability. If a guest invokes the PSCI
6336SYSTEM_SUSPEND function, KVM will exit to userspace with this event
6337type.
6338
6339It is the sole responsibility of userspace to implement the PSCI
6340SYSTEM_SUSPEND call according to ARM DEN0022D.b 5.19 "SYSTEM_SUSPEND".
6341KVM does not change the vCPU's state before exiting to userspace, so
6342the call parameters are left in-place in the vCPU registers.
6343
6344Userspace is _required_ to take action for such an exit. It must
6345either:
6346
6347 - Honor the guest request to suspend the VM. Userspace can request
6348   in-kernel emulation of suspension by setting the calling vCPU's
6349   state to KVM_MP_STATE_SUSPENDED. Userspace must configure the vCPU's
6350   state according to the parameters passed to the PSCI function when
6351   the calling vCPU is resumed. See ARM DEN0022D.b 5.19.1 "Intended use"
6352   for details on the function parameters.
6353
6354 - Deny the guest request to suspend the VM. See ARM DEN0022D.b 5.19.2
6355   "Caller responsibilities" for possible return values.
6356
6357::
6358
6359		/* KVM_EXIT_IOAPIC_EOI */
6360		struct {
6361			__u8 vector;
6362		} eoi;
6363
6364Indicates that the VCPU's in-kernel local APIC received an EOI for a
6365level-triggered IOAPIC interrupt.  This exit only triggers when the
6366IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
6367the userspace IOAPIC should process the EOI and retrigger the interrupt if
6368it is still asserted.  Vector is the LAPIC interrupt vector for which the
6369EOI was received.
6370
6371::
6372
6373		struct kvm_hyperv_exit {
6374  #define KVM_EXIT_HYPERV_SYNIC          1
6375  #define KVM_EXIT_HYPERV_HCALL          2
6376  #define KVM_EXIT_HYPERV_SYNDBG         3
6377			__u32 type;
6378			__u32 pad1;
6379			union {
6380				struct {
6381					__u32 msr;
6382					__u32 pad2;
6383					__u64 control;
6384					__u64 evt_page;
6385					__u64 msg_page;
6386				} synic;
6387				struct {
6388					__u64 input;
6389					__u64 result;
6390					__u64 params[2];
6391				} hcall;
6392				struct {
6393					__u32 msr;
6394					__u32 pad2;
6395					__u64 control;
6396					__u64 status;
6397					__u64 send_page;
6398					__u64 recv_page;
6399					__u64 pending_page;
6400				} syndbg;
6401			} u;
6402		};
6403		/* KVM_EXIT_HYPERV */
6404                struct kvm_hyperv_exit hyperv;
6405
6406Indicates that the VCPU exits into userspace to process some tasks
6407related to Hyper-V emulation.
6408
6409Valid values for 'type' are:
6410
6411	- KVM_EXIT_HYPERV_SYNIC -- synchronously notify user-space about
6412
6413Hyper-V SynIC state change. Notification is used to remap SynIC
6414event/message pages and to enable/disable SynIC messages/events processing
6415in userspace.
6416
6417	- KVM_EXIT_HYPERV_SYNDBG -- synchronously notify user-space about
6418
6419Hyper-V Synthetic debugger state change. Notification is used to either update
6420the pending_page location or to send a control command (send the buffer located
6421in send_page or recv a buffer to recv_page).
6422
6423::
6424
6425		/* KVM_EXIT_ARM_NISV */
6426		struct {
6427			__u64 esr_iss;
6428			__u64 fault_ipa;
6429		} arm_nisv;
6430
6431Used on arm64 systems. If a guest accesses memory not in a memslot,
6432KVM will typically return to userspace and ask it to do MMIO emulation on its
6433behalf. However, for certain classes of instructions, no instruction decode
6434(direction, length of memory access) is provided, and fetching and decoding
6435the instruction from the VM is overly complicated to live in the kernel.
6436
6437Historically, when this situation occurred, KVM would print a warning and kill
6438the VM. KVM assumed that if the guest accessed non-memslot memory, it was
6439trying to do I/O, which just couldn't be emulated, and the warning message was
6440phrased accordingly. However, what happened more often was that a guest bug
6441caused access outside the guest memory areas which should lead to a more
6442meaningful warning message and an external abort in the guest, if the access
6443did not fall within an I/O window.
6444
6445Userspace implementations can query for KVM_CAP_ARM_NISV_TO_USER, and enable
6446this capability at VM creation. Once this is done, these types of errors will
6447instead return to userspace with KVM_EXIT_ARM_NISV, with the valid bits from
6448the ESR_EL2 in the esr_iss field, and the faulting IPA in the fault_ipa field.
6449Userspace can either fix up the access if it's actually an I/O access by
6450decoding the instruction from guest memory (if it's very brave) and continue
6451executing the guest, or it can decide to suspend, dump, or restart the guest.
6452
6453Note that KVM does not skip the faulting instruction as it does for
6454KVM_EXIT_MMIO, but userspace has to emulate any change to the processing state
6455if it decides to decode and emulate the instruction.
6456
6457::
6458
6459		/* KVM_EXIT_X86_RDMSR / KVM_EXIT_X86_WRMSR */
6460		struct {
6461			__u8 error; /* user -> kernel */
6462			__u8 pad[7];
6463			__u32 reason; /* kernel -> user */
6464			__u32 index; /* kernel -> user */
6465			__u64 data; /* kernel <-> user */
6466		} msr;
6467
6468Used on x86 systems. When the VM capability KVM_CAP_X86_USER_SPACE_MSR is
6469enabled, MSR accesses to registers that would invoke a #GP by KVM kernel code
6470may instead trigger a KVM_EXIT_X86_RDMSR exit for reads and KVM_EXIT_X86_WRMSR
6471exit for writes.
6472
6473The "reason" field specifies why the MSR interception occurred. Userspace will
6474only receive MSR exits when a particular reason was requested during through
6475ENABLE_CAP. Currently valid exit reasons are:
6476
6477============================ ========================================
6478 KVM_MSR_EXIT_REASON_UNKNOWN access to MSR that is unknown to KVM
6479 KVM_MSR_EXIT_REASON_INVAL   access to invalid MSRs or reserved bits
6480 KVM_MSR_EXIT_REASON_FILTER  access blocked by KVM_X86_SET_MSR_FILTER
6481============================ ========================================
6482
6483For KVM_EXIT_X86_RDMSR, the "index" field tells userspace which MSR the guest
6484wants to read. To respond to this request with a successful read, userspace
6485writes the respective data into the "data" field and must continue guest
6486execution to ensure the read data is transferred into guest register state.
6487
6488If the RDMSR request was unsuccessful, userspace indicates that with a "1" in
6489the "error" field. This will inject a #GP into the guest when the VCPU is
6490executed again.
6491
6492For KVM_EXIT_X86_WRMSR, the "index" field tells userspace which MSR the guest
6493wants to write. Once finished processing the event, userspace must continue
6494vCPU execution. If the MSR write was unsuccessful, userspace also sets the
6495"error" field to "1".
6496
6497See KVM_X86_SET_MSR_FILTER for details on the interaction with MSR filtering.
6498
6499::
6500
6501
6502		struct kvm_xen_exit {
6503  #define KVM_EXIT_XEN_HCALL          1
6504			__u32 type;
6505			union {
6506				struct {
6507					__u32 longmode;
6508					__u32 cpl;
6509					__u64 input;
6510					__u64 result;
6511					__u64 params[6];
6512				} hcall;
6513			} u;
6514		};
6515		/* KVM_EXIT_XEN */
6516                struct kvm_hyperv_exit xen;
6517
6518Indicates that the VCPU exits into userspace to process some tasks
6519related to Xen emulation.
6520
6521Valid values for 'type' are:
6522
6523  - KVM_EXIT_XEN_HCALL -- synchronously notify user-space about Xen hypercall.
6524    Userspace is expected to place the hypercall result into the appropriate
6525    field before invoking KVM_RUN again.
6526
6527::
6528
6529		/* KVM_EXIT_RISCV_SBI */
6530		struct {
6531			unsigned long extension_id;
6532			unsigned long function_id;
6533			unsigned long args[6];
6534			unsigned long ret[2];
6535		} riscv_sbi;
6536
6537If exit reason is KVM_EXIT_RISCV_SBI then it indicates that the VCPU has
6538done a SBI call which is not handled by KVM RISC-V kernel module. The details
6539of the SBI call are available in 'riscv_sbi' member of kvm_run structure. The
6540'extension_id' field of 'riscv_sbi' represents SBI extension ID whereas the
6541'function_id' field represents function ID of given SBI extension. The 'args'
6542array field of 'riscv_sbi' represents parameters for the SBI call and 'ret'
6543array field represents return values. The userspace should update the return
6544values of SBI call before resuming the VCPU. For more details on RISC-V SBI
6545spec refer, https://github.com/riscv/riscv-sbi-doc.
6546
6547::
6548
6549    /* KVM_EXIT_NOTIFY */
6550    struct {
6551  #define KVM_NOTIFY_CONTEXT_INVALID	(1 << 0)
6552      __u32 flags;
6553    } notify;
6554
6555Used on x86 systems. When the VM capability KVM_CAP_X86_NOTIFY_VMEXIT is
6556enabled, a VM exit generated if no event window occurs in VM non-root mode
6557for a specified amount of time. Once KVM_X86_NOTIFY_VMEXIT_USER is set when
6558enabling the cap, it would exit to userspace with the exit reason
6559KVM_EXIT_NOTIFY for further handling. The "flags" field contains more
6560detailed info.
6561
6562The valid value for 'flags' is:
6563
6564  - KVM_NOTIFY_CONTEXT_INVALID -- the VM context is corrupted and not valid
6565    in VMCS. It would run into unknown result if resume the target VM.
6566
6567::
6568
6569		/* Fix the size of the union. */
6570		char padding[256];
6571	};
6572
6573	/*
6574	 * shared registers between kvm and userspace.
6575	 * kvm_valid_regs specifies the register classes set by the host
6576	 * kvm_dirty_regs specified the register classes dirtied by userspace
6577	 * struct kvm_sync_regs is architecture specific, as well as the
6578	 * bits for kvm_valid_regs and kvm_dirty_regs
6579	 */
6580	__u64 kvm_valid_regs;
6581	__u64 kvm_dirty_regs;
6582	union {
6583		struct kvm_sync_regs regs;
6584		char padding[SYNC_REGS_SIZE_BYTES];
6585	} s;
6586
6587If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
6588certain guest registers without having to call SET/GET_*REGS. Thus we can
6589avoid some system call overhead if userspace has to handle the exit.
6590Userspace can query the validity of the structure by checking
6591kvm_valid_regs for specific bits. These bits are architecture specific
6592and usually define the validity of a groups of registers. (e.g. one bit
6593for general purpose registers)
6594
6595Please note that the kernel is allowed to use the kvm_run structure as the
6596primary storage for certain register types. Therefore, the kernel may use the
6597values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
6598
6599
66006. Capabilities that can be enabled on vCPUs
6601============================================
6602
6603There are certain capabilities that change the behavior of the virtual CPU or
6604the virtual machine when enabled. To enable them, please see section 4.37.
6605Below you can find a list of capabilities and what their effect on the vCPU or
6606the virtual machine is when enabling them.
6607
6608The following information is provided along with the description:
6609
6610  Architectures:
6611      which instruction set architectures provide this ioctl.
6612      x86 includes both i386 and x86_64.
6613
6614  Target:
6615      whether this is a per-vcpu or per-vm capability.
6616
6617  Parameters:
6618      what parameters are accepted by the capability.
6619
6620  Returns:
6621      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
6622      are not detailed, but errors with specific meanings are.
6623
6624
66256.1 KVM_CAP_PPC_OSI
6626-------------------
6627
6628:Architectures: ppc
6629:Target: vcpu
6630:Parameters: none
6631:Returns: 0 on success; -1 on error
6632
6633This capability enables interception of OSI hypercalls that otherwise would
6634be treated as normal system calls to be injected into the guest. OSI hypercalls
6635were invented by Mac-on-Linux to have a standardized communication mechanism
6636between the guest and the host.
6637
6638When this capability is enabled, KVM_EXIT_OSI can occur.
6639
6640
66416.2 KVM_CAP_PPC_PAPR
6642--------------------
6643
6644:Architectures: ppc
6645:Target: vcpu
6646:Parameters: none
6647:Returns: 0 on success; -1 on error
6648
6649This capability enables interception of PAPR hypercalls. PAPR hypercalls are
6650done using the hypercall instruction "sc 1".
6651
6652It also sets the guest privilege level to "supervisor" mode. Usually the guest
6653runs in "hypervisor" privilege mode with a few missing features.
6654
6655In addition to the above, it changes the semantics of SDR1. In this mode, the
6656HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
6657HTAB invisible to the guest.
6658
6659When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
6660
6661
66626.3 KVM_CAP_SW_TLB
6663------------------
6664
6665:Architectures: ppc
6666:Target: vcpu
6667:Parameters: args[0] is the address of a struct kvm_config_tlb
6668:Returns: 0 on success; -1 on error
6669
6670::
6671
6672  struct kvm_config_tlb {
6673	__u64 params;
6674	__u64 array;
6675	__u32 mmu_type;
6676	__u32 array_len;
6677  };
6678
6679Configures the virtual CPU's TLB array, establishing a shared memory area
6680between userspace and KVM.  The "params" and "array" fields are userspace
6681addresses of mmu-type-specific data structures.  The "array_len" field is an
6682safety mechanism, and should be set to the size in bytes of the memory that
6683userspace has reserved for the array.  It must be at least the size dictated
6684by "mmu_type" and "params".
6685
6686While KVM_RUN is active, the shared region is under control of KVM.  Its
6687contents are undefined, and any modification by userspace results in
6688boundedly undefined behavior.
6689
6690On return from KVM_RUN, the shared region will reflect the current state of
6691the guest's TLB.  If userspace makes any changes, it must call KVM_DIRTY_TLB
6692to tell KVM which entries have been changed, prior to calling KVM_RUN again
6693on this vcpu.
6694
6695For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
6696
6697 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
6698 - The "array" field points to an array of type "struct
6699   kvm_book3e_206_tlb_entry".
6700 - The array consists of all entries in the first TLB, followed by all
6701   entries in the second TLB.
6702 - Within a TLB, entries are ordered first by increasing set number.  Within a
6703   set, entries are ordered by way (increasing ESEL).
6704 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
6705   where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
6706 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
6707   hardware ignores this value for TLB0.
6708
67096.4 KVM_CAP_S390_CSS_SUPPORT
6710----------------------------
6711
6712:Architectures: s390
6713:Target: vcpu
6714:Parameters: none
6715:Returns: 0 on success; -1 on error
6716
6717This capability enables support for handling of channel I/O instructions.
6718
6719TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
6720handled in-kernel, while the other I/O instructions are passed to userspace.
6721
6722When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
6723SUBCHANNEL intercepts.
6724
6725Note that even though this capability is enabled per-vcpu, the complete
6726virtual machine is affected.
6727
67286.5 KVM_CAP_PPC_EPR
6729-------------------
6730
6731:Architectures: ppc
6732:Target: vcpu
6733:Parameters: args[0] defines whether the proxy facility is active
6734:Returns: 0 on success; -1 on error
6735
6736This capability enables or disables the delivery of interrupts through the
6737external proxy facility.
6738
6739When enabled (args[0] != 0), every time the guest gets an external interrupt
6740delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
6741to receive the topmost interrupt vector.
6742
6743When disabled (args[0] == 0), behavior is as if this facility is unsupported.
6744
6745When this capability is enabled, KVM_EXIT_EPR can occur.
6746
67476.6 KVM_CAP_IRQ_MPIC
6748--------------------
6749
6750:Architectures: ppc
6751:Parameters: args[0] is the MPIC device fd;
6752             args[1] is the MPIC CPU number for this vcpu
6753
6754This capability connects the vcpu to an in-kernel MPIC device.
6755
67566.7 KVM_CAP_IRQ_XICS
6757--------------------
6758
6759:Architectures: ppc
6760:Target: vcpu
6761:Parameters: args[0] is the XICS device fd;
6762             args[1] is the XICS CPU number (server ID) for this vcpu
6763
6764This capability connects the vcpu to an in-kernel XICS device.
6765
67666.8 KVM_CAP_S390_IRQCHIP
6767------------------------
6768
6769:Architectures: s390
6770:Target: vm
6771:Parameters: none
6772
6773This capability enables the in-kernel irqchip for s390. Please refer to
6774"4.24 KVM_CREATE_IRQCHIP" for details.
6775
67766.9 KVM_CAP_MIPS_FPU
6777--------------------
6778
6779:Architectures: mips
6780:Target: vcpu
6781:Parameters: args[0] is reserved for future use (should be 0).
6782
6783This capability allows the use of the host Floating Point Unit by the guest. It
6784allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
6785done the ``KVM_REG_MIPS_FPR_*`` and ``KVM_REG_MIPS_FCR_*`` registers can be
6786accessed (depending on the current guest FPU register mode), and the Status.FR,
6787Config5.FRE bits are accessible via the KVM API and also from the guest,
6788depending on them being supported by the FPU.
6789
67906.10 KVM_CAP_MIPS_MSA
6791---------------------
6792
6793:Architectures: mips
6794:Target: vcpu
6795:Parameters: args[0] is reserved for future use (should be 0).
6796
6797This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
6798It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
6799Once this is done the ``KVM_REG_MIPS_VEC_*`` and ``KVM_REG_MIPS_MSA_*``
6800registers can be accessed, and the Config5.MSAEn bit is accessible via the
6801KVM API and also from the guest.
6802
68036.74 KVM_CAP_SYNC_REGS
6804----------------------
6805
6806:Architectures: s390, x86
6807:Target: s390: always enabled, x86: vcpu
6808:Parameters: none
6809:Returns: x86: KVM_CHECK_EXTENSION returns a bit-array indicating which register
6810          sets are supported
6811          (bitfields defined in arch/x86/include/uapi/asm/kvm.h).
6812
6813As described above in the kvm_sync_regs struct info in section 5 (kvm_run):
6814KVM_CAP_SYNC_REGS "allow[s] userspace to access certain guest registers
6815without having to call SET/GET_*REGS". This reduces overhead by eliminating
6816repeated ioctl calls for setting and/or getting register values. This is
6817particularly important when userspace is making synchronous guest state
6818modifications, e.g. when emulating and/or intercepting instructions in
6819userspace.
6820
6821For s390 specifics, please refer to the source code.
6822
6823For x86:
6824
6825- the register sets to be copied out to kvm_run are selectable
6826  by userspace (rather that all sets being copied out for every exit).
6827- vcpu_events are available in addition to regs and sregs.
6828
6829For x86, the 'kvm_valid_regs' field of struct kvm_run is overloaded to
6830function as an input bit-array field set by userspace to indicate the
6831specific register sets to be copied out on the next exit.
6832
6833To indicate when userspace has modified values that should be copied into
6834the vCPU, the all architecture bitarray field, 'kvm_dirty_regs' must be set.
6835This is done using the same bitflags as for the 'kvm_valid_regs' field.
6836If the dirty bit is not set, then the register set values will not be copied
6837into the vCPU even if they've been modified.
6838
6839Unused bitfields in the bitarrays must be set to zero.
6840
6841::
6842
6843  struct kvm_sync_regs {
6844        struct kvm_regs regs;
6845        struct kvm_sregs sregs;
6846        struct kvm_vcpu_events events;
6847  };
6848
68496.75 KVM_CAP_PPC_IRQ_XIVE
6850-------------------------
6851
6852:Architectures: ppc
6853:Target: vcpu
6854:Parameters: args[0] is the XIVE device fd;
6855             args[1] is the XIVE CPU number (server ID) for this vcpu
6856
6857This capability connects the vcpu to an in-kernel XIVE device.
6858
68597. Capabilities that can be enabled on VMs
6860==========================================
6861
6862There are certain capabilities that change the behavior of the virtual
6863machine when enabled. To enable them, please see section 4.37. Below
6864you can find a list of capabilities and what their effect on the VM
6865is when enabling them.
6866
6867The following information is provided along with the description:
6868
6869  Architectures:
6870      which instruction set architectures provide this ioctl.
6871      x86 includes both i386 and x86_64.
6872
6873  Parameters:
6874      what parameters are accepted by the capability.
6875
6876  Returns:
6877      the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
6878      are not detailed, but errors with specific meanings are.
6879
6880
68817.1 KVM_CAP_PPC_ENABLE_HCALL
6882----------------------------
6883
6884:Architectures: ppc
6885:Parameters: args[0] is the sPAPR hcall number;
6886	     args[1] is 0 to disable, 1 to enable in-kernel handling
6887
6888This capability controls whether individual sPAPR hypercalls (hcalls)
6889get handled by the kernel or not.  Enabling or disabling in-kernel
6890handling of an hcall is effective across the VM.  On creation, an
6891initial set of hcalls are enabled for in-kernel handling, which
6892consists of those hcalls for which in-kernel handlers were implemented
6893before this capability was implemented.  If disabled, the kernel will
6894not to attempt to handle the hcall, but will always exit to userspace
6895to handle it.  Note that it may not make sense to enable some and
6896disable others of a group of related hcalls, but KVM does not prevent
6897userspace from doing that.
6898
6899If the hcall number specified is not one that has an in-kernel
6900implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
6901error.
6902
69037.2 KVM_CAP_S390_USER_SIGP
6904--------------------------
6905
6906:Architectures: s390
6907:Parameters: none
6908
6909This capability controls which SIGP orders will be handled completely in user
6910space. With this capability enabled, all fast orders will be handled completely
6911in the kernel:
6912
6913- SENSE
6914- SENSE RUNNING
6915- EXTERNAL CALL
6916- EMERGENCY SIGNAL
6917- CONDITIONAL EMERGENCY SIGNAL
6918
6919All other orders will be handled completely in user space.
6920
6921Only privileged operation exceptions will be checked for in the kernel (or even
6922in the hardware prior to interception). If this capability is not enabled, the
6923old way of handling SIGP orders is used (partially in kernel and user space).
6924
69257.3 KVM_CAP_S390_VECTOR_REGISTERS
6926---------------------------------
6927
6928:Architectures: s390
6929:Parameters: none
6930:Returns: 0 on success, negative value on error
6931
6932Allows use of the vector registers introduced with z13 processor, and
6933provides for the synchronization between host and user space.  Will
6934return -EINVAL if the machine does not support vectors.
6935
69367.4 KVM_CAP_S390_USER_STSI
6937--------------------------
6938
6939:Architectures: s390
6940:Parameters: none
6941
6942This capability allows post-handlers for the STSI instruction. After
6943initial handling in the kernel, KVM exits to user space with
6944KVM_EXIT_S390_STSI to allow user space to insert further data.
6945
6946Before exiting to userspace, kvm handlers should fill in s390_stsi field of
6947vcpu->run::
6948
6949  struct {
6950	__u64 addr;
6951	__u8 ar;
6952	__u8 reserved;
6953	__u8 fc;
6954	__u8 sel1;
6955	__u16 sel2;
6956  } s390_stsi;
6957
6958  @addr - guest address of STSI SYSIB
6959  @fc   - function code
6960  @sel1 - selector 1
6961  @sel2 - selector 2
6962  @ar   - access register number
6963
6964KVM handlers should exit to userspace with rc = -EREMOTE.
6965
69667.5 KVM_CAP_SPLIT_IRQCHIP
6967-------------------------
6968
6969:Architectures: x86
6970:Parameters: args[0] - number of routes reserved for userspace IOAPICs
6971:Returns: 0 on success, -1 on error
6972
6973Create a local apic for each processor in the kernel. This can be used
6974instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
6975IOAPIC and PIC (and also the PIT, even though this has to be enabled
6976separately).
6977
6978This capability also enables in kernel routing of interrupt requests;
6979when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
6980used in the IRQ routing table.  The first args[0] MSI routes are reserved
6981for the IOAPIC pins.  Whenever the LAPIC receives an EOI for these routes,
6982a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
6983
6984Fails if VCPU has already been created, or if the irqchip is already in the
6985kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
6986
69877.6 KVM_CAP_S390_RI
6988-------------------
6989
6990:Architectures: s390
6991:Parameters: none
6992
6993Allows use of runtime-instrumentation introduced with zEC12 processor.
6994Will return -EINVAL if the machine does not support runtime-instrumentation.
6995Will return -EBUSY if a VCPU has already been created.
6996
69977.7 KVM_CAP_X2APIC_API
6998----------------------
6999
7000:Architectures: x86
7001:Parameters: args[0] - features that should be enabled
7002:Returns: 0 on success, -EINVAL when args[0] contains invalid features
7003
7004Valid feature flags in args[0] are::
7005
7006  #define KVM_X2APIC_API_USE_32BIT_IDS            (1ULL << 0)
7007  #define KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK  (1ULL << 1)
7008
7009Enabling KVM_X2APIC_API_USE_32BIT_IDS changes the behavior of
7010KVM_SET_GSI_ROUTING, KVM_SIGNAL_MSI, KVM_SET_LAPIC, and KVM_GET_LAPIC,
7011allowing the use of 32-bit APIC IDs.  See KVM_CAP_X2APIC_API in their
7012respective sections.
7013
7014KVM_X2APIC_API_DISABLE_BROADCAST_QUIRK must be enabled for x2APIC to work
7015in logical mode or with more than 255 VCPUs.  Otherwise, KVM treats 0xff
7016as a broadcast even in x2APIC mode in order to support physical x2APIC
7017without interrupt remapping.  This is undesirable in logical mode,
7018where 0xff represents CPUs 0-7 in cluster 0.
7019
70207.8 KVM_CAP_S390_USER_INSTR0
7021----------------------------
7022
7023:Architectures: s390
7024:Parameters: none
7025
7026With this capability enabled, all illegal instructions 0x0000 (2 bytes) will
7027be intercepted and forwarded to user space. User space can use this
7028mechanism e.g. to realize 2-byte software breakpoints. The kernel will
7029not inject an operating exception for these instructions, user space has
7030to take care of that.
7031
7032This capability can be enabled dynamically even if VCPUs were already
7033created and are running.
7034
70357.9 KVM_CAP_S390_GS
7036-------------------
7037
7038:Architectures: s390
7039:Parameters: none
7040:Returns: 0 on success; -EINVAL if the machine does not support
7041          guarded storage; -EBUSY if a VCPU has already been created.
7042
7043Allows use of guarded storage for the KVM guest.
7044
70457.10 KVM_CAP_S390_AIS
7046---------------------
7047
7048:Architectures: s390
7049:Parameters: none
7050
7051Allow use of adapter-interruption suppression.
7052:Returns: 0 on success; -EBUSY if a VCPU has already been created.
7053
70547.11 KVM_CAP_PPC_SMT
7055--------------------
7056
7057:Architectures: ppc
7058:Parameters: vsmt_mode, flags
7059
7060Enabling this capability on a VM provides userspace with a way to set
7061the desired virtual SMT mode (i.e. the number of virtual CPUs per
7062virtual core).  The virtual SMT mode, vsmt_mode, must be a power of 2
7063between 1 and 8.  On POWER8, vsmt_mode must also be no greater than
7064the number of threads per subcore for the host.  Currently flags must
7065be 0.  A successful call to enable this capability will result in
7066vsmt_mode being returned when the KVM_CAP_PPC_SMT capability is
7067subsequently queried for the VM.  This capability is only supported by
7068HV KVM, and can only be set before any VCPUs have been created.
7069The KVM_CAP_PPC_SMT_POSSIBLE capability indicates which virtual SMT
7070modes are available.
7071
70727.12 KVM_CAP_PPC_FWNMI
7073----------------------
7074
7075:Architectures: ppc
7076:Parameters: none
7077
7078With this capability a machine check exception in the guest address
7079space will cause KVM to exit the guest with NMI exit reason. This
7080enables QEMU to build error log and branch to guest kernel registered
7081machine check handling routine. Without this capability KVM will
7082branch to guests' 0x200 interrupt vector.
7083
70847.13 KVM_CAP_X86_DISABLE_EXITS
7085------------------------------
7086
7087:Architectures: x86
7088:Parameters: args[0] defines which exits are disabled
7089:Returns: 0 on success, -EINVAL when args[0] contains invalid exits
7090
7091Valid bits in args[0] are::
7092
7093  #define KVM_X86_DISABLE_EXITS_MWAIT            (1 << 0)
7094  #define KVM_X86_DISABLE_EXITS_HLT              (1 << 1)
7095  #define KVM_X86_DISABLE_EXITS_PAUSE            (1 << 2)
7096  #define KVM_X86_DISABLE_EXITS_CSTATE           (1 << 3)
7097
7098Enabling this capability on a VM provides userspace with a way to no
7099longer intercept some instructions for improved latency in some
7100workloads, and is suggested when vCPUs are associated to dedicated
7101physical CPUs.  More bits can be added in the future; userspace can
7102just pass the KVM_CHECK_EXTENSION result to KVM_ENABLE_CAP to disable
7103all such vmexits.
7104
7105Do not enable KVM_FEATURE_PV_UNHALT if you disable HLT exits.
7106
71077.14 KVM_CAP_S390_HPAGE_1M
7108--------------------------
7109
7110:Architectures: s390
7111:Parameters: none
7112:Returns: 0 on success, -EINVAL if hpage module parameter was not set
7113	  or cmma is enabled, or the VM has the KVM_VM_S390_UCONTROL
7114	  flag set
7115
7116With this capability the KVM support for memory backing with 1m pages
7117through hugetlbfs can be enabled for a VM. After the capability is
7118enabled, cmma can't be enabled anymore and pfmfi and the storage key
7119interpretation are disabled. If cmma has already been enabled or the
7120hpage module parameter is not set to 1, -EINVAL is returned.
7121
7122While it is generally possible to create a huge page backed VM without
7123this capability, the VM will not be able to run.
7124
71257.15 KVM_CAP_MSR_PLATFORM_INFO
7126------------------------------
7127
7128:Architectures: x86
7129:Parameters: args[0] whether feature should be enabled or not
7130
7131With this capability, a guest may read the MSR_PLATFORM_INFO MSR. Otherwise,
7132a #GP would be raised when the guest tries to access. Currently, this
7133capability does not enable write permissions of this MSR for the guest.
7134
71357.16 KVM_CAP_PPC_NESTED_HV
7136--------------------------
7137
7138:Architectures: ppc
7139:Parameters: none
7140:Returns: 0 on success, -EINVAL when the implementation doesn't support
7141	  nested-HV virtualization.
7142
7143HV-KVM on POWER9 and later systems allows for "nested-HV"
7144virtualization, which provides a way for a guest VM to run guests that
7145can run using the CPU's supervisor mode (privileged non-hypervisor
7146state).  Enabling this capability on a VM depends on the CPU having
7147the necessary functionality and on the facility being enabled with a
7148kvm-hv module parameter.
7149
71507.17 KVM_CAP_EXCEPTION_PAYLOAD
7151------------------------------
7152
7153:Architectures: x86
7154:Parameters: args[0] whether feature should be enabled or not
7155
7156With this capability enabled, CR2 will not be modified prior to the
7157emulated VM-exit when L1 intercepts a #PF exception that occurs in
7158L2. Similarly, for kvm-intel only, DR6 will not be modified prior to
7159the emulated VM-exit when L1 intercepts a #DB exception that occurs in
7160L2. As a result, when KVM_GET_VCPU_EVENTS reports a pending #PF (or
7161#DB) exception for L2, exception.has_payload will be set and the
7162faulting address (or the new DR6 bits*) will be reported in the
7163exception_payload field. Similarly, when userspace injects a #PF (or
7164#DB) into L2 using KVM_SET_VCPU_EVENTS, it is expected to set
7165exception.has_payload and to put the faulting address - or the new DR6
7166bits\ [#]_ - in the exception_payload field.
7167
7168This capability also enables exception.pending in struct
7169kvm_vcpu_events, which allows userspace to distinguish between pending
7170and injected exceptions.
7171
7172
7173.. [#] For the new DR6 bits, note that bit 16 is set iff the #DB exception
7174       will clear DR6.RTM.
7175
71767.18 KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
7177
7178:Architectures: x86, arm64, mips
7179:Parameters: args[0] whether feature should be enabled or not
7180
7181Valid flags are::
7182
7183  #define KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE   (1 << 0)
7184  #define KVM_DIRTY_LOG_INITIALLY_SET           (1 << 1)
7185
7186With KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE is set, KVM_GET_DIRTY_LOG will not
7187automatically clear and write-protect all pages that are returned as dirty.
7188Rather, userspace will have to do this operation separately using
7189KVM_CLEAR_DIRTY_LOG.
7190
7191At the cost of a slightly more complicated operation, this provides better
7192scalability and responsiveness for two reasons.  First,
7193KVM_CLEAR_DIRTY_LOG ioctl can operate on a 64-page granularity rather
7194than requiring to sync a full memslot; this ensures that KVM does not
7195take spinlocks for an extended period of time.  Second, in some cases a
7196large amount of time can pass between a call to KVM_GET_DIRTY_LOG and
7197userspace actually using the data in the page.  Pages can be modified
7198during this time, which is inefficient for both the guest and userspace:
7199the guest will incur a higher penalty due to write protection faults,
7200while userspace can see false reports of dirty pages.  Manual reprotection
7201helps reducing this time, improving guest performance and reducing the
7202number of dirty log false positives.
7203
7204With KVM_DIRTY_LOG_INITIALLY_SET set, all the bits of the dirty bitmap
7205will be initialized to 1 when created.  This also improves performance because
7206dirty logging can be enabled gradually in small chunks on the first call
7207to KVM_CLEAR_DIRTY_LOG.  KVM_DIRTY_LOG_INITIALLY_SET depends on
7208KVM_DIRTY_LOG_MANUAL_PROTECT_ENABLE (it is also only available on
7209x86 and arm64 for now).
7210
7211KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 was previously available under the name
7212KVM_CAP_MANUAL_DIRTY_LOG_PROTECT, but the implementation had bugs that make
7213it hard or impossible to use it correctly.  The availability of
7214KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2 signals that those bugs are fixed.
7215Userspace should not try to use KVM_CAP_MANUAL_DIRTY_LOG_PROTECT.
7216
72177.19 KVM_CAP_PPC_SECURE_GUEST
7218------------------------------
7219
7220:Architectures: ppc
7221
7222This capability indicates that KVM is running on a host that has
7223ultravisor firmware and thus can support a secure guest.  On such a
7224system, a guest can ask the ultravisor to make it a secure guest,
7225one whose memory is inaccessible to the host except for pages which
7226are explicitly requested to be shared with the host.  The ultravisor
7227notifies KVM when a guest requests to become a secure guest, and KVM
7228has the opportunity to veto the transition.
7229
7230If present, this capability can be enabled for a VM, meaning that KVM
7231will allow the transition to secure guest mode.  Otherwise KVM will
7232veto the transition.
7233
72347.20 KVM_CAP_HALT_POLL
7235----------------------
7236
7237:Architectures: all
7238:Target: VM
7239:Parameters: args[0] is the maximum poll time in nanoseconds
7240:Returns: 0 on success; -1 on error
7241
7242KVM_CAP_HALT_POLL overrides the kvm.halt_poll_ns module parameter to set the
7243maximum halt-polling time for all vCPUs in the target VM. This capability can
7244be invoked at any time and any number of times to dynamically change the
7245maximum halt-polling time.
7246
7247See Documentation/virt/kvm/halt-polling.rst for more information on halt
7248polling.
7249
72507.21 KVM_CAP_X86_USER_SPACE_MSR
7251-------------------------------
7252
7253:Architectures: x86
7254:Target: VM
7255:Parameters: args[0] contains the mask of KVM_MSR_EXIT_REASON_* events to report
7256:Returns: 0 on success; -1 on error
7257
7258This capability allows userspace to intercept RDMSR and WRMSR instructions if
7259access to an MSR is denied.  By default, KVM injects #GP on denied accesses.
7260
7261When a guest requests to read or write an MSR, KVM may not implement all MSRs
7262that are relevant to a respective system. It also does not differentiate by
7263CPU type.
7264
7265To allow more fine grained control over MSR handling, userspace may enable
7266this capability. With it enabled, MSR accesses that match the mask specified in
7267args[0] and would trigger a #GP inside the guest will instead trigger
7268KVM_EXIT_X86_RDMSR and KVM_EXIT_X86_WRMSR exit notifications.  Userspace
7269can then implement model specific MSR handling and/or user notifications
7270to inform a user that an MSR was not emulated/virtualized by KVM.
7271
7272The valid mask flags are:
7273
7274============================ ===============================================
7275 KVM_MSR_EXIT_REASON_UNKNOWN intercept accesses to unknown (to KVM) MSRs
7276 KVM_MSR_EXIT_REASON_INVAL   intercept accesses that are architecturally
7277                             invalid according to the vCPU model and/or mode
7278 KVM_MSR_EXIT_REASON_FILTER  intercept accesses that are denied by userspace
7279                             via KVM_X86_SET_MSR_FILTER
7280============================ ===============================================
7281
72827.22 KVM_CAP_X86_BUS_LOCK_EXIT
7283-------------------------------
7284
7285:Architectures: x86
7286:Target: VM
7287:Parameters: args[0] defines the policy used when bus locks detected in guest
7288:Returns: 0 on success, -EINVAL when args[0] contains invalid bits
7289
7290Valid bits in args[0] are::
7291
7292  #define KVM_BUS_LOCK_DETECTION_OFF      (1 << 0)
7293  #define KVM_BUS_LOCK_DETECTION_EXIT     (1 << 1)
7294
7295Enabling this capability on a VM provides userspace with a way to select
7296a policy to handle the bus locks detected in guest. Userspace can obtain
7297the supported modes from the result of KVM_CHECK_EXTENSION and define it
7298through the KVM_ENABLE_CAP.
7299
7300KVM_BUS_LOCK_DETECTION_OFF and KVM_BUS_LOCK_DETECTION_EXIT are supported
7301currently and mutually exclusive with each other. More bits can be added in
7302the future.
7303
7304With KVM_BUS_LOCK_DETECTION_OFF set, bus locks in guest will not cause vm exits
7305so that no additional actions are needed. This is the default mode.
7306
7307With KVM_BUS_LOCK_DETECTION_EXIT set, vm exits happen when bus lock detected
7308in VM. KVM just exits to userspace when handling them. Userspace can enforce
7309its own throttling or other policy based mitigations.
7310
7311This capability is aimed to address the thread that VM can exploit bus locks to
7312degree the performance of the whole system. Once the userspace enable this
7313capability and select the KVM_BUS_LOCK_DETECTION_EXIT mode, KVM will set the
7314KVM_RUN_BUS_LOCK flag in vcpu-run->flags field and exit to userspace. Concerning
7315the bus lock vm exit can be preempted by a higher priority VM exit, the exit
7316notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons.
7317KVM_RUN_BUS_LOCK flag is used to distinguish between them.
7318
73197.23 KVM_CAP_PPC_DAWR1
7320----------------------
7321
7322:Architectures: ppc
7323:Parameters: none
7324:Returns: 0 on success, -EINVAL when CPU doesn't support 2nd DAWR
7325
7326This capability can be used to check / enable 2nd DAWR feature provided
7327by POWER10 processor.
7328
7329
73307.24 KVM_CAP_VM_COPY_ENC_CONTEXT_FROM
7331-------------------------------------
7332
7333Architectures: x86 SEV enabled
7334Type: vm
7335Parameters: args[0] is the fd of the source vm
7336Returns: 0 on success; ENOTTY on error
7337
7338This capability enables userspace to copy encryption context from the vm
7339indicated by the fd to the vm this is called on.
7340
7341This is intended to support in-guest workloads scheduled by the host. This
7342allows the in-guest workload to maintain its own NPTs and keeps the two vms
7343from accidentally clobbering each other with interrupts and the like (separate
7344APIC/MSRs/etc).
7345
73467.25 KVM_CAP_SGX_ATTRIBUTE
7347--------------------------
7348
7349:Architectures: x86
7350:Target: VM
7351:Parameters: args[0] is a file handle of a SGX attribute file in securityfs
7352:Returns: 0 on success, -EINVAL if the file handle is invalid or if a requested
7353          attribute is not supported by KVM.
7354
7355KVM_CAP_SGX_ATTRIBUTE enables a userspace VMM to grant a VM access to one or
7356more priveleged enclave attributes.  args[0] must hold a file handle to a valid
7357SGX attribute file corresponding to an attribute that is supported/restricted
7358by KVM (currently only PROVISIONKEY).
7359
7360The SGX subsystem restricts access to a subset of enclave attributes to provide
7361additional security for an uncompromised kernel, e.g. use of the PROVISIONKEY
7362is restricted to deter malware from using the PROVISIONKEY to obtain a stable
7363system fingerprint.  To prevent userspace from circumventing such restrictions
7364by running an enclave in a VM, KVM prevents access to privileged attributes by
7365default.
7366
7367See Documentation/x86/sgx.rst for more details.
7368
73697.26 KVM_CAP_PPC_RPT_INVALIDATE
7370-------------------------------
7371
7372:Capability: KVM_CAP_PPC_RPT_INVALIDATE
7373:Architectures: ppc
7374:Type: vm
7375
7376This capability indicates that the kernel is capable of handling
7377H_RPT_INVALIDATE hcall.
7378
7379In order to enable the use of H_RPT_INVALIDATE in the guest,
7380user space might have to advertise it for the guest. For example,
7381IBM pSeries (sPAPR) guest starts using it if "hcall-rpt-invalidate" is
7382present in the "ibm,hypertas-functions" device-tree property.
7383
7384This capability is enabled for hypervisors on platforms like POWER9
7385that support radix MMU.
7386
73877.27 KVM_CAP_EXIT_ON_EMULATION_FAILURE
7388--------------------------------------
7389
7390:Architectures: x86
7391:Parameters: args[0] whether the feature should be enabled or not
7392
7393When this capability is enabled, an emulation failure will result in an exit
7394to userspace with KVM_INTERNAL_ERROR (except when the emulator was invoked
7395to handle a VMware backdoor instruction). Furthermore, KVM will now provide up
7396to 15 instruction bytes for any exit to userspace resulting from an emulation
7397failure.  When these exits to userspace occur use the emulation_failure struct
7398instead of the internal struct.  They both have the same layout, but the
7399emulation_failure struct matches the content better.  It also explicitly
7400defines the 'flags' field which is used to describe the fields in the struct
7401that are valid (ie: if KVM_INTERNAL_ERROR_EMULATION_FLAG_INSTRUCTION_BYTES is
7402set in the 'flags' field then both 'insn_size' and 'insn_bytes' have valid data
7403in them.)
7404
74057.28 KVM_CAP_ARM_MTE
7406--------------------
7407
7408:Architectures: arm64
7409:Parameters: none
7410
7411This capability indicates that KVM (and the hardware) supports exposing the
7412Memory Tagging Extensions (MTE) to the guest. It must also be enabled by the
7413VMM before creating any VCPUs to allow the guest access. Note that MTE is only
7414available to a guest running in AArch64 mode and enabling this capability will
7415cause attempts to create AArch32 VCPUs to fail.
7416
7417When enabled the guest is able to access tags associated with any memory given
7418to the guest. KVM will ensure that the tags are maintained during swap or
7419hibernation of the host; however the VMM needs to manually save/restore the
7420tags as appropriate if the VM is migrated.
7421
7422When this capability is enabled all memory in memslots must be mapped as
7423``MAP_ANONYMOUS`` or with a RAM-based file mapping (``tmpfs``, ``memfd``),
7424attempts to create a memslot with an invalid mmap will result in an
7425-EINVAL return.
7426
7427When enabled the VMM may make use of the ``KVM_ARM_MTE_COPY_TAGS`` ioctl to
7428perform a bulk copy of tags to/from the guest.
7429
74307.29 KVM_CAP_VM_MOVE_ENC_CONTEXT_FROM
7431-------------------------------------
7432
7433Architectures: x86 SEV enabled
7434Type: vm
7435Parameters: args[0] is the fd of the source vm
7436Returns: 0 on success
7437
7438This capability enables userspace to migrate the encryption context from the VM
7439indicated by the fd to the VM this is called on.
7440
7441This is intended to support intra-host migration of VMs between userspace VMMs,
7442upgrading the VMM process without interrupting the guest.
7443
74447.30 KVM_CAP_PPC_AIL_MODE_3
7445-------------------------------
7446
7447:Capability: KVM_CAP_PPC_AIL_MODE_3
7448:Architectures: ppc
7449:Type: vm
7450
7451This capability indicates that the kernel supports the mode 3 setting for the
7452"Address Translation Mode on Interrupt" aka "Alternate Interrupt Location"
7453resource that is controlled with the H_SET_MODE hypercall.
7454
7455This capability allows a guest kernel to use a better-performance mode for
7456handling interrupts and system calls.
7457
74587.31 KVM_CAP_DISABLE_QUIRKS2
7459----------------------------
7460
7461:Capability: KVM_CAP_DISABLE_QUIRKS2
7462:Parameters: args[0] - set of KVM quirks to disable
7463:Architectures: x86
7464:Type: vm
7465
7466This capability, if enabled, will cause KVM to disable some behavior
7467quirks.
7468
7469Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
7470quirks that can be disabled in KVM.
7471
7472The argument to KVM_ENABLE_CAP for this capability is a bitmask of
7473quirks to disable, and must be a subset of the bitmask returned by
7474KVM_CHECK_EXTENSION.
7475
7476The valid bits in cap.args[0] are:
7477
7478=================================== ============================================
7479 KVM_X86_QUIRK_LINT0_REENABLED      By default, the reset value for the LVT
7480                                    LINT0 register is 0x700 (APIC_MODE_EXTINT).
7481                                    When this quirk is disabled, the reset value
7482                                    is 0x10000 (APIC_LVT_MASKED).
7483
7484 KVM_X86_QUIRK_CD_NW_CLEARED        By default, KVM clears CR0.CD and CR0.NW.
7485                                    When this quirk is disabled, KVM does not
7486                                    change the value of CR0.CD and CR0.NW.
7487
7488 KVM_X86_QUIRK_LAPIC_MMIO_HOLE      By default, the MMIO LAPIC interface is
7489                                    available even when configured for x2APIC
7490                                    mode. When this quirk is disabled, KVM
7491                                    disables the MMIO LAPIC interface if the
7492                                    LAPIC is in x2APIC mode.
7493
7494 KVM_X86_QUIRK_OUT_7E_INC_RIP       By default, KVM pre-increments %rip before
7495                                    exiting to userspace for an OUT instruction
7496                                    to port 0x7e. When this quirk is disabled,
7497                                    KVM does not pre-increment %rip before
7498                                    exiting to userspace.
7499
7500 KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT When this quirk is disabled, KVM sets
7501                                    CPUID.01H:ECX[bit 3] (MONITOR/MWAIT) if
7502                                    IA32_MISC_ENABLE[bit 18] (MWAIT) is set.
7503                                    Additionally, when this quirk is disabled,
7504                                    KVM clears CPUID.01H:ECX[bit 3] if
7505                                    IA32_MISC_ENABLE[bit 18] is cleared.
7506
7507 KVM_X86_QUIRK_FIX_HYPERCALL_INSN   By default, KVM rewrites guest
7508                                    VMMCALL/VMCALL instructions to match the
7509                                    vendor's hypercall instruction for the
7510                                    system. When this quirk is disabled, KVM
7511                                    will no longer rewrite invalid guest
7512                                    hypercall instructions. Executing the
7513                                    incorrect hypercall instruction will
7514                                    generate a #UD within the guest.
7515
7516KVM_X86_QUIRK_MWAIT_NEVER_UD_FAULTS By default, KVM emulates MONITOR/MWAIT (if
7517                                    they are intercepted) as NOPs regardless of
7518                                    whether or not MONITOR/MWAIT are supported
7519                                    according to guest CPUID.  When this quirk
7520                                    is disabled and KVM_X86_DISABLE_EXITS_MWAIT
7521                                    is not set (MONITOR/MWAIT are intercepted),
7522                                    KVM will inject a #UD on MONITOR/MWAIT if
7523                                    they're unsupported per guest CPUID.  Note,
7524                                    KVM will modify MONITOR/MWAIT support in
7525                                    guest CPUID on writes to MISC_ENABLE if
7526                                    KVM_X86_QUIRK_MISC_ENABLE_NO_MWAIT is
7527                                    disabled.
7528=================================== ============================================
7529
75307.32 KVM_CAP_MAX_VCPU_ID
7531------------------------
7532
7533:Architectures: x86
7534:Target: VM
7535:Parameters: args[0] - maximum APIC ID value set for current VM
7536:Returns: 0 on success, -EINVAL if args[0] is beyond KVM_MAX_VCPU_IDS
7537          supported in KVM or if it has been set.
7538
7539This capability allows userspace to specify maximum possible APIC ID
7540assigned for current VM session prior to the creation of vCPUs, saving
7541memory for data structures indexed by the APIC ID.  Userspace is able
7542to calculate the limit to APIC ID values from designated
7543CPU topology.
7544
7545The value can be changed only until KVM_ENABLE_CAP is set to a nonzero
7546value or until a vCPU is created.  Upon creation of the first vCPU,
7547if the value was set to zero or KVM_ENABLE_CAP was not invoked, KVM
7548uses the return value of KVM_CHECK_EXTENSION(KVM_CAP_MAX_VCPU_ID) as
7549the maximum APIC ID.
7550
75517.33 KVM_CAP_X86_NOTIFY_VMEXIT
7552------------------------------
7553
7554:Architectures: x86
7555:Target: VM
7556:Parameters: args[0] is the value of notify window as well as some flags
7557:Returns: 0 on success, -EINVAL if args[0] contains invalid flags or notify
7558          VM exit is unsupported.
7559
7560Bits 63:32 of args[0] are used for notify window.
7561Bits 31:0 of args[0] are for some flags. Valid bits are::
7562
7563  #define KVM_X86_NOTIFY_VMEXIT_ENABLED    (1 << 0)
7564  #define KVM_X86_NOTIFY_VMEXIT_USER       (1 << 1)
7565
7566This capability allows userspace to configure the notify VM exit on/off
7567in per-VM scope during VM creation. Notify VM exit is disabled by default.
7568When userspace sets KVM_X86_NOTIFY_VMEXIT_ENABLED bit in args[0], VMM will
7569enable this feature with the notify window provided, which will generate
7570a VM exit if no event window occurs in VM non-root mode for a specified of
7571time (notify window).
7572
7573If KVM_X86_NOTIFY_VMEXIT_USER is set in args[0], upon notify VM exits happen,
7574KVM would exit to userspace for handling.
7575
7576This capability is aimed to mitigate the threat that malicious VMs can
7577cause CPU stuck (due to event windows don't open up) and make the CPU
7578unavailable to host or other VMs.
7579
75808. Other capabilities.
7581======================
7582
7583This section lists capabilities that give information about other
7584features of the KVM implementation.
7585
75868.1 KVM_CAP_PPC_HWRNG
7587---------------------
7588
7589:Architectures: ppc
7590
7591This capability, if KVM_CHECK_EXTENSION indicates that it is
7592available, means that the kernel has an implementation of the
7593H_RANDOM hypercall backed by a hardware random-number generator.
7594If present, the kernel H_RANDOM handler can be enabled for guest use
7595with the KVM_CAP_PPC_ENABLE_HCALL capability.
7596
75978.2 KVM_CAP_HYPERV_SYNIC
7598------------------------
7599
7600:Architectures: x86
7601
7602This capability, if KVM_CHECK_EXTENSION indicates that it is
7603available, means that the kernel has an implementation of the
7604Hyper-V Synthetic interrupt controller(SynIC). Hyper-V SynIC is
7605used to support Windows Hyper-V based guest paravirt drivers(VMBus).
7606
7607In order to use SynIC, it has to be activated by setting this
7608capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
7609will disable the use of APIC hardware virtualization even if supported
7610by the CPU, as it's incompatible with SynIC auto-EOI behavior.
7611
76128.3 KVM_CAP_PPC_RADIX_MMU
7613-------------------------
7614
7615:Architectures: ppc
7616
7617This capability, if KVM_CHECK_EXTENSION indicates that it is
7618available, means that the kernel can support guests using the
7619radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
7620processor).
7621
76228.4 KVM_CAP_PPC_HASH_MMU_V3
7623---------------------------
7624
7625:Architectures: ppc
7626
7627This capability, if KVM_CHECK_EXTENSION indicates that it is
7628available, means that the kernel can support guests using the
7629hashed page table MMU defined in Power ISA V3.00 (as implemented in
7630the POWER9 processor), including in-memory segment tables.
7631
76328.5 KVM_CAP_MIPS_VZ
7633-------------------
7634
7635:Architectures: mips
7636
7637This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7638it is available, means that full hardware assisted virtualization capabilities
7639of the hardware are available for use through KVM. An appropriate
7640KVM_VM_MIPS_* type must be passed to KVM_CREATE_VM to create a VM which
7641utilises it.
7642
7643If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7644available, it means that the VM is using full hardware assisted virtualization
7645capabilities of the hardware. This is useful to check after creating a VM with
7646KVM_VM_MIPS_DEFAULT.
7647
7648The value returned by KVM_CHECK_EXTENSION should be compared against known
7649values (see below). All other values are reserved. This is to allow for the
7650possibility of other hardware assisted virtualization implementations which
7651may be incompatible with the MIPS VZ ASE.
7652
7653==  ==========================================================================
7654 0  The trap & emulate implementation is in use to run guest code in user
7655    mode. Guest virtual memory segments are rearranged to fit the guest in the
7656    user mode address space.
7657
7658 1  The MIPS VZ ASE is in use, providing full hardware assisted
7659    virtualization, including standard guest virtual memory segments.
7660==  ==========================================================================
7661
76628.6 KVM_CAP_MIPS_TE
7663-------------------
7664
7665:Architectures: mips
7666
7667This capability, if KVM_CHECK_EXTENSION on the main kvm handle indicates that
7668it is available, means that the trap & emulate implementation is available to
7669run guest code in user mode, even if KVM_CAP_MIPS_VZ indicates that hardware
7670assisted virtualisation is also available. KVM_VM_MIPS_TE (0) must be passed
7671to KVM_CREATE_VM to create a VM which utilises it.
7672
7673If KVM_CHECK_EXTENSION on a kvm VM handle indicates that this capability is
7674available, it means that the VM is using trap & emulate.
7675
76768.7 KVM_CAP_MIPS_64BIT
7677----------------------
7678
7679:Architectures: mips
7680
7681This capability indicates the supported architecture type of the guest, i.e. the
7682supported register and address width.
7683
7684The values returned when this capability is checked by KVM_CHECK_EXTENSION on a
7685kvm VM handle correspond roughly to the CP0_Config.AT register field, and should
7686be checked specifically against known values (see below). All other values are
7687reserved.
7688
7689==  ========================================================================
7690 0  MIPS32 or microMIPS32.
7691    Both registers and addresses are 32-bits wide.
7692    It will only be possible to run 32-bit guest code.
7693
7694 1  MIPS64 or microMIPS64 with access only to 32-bit compatibility segments.
7695    Registers are 64-bits wide, but addresses are 32-bits wide.
7696    64-bit guest code may run but cannot access MIPS64 memory segments.
7697    It will also be possible to run 32-bit guest code.
7698
7699 2  MIPS64 or microMIPS64 with access to all address segments.
7700    Both registers and addresses are 64-bits wide.
7701    It will be possible to run 64-bit or 32-bit guest code.
7702==  ========================================================================
7703
77048.9 KVM_CAP_ARM_USER_IRQ
7705------------------------
7706
7707:Architectures: arm64
7708
7709This capability, if KVM_CHECK_EXTENSION indicates that it is available, means
7710that if userspace creates a VM without an in-kernel interrupt controller, it
7711will be notified of changes to the output level of in-kernel emulated devices,
7712which can generate virtual interrupts, presented to the VM.
7713For such VMs, on every return to userspace, the kernel
7714updates the vcpu's run->s.regs.device_irq_level field to represent the actual
7715output level of the device.
7716
7717Whenever kvm detects a change in the device output level, kvm guarantees at
7718least one return to userspace before running the VM.  This exit could either
7719be a KVM_EXIT_INTR or any other exit event, like KVM_EXIT_MMIO. This way,
7720userspace can always sample the device output level and re-compute the state of
7721the userspace interrupt controller.  Userspace should always check the state
7722of run->s.regs.device_irq_level on every kvm exit.
7723The value in run->s.regs.device_irq_level can represent both level and edge
7724triggered interrupt signals, depending on the device.  Edge triggered interrupt
7725signals will exit to userspace with the bit in run->s.regs.device_irq_level
7726set exactly once per edge signal.
7727
7728The field run->s.regs.device_irq_level is available independent of
7729run->kvm_valid_regs or run->kvm_dirty_regs bits.
7730
7731If KVM_CAP_ARM_USER_IRQ is supported, the KVM_CHECK_EXTENSION ioctl returns a
7732number larger than 0 indicating the version of this capability is implemented
7733and thereby which bits in run->s.regs.device_irq_level can signal values.
7734
7735Currently the following bits are defined for the device_irq_level bitmap::
7736
7737  KVM_CAP_ARM_USER_IRQ >= 1:
7738
7739    KVM_ARM_DEV_EL1_VTIMER -  EL1 virtual timer
7740    KVM_ARM_DEV_EL1_PTIMER -  EL1 physical timer
7741    KVM_ARM_DEV_PMU        -  ARM PMU overflow interrupt signal
7742
7743Future versions of kvm may implement additional events. These will get
7744indicated by returning a higher number from KVM_CHECK_EXTENSION and will be
7745listed above.
7746
77478.10 KVM_CAP_PPC_SMT_POSSIBLE
7748-----------------------------
7749
7750:Architectures: ppc
7751
7752Querying this capability returns a bitmap indicating the possible
7753virtual SMT modes that can be set using KVM_CAP_PPC_SMT.  If bit N
7754(counting from the right) is set, then a virtual SMT mode of 2^N is
7755available.
7756
77578.11 KVM_CAP_HYPERV_SYNIC2
7758--------------------------
7759
7760:Architectures: x86
7761
7762This capability enables a newer version of Hyper-V Synthetic interrupt
7763controller (SynIC).  The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
7764doesn't clear SynIC message and event flags pages when they are enabled by
7765writing to the respective MSRs.
7766
77678.12 KVM_CAP_HYPERV_VP_INDEX
7768----------------------------
7769
7770:Architectures: x86
7771
7772This capability indicates that userspace can load HV_X64_MSR_VP_INDEX msr.  Its
7773value is used to denote the target vcpu for a SynIC interrupt.  For
7774compatibilty, KVM initializes this msr to KVM's internal vcpu index.  When this
7775capability is absent, userspace can still query this msr's value.
7776
77778.13 KVM_CAP_S390_AIS_MIGRATION
7778-------------------------------
7779
7780:Architectures: s390
7781:Parameters: none
7782
7783This capability indicates if the flic device will be able to get/set the
7784AIS states for migration via the KVM_DEV_FLIC_AISM_ALL attribute and allows
7785to discover this without having to create a flic device.
7786
77878.14 KVM_CAP_S390_PSW
7788---------------------
7789
7790:Architectures: s390
7791
7792This capability indicates that the PSW is exposed via the kvm_run structure.
7793
77948.15 KVM_CAP_S390_GMAP
7795----------------------
7796
7797:Architectures: s390
7798
7799This capability indicates that the user space memory used as guest mapping can
7800be anywhere in the user memory address space, as long as the memory slots are
7801aligned and sized to a segment (1MB) boundary.
7802
78038.16 KVM_CAP_S390_COW
7804---------------------
7805
7806:Architectures: s390
7807
7808This capability indicates that the user space memory used as guest mapping can
7809use copy-on-write semantics as well as dirty pages tracking via read-only page
7810tables.
7811
78128.17 KVM_CAP_S390_BPB
7813---------------------
7814
7815:Architectures: s390
7816
7817This capability indicates that kvm will implement the interfaces to handle
7818reset, migration and nested KVM for branch prediction blocking. The stfle
7819facility 82 should not be provided to the guest without this capability.
7820
78218.18 KVM_CAP_HYPERV_TLBFLUSH
7822----------------------------
7823
7824:Architectures: x86
7825
7826This capability indicates that KVM supports paravirtualized Hyper-V TLB Flush
7827hypercalls:
7828HvFlushVirtualAddressSpace, HvFlushVirtualAddressSpaceEx,
7829HvFlushVirtualAddressList, HvFlushVirtualAddressListEx.
7830
78318.19 KVM_CAP_ARM_INJECT_SERROR_ESR
7832----------------------------------
7833
7834:Architectures: arm64
7835
7836This capability indicates that userspace can specify (via the
7837KVM_SET_VCPU_EVENTS ioctl) the syndrome value reported to the guest when it
7838takes a virtual SError interrupt exception.
7839If KVM advertises this capability, userspace can only specify the ISS field for
7840the ESR syndrome. Other parts of the ESR, such as the EC are generated by the
7841CPU when the exception is taken. If this virtual SError is taken to EL1 using
7842AArch64, this value will be reported in the ISS field of ESR_ELx.
7843
7844See KVM_CAP_VCPU_EVENTS for more details.
7845
78468.20 KVM_CAP_HYPERV_SEND_IPI
7847----------------------------
7848
7849:Architectures: x86
7850
7851This capability indicates that KVM supports paravirtualized Hyper-V IPI send
7852hypercalls:
7853HvCallSendSyntheticClusterIpi, HvCallSendSyntheticClusterIpiEx.
7854
78558.21 KVM_CAP_HYPERV_DIRECT_TLBFLUSH
7856-----------------------------------
7857
7858:Architectures: x86
7859
7860This capability indicates that KVM running on top of Hyper-V hypervisor
7861enables Direct TLB flush for its guests meaning that TLB flush
7862hypercalls are handled by Level 0 hypervisor (Hyper-V) bypassing KVM.
7863Due to the different ABI for hypercall parameters between Hyper-V and
7864KVM, enabling this capability effectively disables all hypercall
7865handling by KVM (as some KVM hypercall may be mistakenly treated as TLB
7866flush hypercalls by Hyper-V) so userspace should disable KVM identification
7867in CPUID and only exposes Hyper-V identification. In this case, guest
7868thinks it's running on Hyper-V and only use Hyper-V hypercalls.
7869
78708.22 KVM_CAP_S390_VCPU_RESETS
7871-----------------------------
7872
7873:Architectures: s390
7874
7875This capability indicates that the KVM_S390_NORMAL_RESET and
7876KVM_S390_CLEAR_RESET ioctls are available.
7877
78788.23 KVM_CAP_S390_PROTECTED
7879---------------------------
7880
7881:Architectures: s390
7882
7883This capability indicates that the Ultravisor has been initialized and
7884KVM can therefore start protected VMs.
7885This capability governs the KVM_S390_PV_COMMAND ioctl and the
7886KVM_MP_STATE_LOAD MP_STATE. KVM_SET_MP_STATE can fail for protected
7887guests when the state change is invalid.
7888
78898.24 KVM_CAP_STEAL_TIME
7890-----------------------
7891
7892:Architectures: arm64, x86
7893
7894This capability indicates that KVM supports steal time accounting.
7895When steal time accounting is supported it may be enabled with
7896architecture-specific interfaces.  This capability and the architecture-
7897specific interfaces must be consistent, i.e. if one says the feature
7898is supported, than the other should as well and vice versa.  For arm64
7899see Documentation/virt/kvm/devices/vcpu.rst "KVM_ARM_VCPU_PVTIME_CTRL".
7900For x86 see Documentation/virt/kvm/x86/msr.rst "MSR_KVM_STEAL_TIME".
7901
79028.25 KVM_CAP_S390_DIAG318
7903-------------------------
7904
7905:Architectures: s390
7906
7907This capability enables a guest to set information about its control program
7908(i.e. guest kernel type and version). The information is helpful during
7909system/firmware service events, providing additional data about the guest
7910environments running on the machine.
7911
7912The information is associated with the DIAGNOSE 0x318 instruction, which sets
7913an 8-byte value consisting of a one-byte Control Program Name Code (CPNC) and
7914a 7-byte Control Program Version Code (CPVC). The CPNC determines what
7915environment the control program is running in (e.g. Linux, z/VM...), and the
7916CPVC is used for information specific to OS (e.g. Linux version, Linux
7917distribution...)
7918
7919If this capability is available, then the CPNC and CPVC can be synchronized
7920between KVM and userspace via the sync regs mechanism (KVM_SYNC_DIAG318).
7921
79228.26 KVM_CAP_X86_USER_SPACE_MSR
7923-------------------------------
7924
7925:Architectures: x86
7926
7927This capability indicates that KVM supports deflection of MSR reads and
7928writes to user space. It can be enabled on a VM level. If enabled, MSR
7929accesses that would usually trigger a #GP by KVM into the guest will
7930instead get bounced to user space through the KVM_EXIT_X86_RDMSR and
7931KVM_EXIT_X86_WRMSR exit notifications.
7932
79338.27 KVM_CAP_X86_MSR_FILTER
7934---------------------------
7935
7936:Architectures: x86
7937
7938This capability indicates that KVM supports that accesses to user defined MSRs
7939may be rejected. With this capability exposed, KVM exports new VM ioctl
7940KVM_X86_SET_MSR_FILTER which user space can call to specify bitmaps of MSR
7941ranges that KVM should deny access to.
7942
7943In combination with KVM_CAP_X86_USER_SPACE_MSR, this allows user space to
7944trap and emulate MSRs that are outside of the scope of KVM as well as
7945limit the attack surface on KVM's MSR emulation code.
7946
79478.28 KVM_CAP_ENFORCE_PV_FEATURE_CPUID
7948-------------------------------------
7949
7950Architectures: x86
7951
7952When enabled, KVM will disable paravirtual features provided to the
7953guest according to the bits in the KVM_CPUID_FEATURES CPUID leaf
7954(0x40000001). Otherwise, a guest may use the paravirtual features
7955regardless of what has actually been exposed through the CPUID leaf.
7956
79578.29 KVM_CAP_DIRTY_LOG_RING/KVM_CAP_DIRTY_LOG_RING_ACQ_REL
7958----------------------------------------------------------
7959
7960:Architectures: x86, arm64
7961:Parameters: args[0] - size of the dirty log ring
7962
7963KVM is capable of tracking dirty memory using ring buffers that are
7964mmaped into userspace; there is one dirty ring per vcpu.
7965
7966The dirty ring is available to userspace as an array of
7967``struct kvm_dirty_gfn``.  Each dirty entry it's defined as::
7968
7969  struct kvm_dirty_gfn {
7970          __u32 flags;
7971          __u32 slot; /* as_id | slot_id */
7972          __u64 offset;
7973  };
7974
7975The following values are defined for the flags field to define the
7976current state of the entry::
7977
7978  #define KVM_DIRTY_GFN_F_DIRTY           BIT(0)
7979  #define KVM_DIRTY_GFN_F_RESET           BIT(1)
7980  #define KVM_DIRTY_GFN_F_MASK            0x3
7981
7982Userspace should call KVM_ENABLE_CAP ioctl right after KVM_CREATE_VM
7983ioctl to enable this capability for the new guest and set the size of
7984the rings.  Enabling the capability is only allowed before creating any
7985vCPU, and the size of the ring must be a power of two.  The larger the
7986ring buffer, the less likely the ring is full and the VM is forced to
7987exit to userspace. The optimal size depends on the workload, but it is
7988recommended that it be at least 64 KiB (4096 entries).
7989
7990Just like for dirty page bitmaps, the buffer tracks writes to
7991all user memory regions for which the KVM_MEM_LOG_DIRTY_PAGES flag was
7992set in KVM_SET_USER_MEMORY_REGION.  Once a memory region is registered
7993with the flag set, userspace can start harvesting dirty pages from the
7994ring buffer.
7995
7996An entry in the ring buffer can be unused (flag bits ``00``),
7997dirty (flag bits ``01``) or harvested (flag bits ``1X``).  The
7998state machine for the entry is as follows::
7999
8000          dirtied         harvested        reset
8001     00 -----------> 01 -------------> 1X -------+
8002      ^                                          |
8003      |                                          |
8004      +------------------------------------------+
8005
8006To harvest the dirty pages, userspace accesses the mmaped ring buffer
8007to read the dirty GFNs.  If the flags has the DIRTY bit set (at this stage
8008the RESET bit must be cleared), then it means this GFN is a dirty GFN.
8009The userspace should harvest this GFN and mark the flags from state
8010``01b`` to ``1Xb`` (bit 0 will be ignored by KVM, but bit 1 must be set
8011to show that this GFN is harvested and waiting for a reset), and move
8012on to the next GFN.  The userspace should continue to do this until the
8013flags of a GFN have the DIRTY bit cleared, meaning that it has harvested
8014all the dirty GFNs that were available.
8015
8016Note that on weakly ordered architectures, userspace accesses to the
8017ring buffer (and more specifically the 'flags' field) must be ordered,
8018using load-acquire/store-release accessors when available, or any
8019other memory barrier that will ensure this ordering.
8020
8021It's not necessary for userspace to harvest the all dirty GFNs at once.
8022However it must collect the dirty GFNs in sequence, i.e., the userspace
8023program cannot skip one dirty GFN to collect the one next to it.
8024
8025After processing one or more entries in the ring buffer, userspace
8026calls the VM ioctl KVM_RESET_DIRTY_RINGS to notify the kernel about
8027it, so that the kernel will reprotect those collected GFNs.
8028Therefore, the ioctl must be called *before* reading the content of
8029the dirty pages.
8030
8031The dirty ring can get full.  When it happens, the KVM_RUN of the
8032vcpu will return with exit reason KVM_EXIT_DIRTY_LOG_FULL.
8033
8034The dirty ring interface has a major difference comparing to the
8035KVM_GET_DIRTY_LOG interface in that, when reading the dirty ring from
8036userspace, it's still possible that the kernel has not yet flushed the
8037processor's dirty page buffers into the kernel buffer (with dirty bitmaps, the
8038flushing is done by the KVM_GET_DIRTY_LOG ioctl).  To achieve that, one
8039needs to kick the vcpu out of KVM_RUN using a signal.  The resulting
8040vmexit ensures that all dirty GFNs are flushed to the dirty rings.
8041
8042NOTE: KVM_CAP_DIRTY_LOG_RING_ACQ_REL is the only capability that
8043should be exposed by weakly ordered architecture, in order to indicate
8044the additional memory ordering requirements imposed on userspace when
8045reading the state of an entry and mutating it from DIRTY to HARVESTED.
8046Architecture with TSO-like ordering (such as x86) are allowed to
8047expose both KVM_CAP_DIRTY_LOG_RING and KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8048to userspace.
8049
8050After enabling the dirty rings, the userspace needs to detect the
8051capability of KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP to see whether the
8052ring structures can be backed by per-slot bitmaps. With this capability
8053advertised, it means the architecture can dirty guest pages without
8054vcpu/ring context, so that some of the dirty information will still be
8055maintained in the bitmap structure. KVM_CAP_DIRTY_LOG_RING_WITH_BITMAP
8056can't be enabled if the capability of KVM_CAP_DIRTY_LOG_RING_ACQ_REL
8057hasn't been enabled, or any memslot has been existing.
8058
8059Note that the bitmap here is only a backup of the ring structure. The
8060use of the ring and bitmap combination is only beneficial if there is
8061only a very small amount of memory that is dirtied out of vcpu/ring
8062context. Otherwise, the stand-alone per-slot bitmap mechanism needs to
8063be considered.
8064
8065To collect dirty bits in the backup bitmap, userspace can use the same
8066KVM_GET_DIRTY_LOG ioctl. KVM_CLEAR_DIRTY_LOG isn't needed as long as all
8067the generation of the dirty bits is done in a single pass. Collecting
8068the dirty bitmap should be the very last thing that the VMM does before
8069considering the state as complete. VMM needs to ensure that the dirty
8070state is final and avoid missing dirty pages from another ioctl ordered
8071after the bitmap collection.
8072
8073NOTE: One example of using the backup bitmap is saving arm64 vgic/its
8074tables through KVM_DEV_ARM_{VGIC_GRP_CTRL, ITS_SAVE_TABLES} command on
8075KVM device "kvm-arm-vgic-its" when dirty ring is enabled.
8076
80778.30 KVM_CAP_XEN_HVM
8078--------------------
8079
8080:Architectures: x86
8081
8082This capability indicates the features that Xen supports for hosting Xen
8083PVHVM guests. Valid flags are::
8084
8085  #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR		(1 << 0)
8086  #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL		(1 << 1)
8087  #define KVM_XEN_HVM_CONFIG_SHARED_INFO		(1 << 2)
8088  #define KVM_XEN_HVM_CONFIG_RUNSTATE			(1 << 3)
8089  #define KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL		(1 << 4)
8090  #define KVM_XEN_HVM_CONFIG_EVTCHN_SEND		(1 << 5)
8091  #define KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG	(1 << 6)
8092
8093The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG
8094ioctl is available, for the guest to set its hypercall page.
8095
8096If KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL is also set, the same flag may also be
8097provided in the flags to KVM_XEN_HVM_CONFIG, without providing hypercall page
8098contents, to request that KVM generate hypercall page content automatically
8099and also enable interception of guest hypercalls with KVM_EXIT_XEN.
8100
8101The KVM_XEN_HVM_CONFIG_SHARED_INFO flag indicates the availability of the
8102KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and
8103KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors
8104for event channel upcalls when the evtchn_upcall_pending field of a vcpu's
8105vcpu_info is set.
8106
8107The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related
8108features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are
8109supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls.
8110
8111The KVM_XEN_HVM_CONFIG_EVTCHN_2LEVEL flag indicates that IRQ routing entries
8112of the type KVM_IRQ_ROUTING_XEN_EVTCHN are supported, with the priority
8113field set to indicate 2 level event channel delivery.
8114
8115The KVM_XEN_HVM_CONFIG_EVTCHN_SEND flag indicates that KVM supports
8116injecting event channel events directly into the guest with the
8117KVM_XEN_HVM_EVTCHN_SEND ioctl. It also indicates support for the
8118KVM_XEN_ATTR_TYPE_EVTCHN/XEN_VERSION HVM attributes and the
8119KVM_XEN_VCPU_ATTR_TYPE_VCPU_ID/TIMER/UPCALL_VECTOR vCPU attributes.
8120related to event channel delivery, timers, and the XENVER_version
8121interception.
8122
8123The KVM_XEN_HVM_CONFIG_RUNSTATE_UPDATE_FLAG flag indicates that KVM supports
8124the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute in the KVM_XEN_SET_ATTR
8125and KVM_XEN_GET_ATTR ioctls. This controls whether KVM will set the
8126XEN_RUNSTATE_UPDATE flag in guest memory mapped vcpu_runstate_info during
8127updates of the runstate information. Note that versions of KVM which support
8128the RUNSTATE feature above, but not thie RUNSTATE_UPDATE_FLAG feature, will
8129always set the XEN_RUNSTATE_UPDATE flag when updating the guest structure,
8130which is perhaps counterintuitive. When this flag is advertised, KVM will
8131behave more correctly, not using the XEN_RUNSTATE_UPDATE flag until/unless
8132specifically enabled (by the guest making the hypercall, causing the VMM
8133to enable the KVM_XEN_ATTR_TYPE_RUNSTATE_UPDATE_FLAG attribute).
8134
81358.31 KVM_CAP_PPC_MULTITCE
8136-------------------------
8137
8138:Capability: KVM_CAP_PPC_MULTITCE
8139:Architectures: ppc
8140:Type: vm
8141
8142This capability means the kernel is capable of handling hypercalls
8143H_PUT_TCE_INDIRECT and H_STUFF_TCE without passing those into the user
8144space. This significantly accelerates DMA operations for PPC KVM guests.
8145User space should expect that its handlers for these hypercalls
8146are not going to be called if user space previously registered LIOBN
8147in KVM (via KVM_CREATE_SPAPR_TCE or similar calls).
8148
8149In order to enable H_PUT_TCE_INDIRECT and H_STUFF_TCE use in the guest,
8150user space might have to advertise it for the guest. For example,
8151IBM pSeries (sPAPR) guest starts using them if "hcall-multi-tce" is
8152present in the "ibm,hypertas-functions" device-tree property.
8153
8154The hypercalls mentioned above may or may not be processed successfully
8155in the kernel based fast path. If they can not be handled by the kernel,
8156they will get passed on to user space. So user space still has to have
8157an implementation for these despite the in kernel acceleration.
8158
8159This capability is always enabled.
8160
81618.32 KVM_CAP_PTP_KVM
8162--------------------
8163
8164:Architectures: arm64
8165
8166This capability indicates that the KVM virtual PTP service is
8167supported in the host. A VMM can check whether the service is
8168available to the guest on migration.
8169
81708.33 KVM_CAP_HYPERV_ENFORCE_CPUID
8171---------------------------------
8172
8173Architectures: x86
8174
8175When enabled, KVM will disable emulated Hyper-V features provided to the
8176guest according to the bits Hyper-V CPUID feature leaves. Otherwise, all
8177currently implmented Hyper-V features are provided unconditionally when
8178Hyper-V identification is set in the HYPERV_CPUID_INTERFACE (0x40000001)
8179leaf.
8180
81818.34 KVM_CAP_EXIT_HYPERCALL
8182---------------------------
8183
8184:Capability: KVM_CAP_EXIT_HYPERCALL
8185:Architectures: x86
8186:Type: vm
8187
8188This capability, if enabled, will cause KVM to exit to userspace
8189with KVM_EXIT_HYPERCALL exit reason to process some hypercalls.
8190
8191Calling KVM_CHECK_EXTENSION for this capability will return a bitmask
8192of hypercalls that can be configured to exit to userspace.
8193Right now, the only such hypercall is KVM_HC_MAP_GPA_RANGE.
8194
8195The argument to KVM_ENABLE_CAP is also a bitmask, and must be a subset
8196of the result of KVM_CHECK_EXTENSION.  KVM will forward to userspace
8197the hypercalls whose corresponding bit is in the argument, and return
8198ENOSYS for the others.
8199
82008.35 KVM_CAP_PMU_CAPABILITY
8201---------------------------
8202
8203:Capability KVM_CAP_PMU_CAPABILITY
8204:Architectures: x86
8205:Type: vm
8206:Parameters: arg[0] is bitmask of PMU virtualization capabilities.
8207:Returns 0 on success, -EINVAL when arg[0] contains invalid bits
8208
8209This capability alters PMU virtualization in KVM.
8210
8211Calling KVM_CHECK_EXTENSION for this capability returns a bitmask of
8212PMU virtualization capabilities that can be adjusted on a VM.
8213
8214The argument to KVM_ENABLE_CAP is also a bitmask and selects specific
8215PMU virtualization capabilities to be applied to the VM.  This can
8216only be invoked on a VM prior to the creation of VCPUs.
8217
8218At this time, KVM_PMU_CAP_DISABLE is the only capability.  Setting
8219this capability will disable PMU virtualization for that VM.  Usermode
8220should adjust CPUID leaf 0xA to reflect that the PMU is disabled.
8221
82228.36 KVM_CAP_ARM_SYSTEM_SUSPEND
8223-------------------------------
8224
8225:Capability: KVM_CAP_ARM_SYSTEM_SUSPEND
8226:Architectures: arm64
8227:Type: vm
8228
8229When enabled, KVM will exit to userspace with KVM_EXIT_SYSTEM_EVENT of
8230type KVM_SYSTEM_EVENT_SUSPEND to process the guest suspend request.
8231
82328.37 KVM_CAP_S390_PROTECTED_DUMP
8233--------------------------------
8234
8235:Capability: KVM_CAP_S390_PROTECTED_DUMP
8236:Architectures: s390
8237:Type: vm
8238
8239This capability indicates that KVM and the Ultravisor support dumping
8240PV guests. The `KVM_PV_DUMP` command is available for the
8241`KVM_S390_PV_COMMAND` ioctl and the `KVM_PV_INFO` command provides
8242dump related UV data. Also the vcpu ioctl `KVM_S390_PV_CPU_COMMAND` is
8243available and supports the `KVM_PV_DUMP_CPU` subcommand.
8244
82458.38 KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
8246-------------------------------------
8247
8248:Capability: KVM_CAP_VM_DISABLE_NX_HUGE_PAGES
8249:Architectures: x86
8250:Type: vm
8251:Parameters: arg[0] must be 0.
8252:Returns: 0 on success, -EPERM if the userspace process does not
8253          have CAP_SYS_BOOT, -EINVAL if args[0] is not 0 or any vCPUs have been
8254          created.
8255
8256This capability disables the NX huge pages mitigation for iTLB MULTIHIT.
8257
8258The capability has no effect if the nx_huge_pages module parameter is not set.
8259
8260This capability may only be set before any vCPUs are created.
8261
82628.39 KVM_CAP_S390_CPU_TOPOLOGY
8263------------------------------
8264
8265:Capability: KVM_CAP_S390_CPU_TOPOLOGY
8266:Architectures: s390
8267:Type: vm
8268
8269This capability indicates that KVM will provide the S390 CPU Topology
8270facility which consist of the interpretation of the PTF instruction for
8271the function code 2 along with interception and forwarding of both the
8272PTF instruction with function codes 0 or 1 and the STSI(15,1,x)
8273instruction to the userland hypervisor.
8274
8275The stfle facility 11, CPU Topology facility, should not be indicated
8276to the guest without this capability.
8277
8278When this capability is present, KVM provides a new attribute group
8279on vm fd, KVM_S390_VM_CPU_TOPOLOGY.
8280This new attribute allows to get, set or clear the Modified Change
8281Topology Report (MTCR) bit of the SCA through the kvm_device_attr
8282structure.
8283
8284When getting the Modified Change Topology Report value, the attr->addr
8285must point to a byte where the value will be stored or retrieved from.
8286
82879. Known KVM API problems
8288=========================
8289
8290In some cases, KVM's API has some inconsistencies or common pitfalls
8291that userspace need to be aware of.  This section details some of
8292these issues.
8293
8294Most of them are architecture specific, so the section is split by
8295architecture.
8296
82979.1. x86
8298--------
8299
8300``KVM_GET_SUPPORTED_CPUID`` issues
8301^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8302
8303In general, ``KVM_GET_SUPPORTED_CPUID`` is designed so that it is possible
8304to take its result and pass it directly to ``KVM_SET_CPUID2``.  This section
8305documents some cases in which that requires some care.
8306
8307Local APIC features
8308~~~~~~~~~~~~~~~~~~~
8309
8310CPU[EAX=1]:ECX[21] (X2APIC) is reported by ``KVM_GET_SUPPORTED_CPUID``,
8311but it can only be enabled if ``KVM_CREATE_IRQCHIP`` or
8312``KVM_ENABLE_CAP(KVM_CAP_IRQCHIP_SPLIT)`` are used to enable in-kernel emulation of
8313the local APIC.
8314
8315The same is true for the ``KVM_FEATURE_PV_UNHALT`` paravirtualized feature.
8316
8317CPU[EAX=1]:ECX[24] (TSC_DEADLINE) is not reported by ``KVM_GET_SUPPORTED_CPUID``.
8318It can be enabled if ``KVM_CAP_TSC_DEADLINE_TIMER`` is present and the kernel
8319has enabled in-kernel emulation of the local APIC.
8320
8321CPU topology
8322~~~~~~~~~~~~
8323
8324Several CPUID values include topology information for the host CPU:
83250x0b and 0x1f for Intel systems, 0x8000001e for AMD systems.  Different
8326versions of KVM return different values for this information and userspace
8327should not rely on it.  Currently they return all zeroes.
8328
8329If userspace wishes to set up a guest topology, it should be careful that
8330the values of these three leaves differ for each CPU.  In particular,
8331the APIC ID is found in EDX for all subleaves of 0x0b and 0x1f, and in EAX
8332for 0x8000001e; the latter also encodes the core id and node id in bits
83337:0 of EBX and ECX respectively.
8334
8335Obsolete ioctls and capabilities
8336^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8337
8338KVM_CAP_DISABLE_QUIRKS does not let userspace know which quirks are actually
8339available.  Use ``KVM_CHECK_EXTENSION(KVM_CAP_DISABLE_QUIRKS2)`` instead if
8340available.
8341
8342Ordering of KVM_GET_*/KVM_SET_* ioctls
8343^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
8344
8345TBD
8346