• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1The Definitive KVM (Kernel-based Virtual Machine) API Documentation
2===================================================================
3
41. General description
5----------------------
6
7The kvm API is a set of ioctls that are issued to control various aspects
8of a virtual machine.  The ioctls belong to three classes
9
10 - System ioctls: These query and set global attributes which affect the
11   whole kvm subsystem.  In addition a system ioctl is used to create
12   virtual machines
13
14 - VM ioctls: These query and set attributes that affect an entire virtual
15   machine, for example memory layout.  In addition a VM ioctl is used to
16   create virtual cpus (vcpus) and devices.
17
18   Only run VM ioctls from the same process (address space) that was used
19   to create the VM.
20
21 - vcpu ioctls: These query and set attributes that control the operation
22   of a single virtual cpu.
23
24   Only run vcpu ioctls from the same thread that was used to create the
25   vcpu.
26
27 - device ioctls: These query and set attributes that control the operation
28   of a single device.
29
30   device ioctls must be issued from the same process (address space) that
31   was used to create the VM.
32
332. File descriptors
34-------------------
35
36The kvm API is centered around file descriptors.  An initial
37open("/dev/kvm") obtains a handle to the kvm subsystem; this handle
38can be used to issue system ioctls.  A KVM_CREATE_VM ioctl on this
39handle will create a VM file descriptor which can be used to issue VM
40ioctls.  A KVM_CREATE_VCPU or KVM_CREATE_DEVICE ioctl on a VM fd will
41create a virtual cpu or device and return a file descriptor pointing to
42the new resource.  Finally, ioctls on a vcpu or device fd can be used
43to control the vcpu or device.  For vcpus, this includes the important
44task of actually running guest code.
45
46In general file descriptors can be migrated among processes by means
47of fork() and the SCM_RIGHTS facility of unix domain socket.  These
48kinds of tricks are explicitly not supported by kvm.  While they will
49not cause harm to the host, their actual behavior is not guaranteed by
50the API.  The only supported use is one virtual machine per process,
51and one vcpu per thread.
52
53
543. Extensions
55-------------
56
57As of Linux 2.6.22, the KVM ABI has been stabilized: no backward
58incompatible change are allowed.  However, there is an extension
59facility that allows backward-compatible extensions to the API to be
60queried and used.
61
62The extension mechanism is not based on the Linux version number.
63Instead, kvm defines extension identifiers and a facility to query
64whether a particular extension identifier is available.  If it is, a
65set of ioctls is available for application use.
66
67
684. API description
69------------------
70
71This section describes ioctls that can be used to control kvm guests.
72For each ioctl, the following information is provided along with a
73description:
74
75  Capability: which KVM extension provides this ioctl.  Can be 'basic',
76      which means that is will be provided by any kernel that supports
77      API version 12 (see section 4.1), a KVM_CAP_xyz constant, which
78      means availability needs to be checked with KVM_CHECK_EXTENSION
79      (see section 4.4), or 'none' which means that while not all kernels
80      support this ioctl, there's no capability bit to check its
81      availability: for kernels that don't support the ioctl,
82      the ioctl returns -ENOTTY.
83
84  Architectures: which instruction set architectures provide this ioctl.
85      x86 includes both i386 and x86_64.
86
87  Type: system, vm, or vcpu.
88
89  Parameters: what parameters are accepted by the ioctl.
90
91  Returns: the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
92      are not detailed, but errors with specific meanings are.
93
94
954.1 KVM_GET_API_VERSION
96
97Capability: basic
98Architectures: all
99Type: system ioctl
100Parameters: none
101Returns: the constant KVM_API_VERSION (=12)
102
103This identifies the API version as the stable kvm API. It is not
104expected that this number will change.  However, Linux 2.6.20 and
1052.6.21 report earlier versions; these are not documented and not
106supported.  Applications should refuse to run if KVM_GET_API_VERSION
107returns a value other than 12.  If this check passes, all ioctls
108described as 'basic' will be available.
109
110
1114.2 KVM_CREATE_VM
112
113Capability: basic
114Architectures: all
115Type: system ioctl
116Parameters: machine type identifier (KVM_VM_*)
117Returns: a VM fd that can be used to control the new virtual machine.
118
119The new VM has no virtual cpus and no memory.  An mmap() of a VM fd
120will access the virtual machine's physical address space; offset zero
121corresponds to guest physical address zero.  Use of mmap() on a VM fd
122is discouraged if userspace memory allocation (KVM_CAP_USER_MEMORY) is
123available.
124You most certainly want to use 0 as machine type.
125
126In order to create user controlled virtual machines on S390, check
127KVM_CAP_S390_UCONTROL and use the flag KVM_VM_S390_UCONTROL as
128privileged user (CAP_SYS_ADMIN).
129
130
1314.3 KVM_GET_MSR_INDEX_LIST
132
133Capability: basic
134Architectures: x86
135Type: system
136Parameters: struct kvm_msr_list (in/out)
137Returns: 0 on success; -1 on error
138Errors:
139  E2BIG:     the msr index list is to be to fit in the array specified by
140             the user.
141
142struct kvm_msr_list {
143	__u32 nmsrs; /* number of msrs in entries */
144	__u32 indices[0];
145};
146
147This ioctl returns the guest msrs that are supported.  The list varies
148by kvm version and host processor, but does not change otherwise.  The
149user fills in the size of the indices array in nmsrs, and in return
150kvm adjusts nmsrs to reflect the actual number of msrs and fills in
151the indices array with their numbers.
152
153Note: if kvm indicates supports MCE (KVM_CAP_MCE), then the MCE bank MSRs are
154not returned in the MSR list, as different vcpus can have a different number
155of banks, as set via the KVM_X86_SETUP_MCE ioctl.
156
157
1584.4 KVM_CHECK_EXTENSION
159
160Capability: basic, KVM_CAP_CHECK_EXTENSION_VM for vm ioctl
161Architectures: all
162Type: system ioctl, vm ioctl
163Parameters: extension identifier (KVM_CAP_*)
164Returns: 0 if unsupported; 1 (or some other positive integer) if supported
165
166The API allows the application to query about extensions to the core
167kvm API.  Userspace passes an extension identifier (an integer) and
168receives an integer that describes the extension availability.
169Generally 0 means no and 1 means yes, but some extensions may report
170additional information in the integer return value.
171
172Based on their initialization different VMs may have different capabilities.
173It is thus encouraged to use the vm ioctl to query for capabilities (available
174with KVM_CAP_CHECK_EXTENSION_VM on the vm fd)
175
1764.5 KVM_GET_VCPU_MMAP_SIZE
177
178Capability: basic
179Architectures: all
180Type: system ioctl
181Parameters: none
182Returns: size of vcpu mmap area, in bytes
183
184The KVM_RUN ioctl (cf.) communicates with userspace via a shared
185memory region.  This ioctl returns the size of that region.  See the
186KVM_RUN documentation for details.
187
188
1894.6 KVM_SET_MEMORY_REGION
190
191Capability: basic
192Architectures: all
193Type: vm ioctl
194Parameters: struct kvm_memory_region (in)
195Returns: 0 on success, -1 on error
196
197This ioctl is obsolete and has been removed.
198
199
2004.7 KVM_CREATE_VCPU
201
202Capability: basic
203Architectures: all
204Type: vm ioctl
205Parameters: vcpu id (apic id on x86)
206Returns: vcpu fd on success, -1 on error
207
208This API adds a vcpu to a virtual machine.  The vcpu id is a small integer
209in the range [0, max_vcpus).
210
211The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of
212the KVM_CHECK_EXTENSION ioctl() at run-time.
213The maximum possible value for max_vcpus can be retrieved using the
214KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time.
215
216If the KVM_CAP_NR_VCPUS does not exist, you should assume that max_vcpus is 4
217cpus max.
218If the KVM_CAP_MAX_VCPUS does not exist, you should assume that max_vcpus is
219same as the value returned from KVM_CAP_NR_VCPUS.
220
221On powerpc using book3s_hv mode, the vcpus are mapped onto virtual
222threads in one or more virtual CPU cores.  (This is because the
223hardware requires all the hardware threads in a CPU core to be in the
224same partition.)  The KVM_CAP_PPC_SMT capability indicates the number
225of vcpus per virtual core (vcore).  The vcore id is obtained by
226dividing the vcpu id by the number of vcpus per vcore.  The vcpus in a
227given vcore will always be in the same physical core as each other
228(though that might be a different physical core from time to time).
229Userspace can control the threading (SMT) mode of the guest by its
230allocation of vcpu ids.  For example, if userspace wants
231single-threaded guest vcpus, it should make all vcpu ids be a multiple
232of the number of vcpus per vcore.
233
234For virtual cpus that have been created with S390 user controlled virtual
235machines, the resulting vcpu fd can be memory mapped at page offset
236KVM_S390_SIE_PAGE_OFFSET in order to obtain a memory map of the virtual
237cpu's hardware control block.
238
239
2404.8 KVM_GET_DIRTY_LOG (vm ioctl)
241
242Capability: basic
243Architectures: x86
244Type: vm ioctl
245Parameters: struct kvm_dirty_log (in/out)
246Returns: 0 on success, -1 on error
247
248/* for KVM_GET_DIRTY_LOG */
249struct kvm_dirty_log {
250	__u32 slot;
251	__u32 padding;
252	union {
253		void __user *dirty_bitmap; /* one bit per page */
254		__u64 padding;
255	};
256};
257
258Given a memory slot, return a bitmap containing any pages dirtied
259since the last call to this ioctl.  Bit 0 is the first page in the
260memory slot.  Ensure the entire structure is cleared to avoid padding
261issues.
262
263If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 specifies
264the address space for which you want to return the dirty bitmap.
265They must be less than the value that KVM_CHECK_EXTENSION returns for
266the KVM_CAP_MULTI_ADDRESS_SPACE capability.
267
268
2694.9 KVM_SET_MEMORY_ALIAS
270
271Capability: basic
272Architectures: x86
273Type: vm ioctl
274Parameters: struct kvm_memory_alias (in)
275Returns: 0 (success), -1 (error)
276
277This ioctl is obsolete and has been removed.
278
279
2804.10 KVM_RUN
281
282Capability: basic
283Architectures: all
284Type: vcpu ioctl
285Parameters: none
286Returns: 0 on success, -1 on error
287Errors:
288  EINTR:     an unmasked signal is pending
289
290This ioctl is used to run a guest virtual cpu.  While there are no
291explicit parameters, there is an implicit parameter block that can be
292obtained by mmap()ing the vcpu fd at offset 0, with the size given by
293KVM_GET_VCPU_MMAP_SIZE.  The parameter block is formatted as a 'struct
294kvm_run' (see below).
295
296
2974.11 KVM_GET_REGS
298
299Capability: basic
300Architectures: all except ARM, arm64
301Type: vcpu ioctl
302Parameters: struct kvm_regs (out)
303Returns: 0 on success, -1 on error
304
305Reads the general purpose registers from the vcpu.
306
307/* x86 */
308struct kvm_regs {
309	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
310	__u64 rax, rbx, rcx, rdx;
311	__u64 rsi, rdi, rsp, rbp;
312	__u64 r8,  r9,  r10, r11;
313	__u64 r12, r13, r14, r15;
314	__u64 rip, rflags;
315};
316
317/* mips */
318struct kvm_regs {
319	/* out (KVM_GET_REGS) / in (KVM_SET_REGS) */
320	__u64 gpr[32];
321	__u64 hi;
322	__u64 lo;
323	__u64 pc;
324};
325
326
3274.12 KVM_SET_REGS
328
329Capability: basic
330Architectures: all except ARM, arm64
331Type: vcpu ioctl
332Parameters: struct kvm_regs (in)
333Returns: 0 on success, -1 on error
334
335Writes the general purpose registers into the vcpu.
336
337See KVM_GET_REGS for the data structure.
338
339
3404.13 KVM_GET_SREGS
341
342Capability: basic
343Architectures: x86, ppc
344Type: vcpu ioctl
345Parameters: struct kvm_sregs (out)
346Returns: 0 on success, -1 on error
347
348Reads special registers from the vcpu.
349
350/* x86 */
351struct kvm_sregs {
352	struct kvm_segment cs, ds, es, fs, gs, ss;
353	struct kvm_segment tr, ldt;
354	struct kvm_dtable gdt, idt;
355	__u64 cr0, cr2, cr3, cr4, cr8;
356	__u64 efer;
357	__u64 apic_base;
358	__u64 interrupt_bitmap[(KVM_NR_INTERRUPTS + 63) / 64];
359};
360
361/* ppc -- see arch/powerpc/include/uapi/asm/kvm.h */
362
363interrupt_bitmap is a bitmap of pending external interrupts.  At most
364one bit may be set.  This interrupt has been acknowledged by the APIC
365but not yet injected into the cpu core.
366
367
3684.14 KVM_SET_SREGS
369
370Capability: basic
371Architectures: x86, ppc
372Type: vcpu ioctl
373Parameters: struct kvm_sregs (in)
374Returns: 0 on success, -1 on error
375
376Writes special registers into the vcpu.  See KVM_GET_SREGS for the
377data structures.
378
379
3804.15 KVM_TRANSLATE
381
382Capability: basic
383Architectures: x86
384Type: vcpu ioctl
385Parameters: struct kvm_translation (in/out)
386Returns: 0 on success, -1 on error
387
388Translates a virtual address according to the vcpu's current address
389translation mode.
390
391struct kvm_translation {
392	/* in */
393	__u64 linear_address;
394
395	/* out */
396	__u64 physical_address;
397	__u8  valid;
398	__u8  writeable;
399	__u8  usermode;
400	__u8  pad[5];
401};
402
403
4044.16 KVM_INTERRUPT
405
406Capability: basic
407Architectures: x86, ppc, mips
408Type: vcpu ioctl
409Parameters: struct kvm_interrupt (in)
410Returns: 0 on success, negative on failure.
411
412Queues a hardware interrupt vector to be injected.
413
414/* for KVM_INTERRUPT */
415struct kvm_interrupt {
416	/* in */
417	__u32 irq;
418};
419
420X86:
421
422Returns: 0 on success,
423	 -EEXIST if an interrupt is already enqueued
424	 -EINVAL the the irq number is invalid
425	 -ENXIO if the PIC is in the kernel
426	 -EFAULT if the pointer is invalid
427
428Note 'irq' is an interrupt vector, not an interrupt pin or line. This
429ioctl is useful if the in-kernel PIC is not used.
430
431PPC:
432
433Queues an external interrupt to be injected. This ioctl is overleaded
434with 3 different irq values:
435
436a) KVM_INTERRUPT_SET
437
438  This injects an edge type external interrupt into the guest once it's ready
439  to receive interrupts. When injected, the interrupt is done.
440
441b) KVM_INTERRUPT_UNSET
442
443  This unsets any pending interrupt.
444
445  Only available with KVM_CAP_PPC_UNSET_IRQ.
446
447c) KVM_INTERRUPT_SET_LEVEL
448
449  This injects a level type external interrupt into the guest context. The
450  interrupt stays pending until a specific ioctl with KVM_INTERRUPT_UNSET
451  is triggered.
452
453  Only available with KVM_CAP_PPC_IRQ_LEVEL.
454
455Note that any value for 'irq' other than the ones stated above is invalid
456and incurs unexpected behavior.
457
458MIPS:
459
460Queues an external interrupt to be injected into the virtual CPU. A negative
461interrupt number dequeues the interrupt.
462
463
4644.17 KVM_DEBUG_GUEST
465
466Capability: basic
467Architectures: none
468Type: vcpu ioctl
469Parameters: none)
470Returns: -1 on error
471
472Support for this has been removed.  Use KVM_SET_GUEST_DEBUG instead.
473
474
4754.18 KVM_GET_MSRS
476
477Capability: basic
478Architectures: x86
479Type: vcpu ioctl
480Parameters: struct kvm_msrs (in/out)
481Returns: 0 on success, -1 on error
482
483Reads model-specific registers from the vcpu.  Supported msr indices can
484be obtained using KVM_GET_MSR_INDEX_LIST.
485
486struct kvm_msrs {
487	__u32 nmsrs; /* number of msrs in entries */
488	__u32 pad;
489
490	struct kvm_msr_entry entries[0];
491};
492
493struct kvm_msr_entry {
494	__u32 index;
495	__u32 reserved;
496	__u64 data;
497};
498
499Application code should set the 'nmsrs' member (which indicates the
500size of the entries array) and the 'index' member of each array entry.
501kvm will fill in the 'data' member.
502
503
5044.19 KVM_SET_MSRS
505
506Capability: basic
507Architectures: x86
508Type: vcpu ioctl
509Parameters: struct kvm_msrs (in)
510Returns: 0 on success, -1 on error
511
512Writes model-specific registers to the vcpu.  See KVM_GET_MSRS for the
513data structures.
514
515Application code should set the 'nmsrs' member (which indicates the
516size of the entries array), and the 'index' and 'data' members of each
517array entry.
518
519
5204.20 KVM_SET_CPUID
521
522Capability: basic
523Architectures: x86
524Type: vcpu ioctl
525Parameters: struct kvm_cpuid (in)
526Returns: 0 on success, -1 on error
527
528Defines the vcpu responses to the cpuid instruction.  Applications
529should use the KVM_SET_CPUID2 ioctl if available.
530
531
532struct kvm_cpuid_entry {
533	__u32 function;
534	__u32 eax;
535	__u32 ebx;
536	__u32 ecx;
537	__u32 edx;
538	__u32 padding;
539};
540
541/* for KVM_SET_CPUID */
542struct kvm_cpuid {
543	__u32 nent;
544	__u32 padding;
545	struct kvm_cpuid_entry entries[0];
546};
547
548
5494.21 KVM_SET_SIGNAL_MASK
550
551Capability: basic
552Architectures: all
553Type: vcpu ioctl
554Parameters: struct kvm_signal_mask (in)
555Returns: 0 on success, -1 on error
556
557Defines which signals are blocked during execution of KVM_RUN.  This
558signal mask temporarily overrides the threads signal mask.  Any
559unblocked signal received (except SIGKILL and SIGSTOP, which retain
560their traditional behaviour) will cause KVM_RUN to return with -EINTR.
561
562Note the signal will only be delivered if not blocked by the original
563signal mask.
564
565/* for KVM_SET_SIGNAL_MASK */
566struct kvm_signal_mask {
567	__u32 len;
568	__u8  sigset[0];
569};
570
571
5724.22 KVM_GET_FPU
573
574Capability: basic
575Architectures: x86
576Type: vcpu ioctl
577Parameters: struct kvm_fpu (out)
578Returns: 0 on success, -1 on error
579
580Reads the floating point state from the vcpu.
581
582/* for KVM_GET_FPU and KVM_SET_FPU */
583struct kvm_fpu {
584	__u8  fpr[8][16];
585	__u16 fcw;
586	__u16 fsw;
587	__u8  ftwx;  /* in fxsave format */
588	__u8  pad1;
589	__u16 last_opcode;
590	__u64 last_ip;
591	__u64 last_dp;
592	__u8  xmm[16][16];
593	__u32 mxcsr;
594	__u32 pad2;
595};
596
597
5984.23 KVM_SET_FPU
599
600Capability: basic
601Architectures: x86
602Type: vcpu ioctl
603Parameters: struct kvm_fpu (in)
604Returns: 0 on success, -1 on error
605
606Writes the floating point state to the vcpu.
607
608/* for KVM_GET_FPU and KVM_SET_FPU */
609struct kvm_fpu {
610	__u8  fpr[8][16];
611	__u16 fcw;
612	__u16 fsw;
613	__u8  ftwx;  /* in fxsave format */
614	__u8  pad1;
615	__u16 last_opcode;
616	__u64 last_ip;
617	__u64 last_dp;
618	__u8  xmm[16][16];
619	__u32 mxcsr;
620	__u32 pad2;
621};
622
623
6244.24 KVM_CREATE_IRQCHIP
625
626Capability: KVM_CAP_IRQCHIP, KVM_CAP_S390_IRQCHIP (s390)
627Architectures: x86, ARM, arm64, s390
628Type: vm ioctl
629Parameters: none
630Returns: 0 on success, -1 on error
631
632Creates an interrupt controller model in the kernel.
633On x86, creates a virtual ioapic, a virtual PIC (two PICs, nested), and sets up
634future vcpus to have a local APIC.  IRQ routing for GSIs 0-15 is set to both
635PIC and IOAPIC; GSI 16-23 only go to the IOAPIC.
636On ARM/arm64, a GICv2 is created. Any other GIC versions require the usage of
637KVM_CREATE_DEVICE, which also supports creating a GICv2.  Using
638KVM_CREATE_DEVICE is preferred over KVM_CREATE_IRQCHIP for GICv2.
639On s390, a dummy irq routing table is created.
640
641Note that on s390 the KVM_CAP_S390_IRQCHIP vm capability needs to be enabled
642before KVM_CREATE_IRQCHIP can be used.
643
644
6454.25 KVM_IRQ_LINE
646
647Capability: KVM_CAP_IRQCHIP
648Architectures: x86, arm, arm64
649Type: vm ioctl
650Parameters: struct kvm_irq_level
651Returns: 0 on success, -1 on error
652
653Sets the level of a GSI input to the interrupt controller model in the kernel.
654On some architectures it is required that an interrupt controller model has
655been previously created with KVM_CREATE_IRQCHIP.  Note that edge-triggered
656interrupts require the level to be set to 1 and then back to 0.
657
658On real hardware, interrupt pins can be active-low or active-high.  This
659does not matter for the level field of struct kvm_irq_level: 1 always
660means active (asserted), 0 means inactive (deasserted).
661
662x86 allows the operating system to program the interrupt polarity
663(active-low/active-high) for level-triggered interrupts, and KVM used
664to consider the polarity.  However, due to bitrot in the handling of
665active-low interrupts, the above convention is now valid on x86 too.
666This is signaled by KVM_CAP_X86_IOAPIC_POLARITY_IGNORED.  Userspace
667should not present interrupts to the guest as active-low unless this
668capability is present (or unless it is not using the in-kernel irqchip,
669of course).
670
671
672ARM/arm64 can signal an interrupt either at the CPU level, or at the
673in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
674use PPIs designated for specific cpus.  The irq field is interpreted
675like this:
676
677  bits:  | 31 ... 24 | 23  ... 16 | 15    ...    0 |
678  field: | irq_type  | vcpu_index |     irq_id     |
679
680The irq_type field has the following values:
681- irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
682- irq_type[1]: in-kernel GIC: SPI, irq_id between 32 and 1019 (incl.)
683               (the vcpu_index field is ignored)
684- irq_type[2]: in-kernel GIC: PPI, irq_id between 16 and 31 (incl.)
685
686(The irq_id field thus corresponds nicely to the IRQ ID in the ARM GIC specs)
687
688In both cases, level is used to assert/deassert the line.
689
690struct kvm_irq_level {
691	union {
692		__u32 irq;     /* GSI */
693		__s32 status;  /* not used for KVM_IRQ_LEVEL */
694	};
695	__u32 level;           /* 0 or 1 */
696};
697
698
6994.26 KVM_GET_IRQCHIP
700
701Capability: KVM_CAP_IRQCHIP
702Architectures: x86
703Type: vm ioctl
704Parameters: struct kvm_irqchip (in/out)
705Returns: 0 on success, -1 on error
706
707Reads the state of a kernel interrupt controller created with
708KVM_CREATE_IRQCHIP into a buffer provided by the caller.
709
710struct kvm_irqchip {
711	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
712	__u32 pad;
713        union {
714		char dummy[512];  /* reserving space */
715		struct kvm_pic_state pic;
716		struct kvm_ioapic_state ioapic;
717	} chip;
718};
719
720
7214.27 KVM_SET_IRQCHIP
722
723Capability: KVM_CAP_IRQCHIP
724Architectures: x86
725Type: vm ioctl
726Parameters: struct kvm_irqchip (in)
727Returns: 0 on success, -1 on error
728
729Sets the state of a kernel interrupt controller created with
730KVM_CREATE_IRQCHIP from a buffer provided by the caller.
731
732struct kvm_irqchip {
733	__u32 chip_id;  /* 0 = PIC1, 1 = PIC2, 2 = IOAPIC */
734	__u32 pad;
735        union {
736		char dummy[512];  /* reserving space */
737		struct kvm_pic_state pic;
738		struct kvm_ioapic_state ioapic;
739	} chip;
740};
741
742
7434.28 KVM_XEN_HVM_CONFIG
744
745Capability: KVM_CAP_XEN_HVM
746Architectures: x86
747Type: vm ioctl
748Parameters: struct kvm_xen_hvm_config (in)
749Returns: 0 on success, -1 on error
750
751Sets the MSR that the Xen HVM guest uses to initialize its hypercall
752page, and provides the starting address and size of the hypercall
753blobs in userspace.  When the guest writes the MSR, kvm copies one
754page of a blob (32- or 64-bit, depending on the vcpu mode) to guest
755memory.
756
757struct kvm_xen_hvm_config {
758	__u32 flags;
759	__u32 msr;
760	__u64 blob_addr_32;
761	__u64 blob_addr_64;
762	__u8 blob_size_32;
763	__u8 blob_size_64;
764	__u8 pad2[30];
765};
766
767
7684.29 KVM_GET_CLOCK
769
770Capability: KVM_CAP_ADJUST_CLOCK
771Architectures: x86
772Type: vm ioctl
773Parameters: struct kvm_clock_data (out)
774Returns: 0 on success, -1 on error
775
776Gets the current timestamp of kvmclock as seen by the current guest. In
777conjunction with KVM_SET_CLOCK, it is used to ensure monotonicity on scenarios
778such as migration.
779
780struct kvm_clock_data {
781	__u64 clock;  /* kvmclock current value */
782	__u32 flags;
783	__u32 pad[9];
784};
785
786
7874.30 KVM_SET_CLOCK
788
789Capability: KVM_CAP_ADJUST_CLOCK
790Architectures: x86
791Type: vm ioctl
792Parameters: struct kvm_clock_data (in)
793Returns: 0 on success, -1 on error
794
795Sets the current timestamp of kvmclock to the value specified in its parameter.
796In conjunction with KVM_GET_CLOCK, it is used to ensure monotonicity on scenarios
797such as migration.
798
799struct kvm_clock_data {
800	__u64 clock;  /* kvmclock current value */
801	__u32 flags;
802	__u32 pad[9];
803};
804
805
8064.31 KVM_GET_VCPU_EVENTS
807
808Capability: KVM_CAP_VCPU_EVENTS
809Extended by: KVM_CAP_INTR_SHADOW
810Architectures: x86
811Type: vm ioctl
812Parameters: struct kvm_vcpu_event (out)
813Returns: 0 on success, -1 on error
814
815Gets currently pending exceptions, interrupts, and NMIs as well as related
816states of the vcpu.
817
818struct kvm_vcpu_events {
819	struct {
820		__u8 injected;
821		__u8 nr;
822		__u8 has_error_code;
823		__u8 pad;
824		__u32 error_code;
825	} exception;
826	struct {
827		__u8 injected;
828		__u8 nr;
829		__u8 soft;
830		__u8 shadow;
831	} interrupt;
832	struct {
833		__u8 injected;
834		__u8 pending;
835		__u8 masked;
836		__u8 pad;
837	} nmi;
838	__u32 sipi_vector;
839	__u32 flags;
840	struct {
841		__u8 smm;
842		__u8 pending;
843		__u8 smm_inside_nmi;
844		__u8 latched_init;
845	} smi;
846};
847
848Only two fields are defined in the flags field:
849
850- KVM_VCPUEVENT_VALID_SHADOW may be set in the flags field to signal that
851  interrupt.shadow contains a valid state.
852
853- KVM_VCPUEVENT_VALID_SMM may be set in the flags field to signal that
854  smi contains a valid state.
855
8564.32 KVM_SET_VCPU_EVENTS
857
858Capability: KVM_CAP_VCPU_EVENTS
859Extended by: KVM_CAP_INTR_SHADOW
860Architectures: x86
861Type: vm ioctl
862Parameters: struct kvm_vcpu_event (in)
863Returns: 0 on success, -1 on error
864
865Set pending exceptions, interrupts, and NMIs as well as related states of the
866vcpu.
867
868See KVM_GET_VCPU_EVENTS for the data structure.
869
870Fields that may be modified asynchronously by running VCPUs can be excluded
871from the update. These fields are nmi.pending, sipi_vector, smi.smm,
872smi.pending. Keep the corresponding bits in the flags field cleared to
873suppress overwriting the current in-kernel state. The bits are:
874
875KVM_VCPUEVENT_VALID_NMI_PENDING - transfer nmi.pending to the kernel
876KVM_VCPUEVENT_VALID_SIPI_VECTOR - transfer sipi_vector
877KVM_VCPUEVENT_VALID_SMM         - transfer the smi sub-struct.
878
879If KVM_CAP_INTR_SHADOW is available, KVM_VCPUEVENT_VALID_SHADOW can be set in
880the flags field to signal that interrupt.shadow contains a valid state and
881shall be written into the VCPU.
882
883KVM_VCPUEVENT_VALID_SMM can only be set if KVM_CAP_X86_SMM is available.
884
885
8864.33 KVM_GET_DEBUGREGS
887
888Capability: KVM_CAP_DEBUGREGS
889Architectures: x86
890Type: vm ioctl
891Parameters: struct kvm_debugregs (out)
892Returns: 0 on success, -1 on error
893
894Reads debug registers from the vcpu.
895
896struct kvm_debugregs {
897	__u64 db[4];
898	__u64 dr6;
899	__u64 dr7;
900	__u64 flags;
901	__u64 reserved[9];
902};
903
904
9054.34 KVM_SET_DEBUGREGS
906
907Capability: KVM_CAP_DEBUGREGS
908Architectures: x86
909Type: vm ioctl
910Parameters: struct kvm_debugregs (in)
911Returns: 0 on success, -1 on error
912
913Writes debug registers into the vcpu.
914
915See KVM_GET_DEBUGREGS for the data structure. The flags field is unused
916yet and must be cleared on entry.
917
918
9194.35 KVM_SET_USER_MEMORY_REGION
920
921Capability: KVM_CAP_USER_MEM
922Architectures: all
923Type: vm ioctl
924Parameters: struct kvm_userspace_memory_region (in)
925Returns: 0 on success, -1 on error
926
927struct kvm_userspace_memory_region {
928	__u32 slot;
929	__u32 flags;
930	__u64 guest_phys_addr;
931	__u64 memory_size; /* bytes */
932	__u64 userspace_addr; /* start of the userspace allocated memory */
933};
934
935/* for kvm_memory_region::flags */
936#define KVM_MEM_LOG_DIRTY_PAGES	(1UL << 0)
937#define KVM_MEM_READONLY	(1UL << 1)
938
939This ioctl allows the user to create or modify a guest physical memory
940slot.  When changing an existing slot, it may be moved in the guest
941physical memory space, or its flags may be modified.  It may not be
942resized.  Slots may not overlap in guest physical address space.
943
944If KVM_CAP_MULTI_ADDRESS_SPACE is available, bits 16-31 of "slot"
945specifies the address space which is being modified.  They must be
946less than the value that KVM_CHECK_EXTENSION returns for the
947KVM_CAP_MULTI_ADDRESS_SPACE capability.  Slots in separate address spaces
948are unrelated; the restriction on overlapping slots only applies within
949each address space.
950
951Memory for the region is taken starting at the address denoted by the
952field userspace_addr, which must point at user addressable memory for
953the entire memory slot size.  Any object may back this memory, including
954anonymous memory, ordinary files, and hugetlbfs.
955
956It is recommended that the lower 21 bits of guest_phys_addr and userspace_addr
957be identical.  This allows large pages in the guest to be backed by large
958pages in the host.
959
960The flags field supports two flags: KVM_MEM_LOG_DIRTY_PAGES and
961KVM_MEM_READONLY.  The former can be set to instruct KVM to keep track of
962writes to memory within the slot.  See KVM_GET_DIRTY_LOG ioctl to know how to
963use it.  The latter can be set, if KVM_CAP_READONLY_MEM capability allows it,
964to make a new slot read-only.  In this case, writes to this memory will be
965posted to userspace as KVM_EXIT_MMIO exits.
966
967When the KVM_CAP_SYNC_MMU capability is available, changes in the backing of
968the memory region are automatically reflected into the guest.  For example, an
969mmap() that affects the region will be made visible immediately.  Another
970example is madvise(MADV_DROP).
971
972It is recommended to use this API instead of the KVM_SET_MEMORY_REGION ioctl.
973The KVM_SET_MEMORY_REGION does not allow fine grained control over memory
974allocation and is deprecated.
975
976
9774.36 KVM_SET_TSS_ADDR
978
979Capability: KVM_CAP_SET_TSS_ADDR
980Architectures: x86
981Type: vm ioctl
982Parameters: unsigned long tss_address (in)
983Returns: 0 on success, -1 on error
984
985This ioctl defines the physical address of a three-page region in the guest
986physical address space.  The region must be within the first 4GB of the
987guest physical address space and must not conflict with any memory slot
988or any mmio address.  The guest may malfunction if it accesses this memory
989region.
990
991This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
992because of a quirk in the virtualization implementation (see the internals
993documentation when it pops into existence).
994
995
9964.37 KVM_ENABLE_CAP
997
998Capability: KVM_CAP_ENABLE_CAP, KVM_CAP_ENABLE_CAP_VM
999Architectures: x86 (only KVM_CAP_ENABLE_CAP_VM),
1000	       mips (only KVM_CAP_ENABLE_CAP), ppc, s390
1001Type: vcpu ioctl, vm ioctl (with KVM_CAP_ENABLE_CAP_VM)
1002Parameters: struct kvm_enable_cap (in)
1003Returns: 0 on success; -1 on error
1004
1005+Not all extensions are enabled by default. Using this ioctl the application
1006can enable an extension, making it available to the guest.
1007
1008On systems that do not support this ioctl, it always fails. On systems that
1009do support it, it only works for extensions that are supported for enablement.
1010
1011To check if a capability can be enabled, the KVM_CHECK_EXTENSION ioctl should
1012be used.
1013
1014struct kvm_enable_cap {
1015       /* in */
1016       __u32 cap;
1017
1018The capability that is supposed to get enabled.
1019
1020       __u32 flags;
1021
1022A bitfield indicating future enhancements. Has to be 0 for now.
1023
1024       __u64 args[4];
1025
1026Arguments for enabling a feature. If a feature needs initial values to
1027function properly, this is the place to put them.
1028
1029       __u8  pad[64];
1030};
1031
1032The vcpu ioctl should be used for vcpu-specific capabilities, the vm ioctl
1033for vm-wide capabilities.
1034
10354.38 KVM_GET_MP_STATE
1036
1037Capability: KVM_CAP_MP_STATE
1038Architectures: x86, s390, arm, arm64
1039Type: vcpu ioctl
1040Parameters: struct kvm_mp_state (out)
1041Returns: 0 on success; -1 on error
1042
1043struct kvm_mp_state {
1044	__u32 mp_state;
1045};
1046
1047Returns the vcpu's current "multiprocessing state" (though also valid on
1048uniprocessor guests).
1049
1050Possible values are:
1051
1052 - KVM_MP_STATE_RUNNABLE:        the vcpu is currently running [x86,arm/arm64]
1053 - KVM_MP_STATE_UNINITIALIZED:   the vcpu is an application processor (AP)
1054                                 which has not yet received an INIT signal [x86]
1055 - KVM_MP_STATE_INIT_RECEIVED:   the vcpu has received an INIT signal, and is
1056                                 now ready for a SIPI [x86]
1057 - KVM_MP_STATE_HALTED:          the vcpu has executed a HLT instruction and
1058                                 is waiting for an interrupt [x86]
1059 - KVM_MP_STATE_SIPI_RECEIVED:   the vcpu has just received a SIPI (vector
1060                                 accessible via KVM_GET_VCPU_EVENTS) [x86]
1061 - KVM_MP_STATE_STOPPED:         the vcpu is stopped [s390,arm/arm64]
1062 - KVM_MP_STATE_CHECK_STOP:      the vcpu is in a special error state [s390]
1063 - KVM_MP_STATE_OPERATING:       the vcpu is operating (running or halted)
1064                                 [s390]
1065 - KVM_MP_STATE_LOAD:            the vcpu is in a special load/startup state
1066                                 [s390]
1067
1068On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1069in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1070these architectures.
1071
1072For arm/arm64:
1073
1074The only states that are valid are KVM_MP_STATE_STOPPED and
1075KVM_MP_STATE_RUNNABLE which reflect if the vcpu is paused or not.
1076
10774.39 KVM_SET_MP_STATE
1078
1079Capability: KVM_CAP_MP_STATE
1080Architectures: x86, s390, arm, arm64
1081Type: vcpu ioctl
1082Parameters: struct kvm_mp_state (in)
1083Returns: 0 on success; -1 on error
1084
1085Sets the vcpu's current "multiprocessing state"; see KVM_GET_MP_STATE for
1086arguments.
1087
1088On x86, this ioctl is only useful after KVM_CREATE_IRQCHIP. Without an
1089in-kernel irqchip, the multiprocessing state must be maintained by userspace on
1090these architectures.
1091
1092For arm/arm64:
1093
1094The only states that are valid are KVM_MP_STATE_STOPPED and
1095KVM_MP_STATE_RUNNABLE which reflect if the vcpu should be paused or not.
1096
10974.40 KVM_SET_IDENTITY_MAP_ADDR
1098
1099Capability: KVM_CAP_SET_IDENTITY_MAP_ADDR
1100Architectures: x86
1101Type: vm ioctl
1102Parameters: unsigned long identity (in)
1103Returns: 0 on success, -1 on error
1104
1105This ioctl defines the physical address of a one-page region in the guest
1106physical address space.  The region must be within the first 4GB of the
1107guest physical address space and must not conflict with any memory slot
1108or any mmio address.  The guest may malfunction if it accesses this memory
1109region.
1110
1111This ioctl is required on Intel-based hosts.  This is needed on Intel hardware
1112because of a quirk in the virtualization implementation (see the internals
1113documentation when it pops into existence).
1114
1115
11164.41 KVM_SET_BOOT_CPU_ID
1117
1118Capability: KVM_CAP_SET_BOOT_CPU_ID
1119Architectures: x86
1120Type: vm ioctl
1121Parameters: unsigned long vcpu_id
1122Returns: 0 on success, -1 on error
1123
1124Define which vcpu is the Bootstrap Processor (BSP).  Values are the same
1125as the vcpu id in KVM_CREATE_VCPU.  If this ioctl is not called, the default
1126is vcpu 0.
1127
1128
11294.42 KVM_GET_XSAVE
1130
1131Capability: KVM_CAP_XSAVE
1132Architectures: x86
1133Type: vcpu ioctl
1134Parameters: struct kvm_xsave (out)
1135Returns: 0 on success, -1 on error
1136
1137struct kvm_xsave {
1138	__u32 region[1024];
1139};
1140
1141This ioctl would copy current vcpu's xsave struct to the userspace.
1142
1143
11444.43 KVM_SET_XSAVE
1145
1146Capability: KVM_CAP_XSAVE
1147Architectures: x86
1148Type: vcpu ioctl
1149Parameters: struct kvm_xsave (in)
1150Returns: 0 on success, -1 on error
1151
1152struct kvm_xsave {
1153	__u32 region[1024];
1154};
1155
1156This ioctl would copy userspace's xsave struct to the kernel.
1157
1158
11594.44 KVM_GET_XCRS
1160
1161Capability: KVM_CAP_XCRS
1162Architectures: x86
1163Type: vcpu ioctl
1164Parameters: struct kvm_xcrs (out)
1165Returns: 0 on success, -1 on error
1166
1167struct kvm_xcr {
1168	__u32 xcr;
1169	__u32 reserved;
1170	__u64 value;
1171};
1172
1173struct kvm_xcrs {
1174	__u32 nr_xcrs;
1175	__u32 flags;
1176	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1177	__u64 padding[16];
1178};
1179
1180This ioctl would copy current vcpu's xcrs to the userspace.
1181
1182
11834.45 KVM_SET_XCRS
1184
1185Capability: KVM_CAP_XCRS
1186Architectures: x86
1187Type: vcpu ioctl
1188Parameters: struct kvm_xcrs (in)
1189Returns: 0 on success, -1 on error
1190
1191struct kvm_xcr {
1192	__u32 xcr;
1193	__u32 reserved;
1194	__u64 value;
1195};
1196
1197struct kvm_xcrs {
1198	__u32 nr_xcrs;
1199	__u32 flags;
1200	struct kvm_xcr xcrs[KVM_MAX_XCRS];
1201	__u64 padding[16];
1202};
1203
1204This ioctl would set vcpu's xcr to the value userspace specified.
1205
1206
12074.46 KVM_GET_SUPPORTED_CPUID
1208
1209Capability: KVM_CAP_EXT_CPUID
1210Architectures: x86
1211Type: system ioctl
1212Parameters: struct kvm_cpuid2 (in/out)
1213Returns: 0 on success, -1 on error
1214
1215struct kvm_cpuid2 {
1216	__u32 nent;
1217	__u32 padding;
1218	struct kvm_cpuid_entry2 entries[0];
1219};
1220
1221#define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
1222#define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1)
1223#define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2)
1224
1225struct kvm_cpuid_entry2 {
1226	__u32 function;
1227	__u32 index;
1228	__u32 flags;
1229	__u32 eax;
1230	__u32 ebx;
1231	__u32 ecx;
1232	__u32 edx;
1233	__u32 padding[3];
1234};
1235
1236This ioctl returns x86 cpuid features which are supported by both the hardware
1237and kvm.  Userspace can use the information returned by this ioctl to
1238construct cpuid information (for KVM_SET_CPUID2) that is consistent with
1239hardware, kernel, and userspace capabilities, and with user requirements (for
1240example, the user may wish to constrain cpuid to emulate older hardware,
1241or for feature consistency across a cluster).
1242
1243Userspace invokes KVM_GET_SUPPORTED_CPUID by passing a kvm_cpuid2 structure
1244with the 'nent' field indicating the number of entries in the variable-size
1245array 'entries'.  If the number of entries is too low to describe the cpu
1246capabilities, an error (E2BIG) is returned.  If the number is too high,
1247the 'nent' field is adjusted and an error (ENOMEM) is returned.  If the
1248number is just right, the 'nent' field is adjusted to the number of valid
1249entries in the 'entries' array, which is then filled.
1250
1251The entries returned are the host cpuid as returned by the cpuid instruction,
1252with unknown or unsupported features masked out.  Some features (for example,
1253x2apic), may not be present in the host cpu, but are exposed by kvm if it can
1254emulate them efficiently. The fields in each entry are defined as follows:
1255
1256  function: the eax value used to obtain the entry
1257  index: the ecx value used to obtain the entry (for entries that are
1258         affected by ecx)
1259  flags: an OR of zero or more of the following:
1260        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
1261           if the index field is valid
1262        KVM_CPUID_FLAG_STATEFUL_FUNC:
1263           if cpuid for this function returns different values for successive
1264           invocations; there will be several entries with the same function,
1265           all with this flag set
1266        KVM_CPUID_FLAG_STATE_READ_NEXT:
1267           for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
1268           the first entry to be read by a cpu
1269   eax, ebx, ecx, edx: the values returned by the cpuid instruction for
1270         this function/index combination
1271
1272The TSC deadline timer feature (CPUID leaf 1, ecx[24]) is always returned
1273as false, since the feature depends on KVM_CREATE_IRQCHIP for local APIC
1274support.  Instead it is reported via
1275
1276  ioctl(KVM_CHECK_EXTENSION, KVM_CAP_TSC_DEADLINE_TIMER)
1277
1278if that returns true and you use KVM_CREATE_IRQCHIP, or if you emulate the
1279feature in userspace, then you can enable the feature for KVM_SET_CPUID2.
1280
1281
12824.47 KVM_PPC_GET_PVINFO
1283
1284Capability: KVM_CAP_PPC_GET_PVINFO
1285Architectures: ppc
1286Type: vm ioctl
1287Parameters: struct kvm_ppc_pvinfo (out)
1288Returns: 0 on success, !0 on error
1289
1290struct kvm_ppc_pvinfo {
1291	__u32 flags;
1292	__u32 hcall[4];
1293	__u8  pad[108];
1294};
1295
1296This ioctl fetches PV specific information that need to be passed to the guest
1297using the device tree or other means from vm context.
1298
1299The hcall array defines 4 instructions that make up a hypercall.
1300
1301If any additional field gets added to this structure later on, a bit for that
1302additional piece of information will be set in the flags bitmap.
1303
1304The flags bitmap is defined as:
1305
1306   /* the host supports the ePAPR idle hcall
1307   #define KVM_PPC_PVINFO_FLAGS_EV_IDLE   (1<<0)
1308
13094.48 KVM_ASSIGN_PCI_DEVICE (deprecated)
1310
1311Capability: none
1312Architectures: x86
1313Type: vm ioctl
1314Parameters: struct kvm_assigned_pci_dev (in)
1315Returns: 0 on success, -1 on error
1316
1317Assigns a host PCI device to the VM.
1318
1319struct kvm_assigned_pci_dev {
1320	__u32 assigned_dev_id;
1321	__u32 busnr;
1322	__u32 devfn;
1323	__u32 flags;
1324	__u32 segnr;
1325	union {
1326		__u32 reserved[11];
1327	};
1328};
1329
1330The PCI device is specified by the triple segnr, busnr, and devfn.
1331Identification in succeeding service requests is done via assigned_dev_id. The
1332following flags are specified:
1333
1334/* Depends on KVM_CAP_IOMMU */
1335#define KVM_DEV_ASSIGN_ENABLE_IOMMU	(1 << 0)
1336/* The following two depend on KVM_CAP_PCI_2_3 */
1337#define KVM_DEV_ASSIGN_PCI_2_3		(1 << 1)
1338#define KVM_DEV_ASSIGN_MASK_INTX	(1 << 2)
1339
1340If KVM_DEV_ASSIGN_PCI_2_3 is set, the kernel will manage legacy INTx interrupts
1341via the PCI-2.3-compliant device-level mask, thus enable IRQ sharing with other
1342assigned devices or host devices. KVM_DEV_ASSIGN_MASK_INTX specifies the
1343guest's view on the INTx mask, see KVM_ASSIGN_SET_INTX_MASK for details.
1344
1345The KVM_DEV_ASSIGN_ENABLE_IOMMU flag is a mandatory option to ensure
1346isolation of the device.  Usages not specifying this flag are deprecated.
1347
1348Only PCI header type 0 devices with PCI BAR resources are supported by
1349device assignment.  The user requesting this ioctl must have read/write
1350access to the PCI sysfs resource files associated with the device.
1351
1352Errors:
1353  ENOTTY: kernel does not support this ioctl
1354
1355  Other error conditions may be defined by individual device types or
1356  have their standard meanings.
1357
1358
13594.49 KVM_DEASSIGN_PCI_DEVICE (deprecated)
1360
1361Capability: none
1362Architectures: x86
1363Type: vm ioctl
1364Parameters: struct kvm_assigned_pci_dev (in)
1365Returns: 0 on success, -1 on error
1366
1367Ends PCI device assignment, releasing all associated resources.
1368
1369See KVM_ASSIGN_PCI_DEVICE for the data structure. Only assigned_dev_id is
1370used in kvm_assigned_pci_dev to identify the device.
1371
1372Errors:
1373  ENOTTY: kernel does not support this ioctl
1374
1375  Other error conditions may be defined by individual device types or
1376  have their standard meanings.
1377
13784.50 KVM_ASSIGN_DEV_IRQ (deprecated)
1379
1380Capability: KVM_CAP_ASSIGN_DEV_IRQ
1381Architectures: x86
1382Type: vm ioctl
1383Parameters: struct kvm_assigned_irq (in)
1384Returns: 0 on success, -1 on error
1385
1386Assigns an IRQ to a passed-through device.
1387
1388struct kvm_assigned_irq {
1389	__u32 assigned_dev_id;
1390	__u32 host_irq; /* ignored (legacy field) */
1391	__u32 guest_irq;
1392	__u32 flags;
1393	union {
1394		__u32 reserved[12];
1395	};
1396};
1397
1398The following flags are defined:
1399
1400#define KVM_DEV_IRQ_HOST_INTX    (1 << 0)
1401#define KVM_DEV_IRQ_HOST_MSI     (1 << 1)
1402#define KVM_DEV_IRQ_HOST_MSIX    (1 << 2)
1403
1404#define KVM_DEV_IRQ_GUEST_INTX   (1 << 8)
1405#define KVM_DEV_IRQ_GUEST_MSI    (1 << 9)
1406#define KVM_DEV_IRQ_GUEST_MSIX   (1 << 10)
1407
1408It is not valid to specify multiple types per host or guest IRQ. However, the
1409IRQ type of host and guest can differ or can even be null.
1410
1411Errors:
1412  ENOTTY: kernel does not support this ioctl
1413
1414  Other error conditions may be defined by individual device types or
1415  have their standard meanings.
1416
1417
14184.51 KVM_DEASSIGN_DEV_IRQ (deprecated)
1419
1420Capability: KVM_CAP_ASSIGN_DEV_IRQ
1421Architectures: x86
1422Type: vm ioctl
1423Parameters: struct kvm_assigned_irq (in)
1424Returns: 0 on success, -1 on error
1425
1426Ends an IRQ assignment to a passed-through device.
1427
1428See KVM_ASSIGN_DEV_IRQ for the data structure. The target device is specified
1429by assigned_dev_id, flags must correspond to the IRQ type specified on
1430KVM_ASSIGN_DEV_IRQ. Partial deassignment of host or guest IRQ is allowed.
1431
1432
14334.52 KVM_SET_GSI_ROUTING
1434
1435Capability: KVM_CAP_IRQ_ROUTING
1436Architectures: x86 s390
1437Type: vm ioctl
1438Parameters: struct kvm_irq_routing (in)
1439Returns: 0 on success, -1 on error
1440
1441Sets the GSI routing table entries, overwriting any previously set entries.
1442
1443struct kvm_irq_routing {
1444	__u32 nr;
1445	__u32 flags;
1446	struct kvm_irq_routing_entry entries[0];
1447};
1448
1449No flags are specified so far, the corresponding field must be set to zero.
1450
1451struct kvm_irq_routing_entry {
1452	__u32 gsi;
1453	__u32 type;
1454	__u32 flags;
1455	__u32 pad;
1456	union {
1457		struct kvm_irq_routing_irqchip irqchip;
1458		struct kvm_irq_routing_msi msi;
1459		struct kvm_irq_routing_s390_adapter adapter;
1460		__u32 pad[8];
1461	} u;
1462};
1463
1464/* gsi routing entry types */
1465#define KVM_IRQ_ROUTING_IRQCHIP 1
1466#define KVM_IRQ_ROUTING_MSI 2
1467#define KVM_IRQ_ROUTING_S390_ADAPTER 3
1468
1469No flags are specified so far, the corresponding field must be set to zero.
1470
1471struct kvm_irq_routing_irqchip {
1472	__u32 irqchip;
1473	__u32 pin;
1474};
1475
1476struct kvm_irq_routing_msi {
1477	__u32 address_lo;
1478	__u32 address_hi;
1479	__u32 data;
1480	__u32 pad;
1481};
1482
1483struct kvm_irq_routing_s390_adapter {
1484	__u64 ind_addr;
1485	__u64 summary_addr;
1486	__u64 ind_offset;
1487	__u32 summary_offset;
1488	__u32 adapter_id;
1489};
1490
1491
14924.53 KVM_ASSIGN_SET_MSIX_NR (deprecated)
1493
1494Capability: none
1495Architectures: x86
1496Type: vm ioctl
1497Parameters: struct kvm_assigned_msix_nr (in)
1498Returns: 0 on success, -1 on error
1499
1500Set the number of MSI-X interrupts for an assigned device. The number is
1501reset again by terminating the MSI-X assignment of the device via
1502KVM_DEASSIGN_DEV_IRQ. Calling this service more than once at any earlier
1503point will fail.
1504
1505struct kvm_assigned_msix_nr {
1506	__u32 assigned_dev_id;
1507	__u16 entry_nr;
1508	__u16 padding;
1509};
1510
1511#define KVM_MAX_MSIX_PER_DEV		256
1512
1513
15144.54 KVM_ASSIGN_SET_MSIX_ENTRY (deprecated)
1515
1516Capability: none
1517Architectures: x86
1518Type: vm ioctl
1519Parameters: struct kvm_assigned_msix_entry (in)
1520Returns: 0 on success, -1 on error
1521
1522Specifies the routing of an MSI-X assigned device interrupt to a GSI. Setting
1523the GSI vector to zero means disabling the interrupt.
1524
1525struct kvm_assigned_msix_entry {
1526	__u32 assigned_dev_id;
1527	__u32 gsi;
1528	__u16 entry; /* The index of entry in the MSI-X table */
1529	__u16 padding[3];
1530};
1531
1532Errors:
1533  ENOTTY: kernel does not support this ioctl
1534
1535  Other error conditions may be defined by individual device types or
1536  have their standard meanings.
1537
1538
15394.55 KVM_SET_TSC_KHZ
1540
1541Capability: KVM_CAP_TSC_CONTROL
1542Architectures: x86
1543Type: vcpu ioctl
1544Parameters: virtual tsc_khz
1545Returns: 0 on success, -1 on error
1546
1547Specifies the tsc frequency for the virtual machine. The unit of the
1548frequency is KHz.
1549
1550
15514.56 KVM_GET_TSC_KHZ
1552
1553Capability: KVM_CAP_GET_TSC_KHZ
1554Architectures: x86
1555Type: vcpu ioctl
1556Parameters: none
1557Returns: virtual tsc-khz on success, negative value on error
1558
1559Returns the tsc frequency of the guest. The unit of the return value is
1560KHz. If the host has unstable tsc this ioctl returns -EIO instead as an
1561error.
1562
1563
15644.57 KVM_GET_LAPIC
1565
1566Capability: KVM_CAP_IRQCHIP
1567Architectures: x86
1568Type: vcpu ioctl
1569Parameters: struct kvm_lapic_state (out)
1570Returns: 0 on success, -1 on error
1571
1572#define KVM_APIC_REG_SIZE 0x400
1573struct kvm_lapic_state {
1574	char regs[KVM_APIC_REG_SIZE];
1575};
1576
1577Reads the Local APIC registers and copies them into the input argument.  The
1578data format and layout are the same as documented in the architecture manual.
1579
1580
15814.58 KVM_SET_LAPIC
1582
1583Capability: KVM_CAP_IRQCHIP
1584Architectures: x86
1585Type: vcpu ioctl
1586Parameters: struct kvm_lapic_state (in)
1587Returns: 0 on success, -1 on error
1588
1589#define KVM_APIC_REG_SIZE 0x400
1590struct kvm_lapic_state {
1591	char regs[KVM_APIC_REG_SIZE];
1592};
1593
1594Copies the input argument into the Local APIC registers.  The data format
1595and layout are the same as documented in the architecture manual.
1596
1597
15984.59 KVM_IOEVENTFD
1599
1600Capability: KVM_CAP_IOEVENTFD
1601Architectures: all
1602Type: vm ioctl
1603Parameters: struct kvm_ioeventfd (in)
1604Returns: 0 on success, !0 on error
1605
1606This ioctl attaches or detaches an ioeventfd to a legal pio/mmio address
1607within the guest.  A guest write in the registered address will signal the
1608provided event instead of triggering an exit.
1609
1610struct kvm_ioeventfd {
1611	__u64 datamatch;
1612	__u64 addr;        /* legal pio/mmio address */
1613	__u32 len;         /* 0, 1, 2, 4, or 8 bytes    */
1614	__s32 fd;
1615	__u32 flags;
1616	__u8  pad[36];
1617};
1618
1619For the special case of virtio-ccw devices on s390, the ioevent is matched
1620to a subchannel/virtqueue tuple instead.
1621
1622The following flags are defined:
1623
1624#define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch)
1625#define KVM_IOEVENTFD_FLAG_PIO       (1 << kvm_ioeventfd_flag_nr_pio)
1626#define KVM_IOEVENTFD_FLAG_DEASSIGN  (1 << kvm_ioeventfd_flag_nr_deassign)
1627#define KVM_IOEVENTFD_FLAG_VIRTIO_CCW_NOTIFY \
1628	(1 << kvm_ioeventfd_flag_nr_virtio_ccw_notify)
1629
1630If datamatch flag is set, the event will be signaled only if the written value
1631to the registered address is equal to datamatch in struct kvm_ioeventfd.
1632
1633For virtio-ccw devices, addr contains the subchannel id and datamatch the
1634virtqueue index.
1635
1636With KVM_CAP_IOEVENTFD_ANY_LENGTH, a zero length ioeventfd is allowed, and
1637the kernel will ignore the length of guest write and may get a faster vmexit.
1638The speedup may only apply to specific architectures, but the ioeventfd will
1639work anyway.
1640
16414.60 KVM_DIRTY_TLB
1642
1643Capability: KVM_CAP_SW_TLB
1644Architectures: ppc
1645Type: vcpu ioctl
1646Parameters: struct kvm_dirty_tlb (in)
1647Returns: 0 on success, -1 on error
1648
1649struct kvm_dirty_tlb {
1650	__u64 bitmap;
1651	__u32 num_dirty;
1652};
1653
1654This must be called whenever userspace has changed an entry in the shared
1655TLB, prior to calling KVM_RUN on the associated vcpu.
1656
1657The "bitmap" field is the userspace address of an array.  This array
1658consists of a number of bits, equal to the total number of TLB entries as
1659determined by the last successful call to KVM_CONFIG_TLB, rounded up to the
1660nearest multiple of 64.
1661
1662Each bit corresponds to one TLB entry, ordered the same as in the shared TLB
1663array.
1664
1665The array is little-endian: the bit 0 is the least significant bit of the
1666first byte, bit 8 is the least significant bit of the second byte, etc.
1667This avoids any complications with differing word sizes.
1668
1669The "num_dirty" field is a performance hint for KVM to determine whether it
1670should skip processing the bitmap and just invalidate everything.  It must
1671be set to the number of set bits in the bitmap.
1672
1673
16744.61 KVM_ASSIGN_SET_INTX_MASK (deprecated)
1675
1676Capability: KVM_CAP_PCI_2_3
1677Architectures: x86
1678Type: vm ioctl
1679Parameters: struct kvm_assigned_pci_dev (in)
1680Returns: 0 on success, -1 on error
1681
1682Allows userspace to mask PCI INTx interrupts from the assigned device.  The
1683kernel will not deliver INTx interrupts to the guest between setting and
1684clearing of KVM_ASSIGN_SET_INTX_MASK via this interface.  This enables use of
1685and emulation of PCI 2.3 INTx disable command register behavior.
1686
1687This may be used for both PCI 2.3 devices supporting INTx disable natively and
1688older devices lacking this support. Userspace is responsible for emulating the
1689read value of the INTx disable bit in the guest visible PCI command register.
1690When modifying the INTx disable state, userspace should precede updating the
1691physical device command register by calling this ioctl to inform the kernel of
1692the new intended INTx mask state.
1693
1694Note that the kernel uses the device INTx disable bit to internally manage the
1695device interrupt state for PCI 2.3 devices.  Reads of this register may
1696therefore not match the expected value.  Writes should always use the guest
1697intended INTx disable value rather than attempting to read-copy-update the
1698current physical device state.  Races between user and kernel updates to the
1699INTx disable bit are handled lazily in the kernel.  It's possible the device
1700may generate unintended interrupts, but they will not be injected into the
1701guest.
1702
1703See KVM_ASSIGN_DEV_IRQ for the data structure.  The target device is specified
1704by assigned_dev_id.  In the flags field, only KVM_DEV_ASSIGN_MASK_INTX is
1705evaluated.
1706
1707
17084.62 KVM_CREATE_SPAPR_TCE
1709
1710Capability: KVM_CAP_SPAPR_TCE
1711Architectures: powerpc
1712Type: vm ioctl
1713Parameters: struct kvm_create_spapr_tce (in)
1714Returns: file descriptor for manipulating the created TCE table
1715
1716This creates a virtual TCE (translation control entry) table, which
1717is an IOMMU for PAPR-style virtual I/O.  It is used to translate
1718logical addresses used in virtual I/O into guest physical addresses,
1719and provides a scatter/gather capability for PAPR virtual I/O.
1720
1721/* for KVM_CAP_SPAPR_TCE */
1722struct kvm_create_spapr_tce {
1723	__u64 liobn;
1724	__u32 window_size;
1725};
1726
1727The liobn field gives the logical IO bus number for which to create a
1728TCE table.  The window_size field specifies the size of the DMA window
1729which this TCE table will translate - the table will contain one 64
1730bit TCE entry for every 4kiB of the DMA window.
1731
1732When the guest issues an H_PUT_TCE hcall on a liobn for which a TCE
1733table has been created using this ioctl(), the kernel will handle it
1734in real mode, updating the TCE table.  H_PUT_TCE calls for other
1735liobns will cause a vm exit and must be handled by userspace.
1736
1737The return value is a file descriptor which can be passed to mmap(2)
1738to map the created TCE table into userspace.  This lets userspace read
1739the entries written by kernel-handled H_PUT_TCE calls, and also lets
1740userspace update the TCE table directly which is useful in some
1741circumstances.
1742
1743
17444.63 KVM_ALLOCATE_RMA
1745
1746Capability: KVM_CAP_PPC_RMA
1747Architectures: powerpc
1748Type: vm ioctl
1749Parameters: struct kvm_allocate_rma (out)
1750Returns: file descriptor for mapping the allocated RMA
1751
1752This allocates a Real Mode Area (RMA) from the pool allocated at boot
1753time by the kernel.  An RMA is a physically-contiguous, aligned region
1754of memory used on older POWER processors to provide the memory which
1755will be accessed by real-mode (MMU off) accesses in a KVM guest.
1756POWER processors support a set of sizes for the RMA that usually
1757includes 64MB, 128MB, 256MB and some larger powers of two.
1758
1759/* for KVM_ALLOCATE_RMA */
1760struct kvm_allocate_rma {
1761	__u64 rma_size;
1762};
1763
1764The return value is a file descriptor which can be passed to mmap(2)
1765to map the allocated RMA into userspace.  The mapped area can then be
1766passed to the KVM_SET_USER_MEMORY_REGION ioctl to establish it as the
1767RMA for a virtual machine.  The size of the RMA in bytes (which is
1768fixed at host kernel boot time) is returned in the rma_size field of
1769the argument structure.
1770
1771The KVM_CAP_PPC_RMA capability is 1 or 2 if the KVM_ALLOCATE_RMA ioctl
1772is supported; 2 if the processor requires all virtual machines to have
1773an RMA, or 1 if the processor can use an RMA but doesn't require it,
1774because it supports the Virtual RMA (VRMA) facility.
1775
1776
17774.64 KVM_NMI
1778
1779Capability: KVM_CAP_USER_NMI
1780Architectures: x86
1781Type: vcpu ioctl
1782Parameters: none
1783Returns: 0 on success, -1 on error
1784
1785Queues an NMI on the thread's vcpu.  Note this is well defined only
1786when KVM_CREATE_IRQCHIP has not been called, since this is an interface
1787between the virtual cpu core and virtual local APIC.  After KVM_CREATE_IRQCHIP
1788has been called, this interface is completely emulated within the kernel.
1789
1790To use this to emulate the LINT1 input with KVM_CREATE_IRQCHIP, use the
1791following algorithm:
1792
1793  - pause the vcpu
1794  - read the local APIC's state (KVM_GET_LAPIC)
1795  - check whether changing LINT1 will queue an NMI (see the LVT entry for LINT1)
1796  - if so, issue KVM_NMI
1797  - resume the vcpu
1798
1799Some guests configure the LINT1 NMI input to cause a panic, aiding in
1800debugging.
1801
1802
18034.65 KVM_S390_UCAS_MAP
1804
1805Capability: KVM_CAP_S390_UCONTROL
1806Architectures: s390
1807Type: vcpu ioctl
1808Parameters: struct kvm_s390_ucas_mapping (in)
1809Returns: 0 in case of success
1810
1811The parameter is defined like this:
1812	struct kvm_s390_ucas_mapping {
1813		__u64 user_addr;
1814		__u64 vcpu_addr;
1815		__u64 length;
1816	};
1817
1818This ioctl maps the memory at "user_addr" with the length "length" to
1819the vcpu's address space starting at "vcpu_addr". All parameters need to
1820be aligned by 1 megabyte.
1821
1822
18234.66 KVM_S390_UCAS_UNMAP
1824
1825Capability: KVM_CAP_S390_UCONTROL
1826Architectures: s390
1827Type: vcpu ioctl
1828Parameters: struct kvm_s390_ucas_mapping (in)
1829Returns: 0 in case of success
1830
1831The parameter is defined like this:
1832	struct kvm_s390_ucas_mapping {
1833		__u64 user_addr;
1834		__u64 vcpu_addr;
1835		__u64 length;
1836	};
1837
1838This ioctl unmaps the memory in the vcpu's address space starting at
1839"vcpu_addr" with the length "length". The field "user_addr" is ignored.
1840All parameters need to be aligned by 1 megabyte.
1841
1842
18434.67 KVM_S390_VCPU_FAULT
1844
1845Capability: KVM_CAP_S390_UCONTROL
1846Architectures: s390
1847Type: vcpu ioctl
1848Parameters: vcpu absolute address (in)
1849Returns: 0 in case of success
1850
1851This call creates a page table entry on the virtual cpu's address space
1852(for user controlled virtual machines) or the virtual machine's address
1853space (for regular virtual machines). This only works for minor faults,
1854thus it's recommended to access subject memory page via the user page
1855table upfront. This is useful to handle validity intercepts for user
1856controlled virtual machines to fault in the virtual cpu's lowcore pages
1857prior to calling the KVM_RUN ioctl.
1858
1859
18604.68 KVM_SET_ONE_REG
1861
1862Capability: KVM_CAP_ONE_REG
1863Architectures: all
1864Type: vcpu ioctl
1865Parameters: struct kvm_one_reg (in)
1866Returns: 0 on success, negative value on failure
1867
1868struct kvm_one_reg {
1869       __u64 id;
1870       __u64 addr;
1871};
1872
1873Using this ioctl, a single vcpu register can be set to a specific value
1874defined by user space with the passed in struct kvm_one_reg, where id
1875refers to the register identifier as described below and addr is a pointer
1876to a variable with the respective size. There can be architecture agnostic
1877and architecture specific registers. Each have their own range of operation
1878and their own constants and width. To keep track of the implemented
1879registers, find a list below:
1880
1881  Arch  |           Register            | Width (bits)
1882        |                               |
1883  PPC   | KVM_REG_PPC_HIOR              | 64
1884  PPC   | KVM_REG_PPC_IAC1              | 64
1885  PPC   | KVM_REG_PPC_IAC2              | 64
1886  PPC   | KVM_REG_PPC_IAC3              | 64
1887  PPC   | KVM_REG_PPC_IAC4              | 64
1888  PPC   | KVM_REG_PPC_DAC1              | 64
1889  PPC   | KVM_REG_PPC_DAC2              | 64
1890  PPC   | KVM_REG_PPC_DABR              | 64
1891  PPC   | KVM_REG_PPC_DSCR              | 64
1892  PPC   | KVM_REG_PPC_PURR              | 64
1893  PPC   | KVM_REG_PPC_SPURR             | 64
1894  PPC   | KVM_REG_PPC_DAR               | 64
1895  PPC   | KVM_REG_PPC_DSISR             | 32
1896  PPC   | KVM_REG_PPC_AMR               | 64
1897  PPC   | KVM_REG_PPC_UAMOR             | 64
1898  PPC   | KVM_REG_PPC_MMCR0             | 64
1899  PPC   | KVM_REG_PPC_MMCR1             | 64
1900  PPC   | KVM_REG_PPC_MMCRA             | 64
1901  PPC   | KVM_REG_PPC_MMCR2             | 64
1902  PPC   | KVM_REG_PPC_MMCRS             | 64
1903  PPC   | KVM_REG_PPC_SIAR              | 64
1904  PPC   | KVM_REG_PPC_SDAR              | 64
1905  PPC   | KVM_REG_PPC_SIER              | 64
1906  PPC   | KVM_REG_PPC_PMC1              | 32
1907  PPC   | KVM_REG_PPC_PMC2              | 32
1908  PPC   | KVM_REG_PPC_PMC3              | 32
1909  PPC   | KVM_REG_PPC_PMC4              | 32
1910  PPC   | KVM_REG_PPC_PMC5              | 32
1911  PPC   | KVM_REG_PPC_PMC6              | 32
1912  PPC   | KVM_REG_PPC_PMC7              | 32
1913  PPC   | KVM_REG_PPC_PMC8              | 32
1914  PPC   | KVM_REG_PPC_FPR0              | 64
1915          ...
1916  PPC   | KVM_REG_PPC_FPR31             | 64
1917  PPC   | KVM_REG_PPC_VR0               | 128
1918          ...
1919  PPC   | KVM_REG_PPC_VR31              | 128
1920  PPC   | KVM_REG_PPC_VSR0              | 128
1921          ...
1922  PPC   | KVM_REG_PPC_VSR31             | 128
1923  PPC   | KVM_REG_PPC_FPSCR             | 64
1924  PPC   | KVM_REG_PPC_VSCR              | 32
1925  PPC   | KVM_REG_PPC_VPA_ADDR          | 64
1926  PPC   | KVM_REG_PPC_VPA_SLB           | 128
1927  PPC   | KVM_REG_PPC_VPA_DTL           | 128
1928  PPC   | KVM_REG_PPC_EPCR              | 32
1929  PPC   | KVM_REG_PPC_EPR               | 32
1930  PPC   | KVM_REG_PPC_TCR               | 32
1931  PPC   | KVM_REG_PPC_TSR               | 32
1932  PPC   | KVM_REG_PPC_OR_TSR            | 32
1933  PPC   | KVM_REG_PPC_CLEAR_TSR         | 32
1934  PPC   | KVM_REG_PPC_MAS0              | 32
1935  PPC   | KVM_REG_PPC_MAS1              | 32
1936  PPC   | KVM_REG_PPC_MAS2              | 64
1937  PPC   | KVM_REG_PPC_MAS7_3            | 64
1938  PPC   | KVM_REG_PPC_MAS4              | 32
1939  PPC   | KVM_REG_PPC_MAS6              | 32
1940  PPC   | KVM_REG_PPC_MMUCFG            | 32
1941  PPC   | KVM_REG_PPC_TLB0CFG           | 32
1942  PPC   | KVM_REG_PPC_TLB1CFG           | 32
1943  PPC   | KVM_REG_PPC_TLB2CFG           | 32
1944  PPC   | KVM_REG_PPC_TLB3CFG           | 32
1945  PPC   | KVM_REG_PPC_TLB0PS            | 32
1946  PPC   | KVM_REG_PPC_TLB1PS            | 32
1947  PPC   | KVM_REG_PPC_TLB2PS            | 32
1948  PPC   | KVM_REG_PPC_TLB3PS            | 32
1949  PPC   | KVM_REG_PPC_EPTCFG            | 32
1950  PPC   | KVM_REG_PPC_ICP_STATE         | 64
1951  PPC   | KVM_REG_PPC_TB_OFFSET         | 64
1952  PPC   | KVM_REG_PPC_SPMC1             | 32
1953  PPC   | KVM_REG_PPC_SPMC2             | 32
1954  PPC   | KVM_REG_PPC_IAMR              | 64
1955  PPC   | KVM_REG_PPC_TFHAR             | 64
1956  PPC   | KVM_REG_PPC_TFIAR             | 64
1957  PPC   | KVM_REG_PPC_TEXASR            | 64
1958  PPC   | KVM_REG_PPC_FSCR              | 64
1959  PPC   | KVM_REG_PPC_PSPB              | 32
1960  PPC   | KVM_REG_PPC_EBBHR             | 64
1961  PPC   | KVM_REG_PPC_EBBRR             | 64
1962  PPC   | KVM_REG_PPC_BESCR             | 64
1963  PPC   | KVM_REG_PPC_TAR               | 64
1964  PPC   | KVM_REG_PPC_DPDES             | 64
1965  PPC   | KVM_REG_PPC_DAWR              | 64
1966  PPC   | KVM_REG_PPC_DAWRX             | 64
1967  PPC   | KVM_REG_PPC_CIABR             | 64
1968  PPC   | KVM_REG_PPC_IC                | 64
1969  PPC   | KVM_REG_PPC_VTB               | 64
1970  PPC   | KVM_REG_PPC_CSIGR             | 64
1971  PPC   | KVM_REG_PPC_TACR              | 64
1972  PPC   | KVM_REG_PPC_TCSCR             | 64
1973  PPC   | KVM_REG_PPC_PID               | 64
1974  PPC   | KVM_REG_PPC_ACOP              | 64
1975  PPC   | KVM_REG_PPC_VRSAVE            | 32
1976  PPC   | KVM_REG_PPC_LPCR              | 32
1977  PPC   | KVM_REG_PPC_LPCR_64           | 64
1978  PPC   | KVM_REG_PPC_PPR               | 64
1979  PPC   | KVM_REG_PPC_ARCH_COMPAT       | 32
1980  PPC   | KVM_REG_PPC_DABRX             | 32
1981  PPC   | KVM_REG_PPC_WORT              | 64
1982  PPC	| KVM_REG_PPC_SPRG9             | 64
1983  PPC	| KVM_REG_PPC_DBSR              | 32
1984  PPC   | KVM_REG_PPC_TM_GPR0           | 64
1985          ...
1986  PPC   | KVM_REG_PPC_TM_GPR31          | 64
1987  PPC   | KVM_REG_PPC_TM_VSR0           | 128
1988          ...
1989  PPC   | KVM_REG_PPC_TM_VSR63          | 128
1990  PPC   | KVM_REG_PPC_TM_CR             | 64
1991  PPC   | KVM_REG_PPC_TM_LR             | 64
1992  PPC   | KVM_REG_PPC_TM_CTR            | 64
1993  PPC   | KVM_REG_PPC_TM_FPSCR          | 64
1994  PPC   | KVM_REG_PPC_TM_AMR            | 64
1995  PPC   | KVM_REG_PPC_TM_PPR            | 64
1996  PPC   | KVM_REG_PPC_TM_VRSAVE         | 64
1997  PPC   | KVM_REG_PPC_TM_VSCR           | 32
1998  PPC   | KVM_REG_PPC_TM_DSCR           | 64
1999  PPC   | KVM_REG_PPC_TM_TAR            | 64
2000  PPC   | KVM_REG_PPC_TM_XER            | 64
2001        |                               |
2002  MIPS  | KVM_REG_MIPS_R0               | 64
2003          ...
2004  MIPS  | KVM_REG_MIPS_R31              | 64
2005  MIPS  | KVM_REG_MIPS_HI               | 64
2006  MIPS  | KVM_REG_MIPS_LO               | 64
2007  MIPS  | KVM_REG_MIPS_PC               | 64
2008  MIPS  | KVM_REG_MIPS_CP0_INDEX        | 32
2009  MIPS  | KVM_REG_MIPS_CP0_CONTEXT      | 64
2010  MIPS  | KVM_REG_MIPS_CP0_USERLOCAL    | 64
2011  MIPS  | KVM_REG_MIPS_CP0_PAGEMASK     | 32
2012  MIPS  | KVM_REG_MIPS_CP0_WIRED        | 32
2013  MIPS  | KVM_REG_MIPS_CP0_HWRENA       | 32
2014  MIPS  | KVM_REG_MIPS_CP0_BADVADDR     | 64
2015  MIPS  | KVM_REG_MIPS_CP0_COUNT        | 32
2016  MIPS  | KVM_REG_MIPS_CP0_ENTRYHI      | 64
2017  MIPS  | KVM_REG_MIPS_CP0_COMPARE      | 32
2018  MIPS  | KVM_REG_MIPS_CP0_STATUS       | 32
2019  MIPS  | KVM_REG_MIPS_CP0_CAUSE        | 32
2020  MIPS  | KVM_REG_MIPS_CP0_EPC          | 64
2021  MIPS  | KVM_REG_MIPS_CP0_PRID         | 32
2022  MIPS  | KVM_REG_MIPS_CP0_CONFIG       | 32
2023  MIPS  | KVM_REG_MIPS_CP0_CONFIG1      | 32
2024  MIPS  | KVM_REG_MIPS_CP0_CONFIG2      | 32
2025  MIPS  | KVM_REG_MIPS_CP0_CONFIG3      | 32
2026  MIPS  | KVM_REG_MIPS_CP0_CONFIG4      | 32
2027  MIPS  | KVM_REG_MIPS_CP0_CONFIG5      | 32
2028  MIPS  | KVM_REG_MIPS_CP0_CONFIG7      | 32
2029  MIPS  | KVM_REG_MIPS_CP0_ERROREPC     | 64
2030  MIPS  | KVM_REG_MIPS_COUNT_CTL        | 64
2031  MIPS  | KVM_REG_MIPS_COUNT_RESUME     | 64
2032  MIPS  | KVM_REG_MIPS_COUNT_HZ         | 64
2033  MIPS  | KVM_REG_MIPS_FPR_32(0..31)    | 32
2034  MIPS  | KVM_REG_MIPS_FPR_64(0..31)    | 64
2035  MIPS  | KVM_REG_MIPS_VEC_128(0..31)   | 128
2036  MIPS  | KVM_REG_MIPS_FCR_IR           | 32
2037  MIPS  | KVM_REG_MIPS_FCR_CSR          | 32
2038  MIPS  | KVM_REG_MIPS_MSA_IR           | 32
2039  MIPS  | KVM_REG_MIPS_MSA_CSR          | 32
2040
2041ARM registers are mapped using the lower 32 bits.  The upper 16 of that
2042is the register group type, or coprocessor number:
2043
2044ARM core registers have the following id bit patterns:
2045  0x4020 0000 0010 <index into the kvm_regs struct:16>
2046
2047ARM 32-bit CP15 registers have the following id bit patterns:
2048  0x4020 0000 000F <zero:1> <crn:4> <crm:4> <opc1:4> <opc2:3>
2049
2050ARM 64-bit CP15 registers have the following id bit patterns:
2051  0x4030 0000 000F <zero:1> <zero:4> <crm:4> <opc1:4> <zero:3>
2052
2053ARM CCSIDR registers are demultiplexed by CSSELR value:
2054  0x4020 0000 0011 00 <csselr:8>
2055
2056ARM 32-bit VFP control registers have the following id bit patterns:
2057  0x4020 0000 0012 1 <regno:12>
2058
2059ARM 64-bit FP registers have the following id bit patterns:
2060  0x4030 0000 0012 0 <regno:12>
2061
2062
2063arm64 registers are mapped using the lower 32 bits. The upper 16 of
2064that is the register group type, or coprocessor number:
2065
2066arm64 core/FP-SIMD registers have the following id bit patterns. Note
2067that the size of the access is variable, as the kvm_regs structure
2068contains elements ranging from 32 to 128 bits. The index is a 32bit
2069value in the kvm_regs structure seen as a 32bit array.
2070  0x60x0 0000 0010 <index into the kvm_regs struct:16>
2071
2072arm64 CCSIDR registers are demultiplexed by CSSELR value:
2073  0x6020 0000 0011 00 <csselr:8>
2074
2075arm64 system registers have the following id bit patterns:
2076  0x6030 0000 0013 <op0:2> <op1:3> <crn:4> <crm:4> <op2:3>
2077
2078
2079MIPS registers are mapped using the lower 32 bits.  The upper 16 of that is
2080the register group type:
2081
2082MIPS core registers (see above) have the following id bit patterns:
2083  0x7030 0000 0000 <reg:16>
2084
2085MIPS CP0 registers (see KVM_REG_MIPS_CP0_* above) have the following id bit
2086patterns depending on whether they're 32-bit or 64-bit registers:
2087  0x7020 0000 0001 00 <reg:5> <sel:3>   (32-bit)
2088  0x7030 0000 0001 00 <reg:5> <sel:3>   (64-bit)
2089
2090MIPS KVM control registers (see above) have the following id bit patterns:
2091  0x7030 0000 0002 <reg:16>
2092
2093MIPS FPU registers (see KVM_REG_MIPS_FPR_{32,64}() above) have the following
2094id bit patterns depending on the size of the register being accessed. They are
2095always accessed according to the current guest FPU mode (Status.FR and
2096Config5.FRE), i.e. as the guest would see them, and they become unpredictable
2097if the guest FPU mode is changed. MIPS SIMD Architecture (MSA) vector
2098registers (see KVM_REG_MIPS_VEC_128() above) have similar patterns as they
2099overlap the FPU registers:
2100  0x7020 0000 0003 00 <0:3> <reg:5> (32-bit FPU registers)
2101  0x7030 0000 0003 00 <0:3> <reg:5> (64-bit FPU registers)
2102  0x7040 0000 0003 00 <0:3> <reg:5> (128-bit MSA vector registers)
2103
2104MIPS FPU control registers (see KVM_REG_MIPS_FCR_{IR,CSR} above) have the
2105following id bit patterns:
2106  0x7020 0000 0003 01 <0:3> <reg:5>
2107
2108MIPS MSA control registers (see KVM_REG_MIPS_MSA_{IR,CSR} above) have the
2109following id bit patterns:
2110  0x7020 0000 0003 02 <0:3> <reg:5>
2111
2112
21134.69 KVM_GET_ONE_REG
2114
2115Capability: KVM_CAP_ONE_REG
2116Architectures: all
2117Type: vcpu ioctl
2118Parameters: struct kvm_one_reg (in and out)
2119Returns: 0 on success, negative value on failure
2120
2121This ioctl allows to receive the value of a single register implemented
2122in a vcpu. The register to read is indicated by the "id" field of the
2123kvm_one_reg struct passed in. On success, the register value can be found
2124at the memory location pointed to by "addr".
2125
2126The list of registers accessible using this interface is identical to the
2127list in 4.68.
2128
2129
21304.70 KVM_KVMCLOCK_CTRL
2131
2132Capability: KVM_CAP_KVMCLOCK_CTRL
2133Architectures: Any that implement pvclocks (currently x86 only)
2134Type: vcpu ioctl
2135Parameters: None
2136Returns: 0 on success, -1 on error
2137
2138This signals to the host kernel that the specified guest is being paused by
2139userspace.  The host will set a flag in the pvclock structure that is checked
2140from the soft lockup watchdog.  The flag is part of the pvclock structure that
2141is shared between guest and host, specifically the second bit of the flags
2142field of the pvclock_vcpu_time_info structure.  It will be set exclusively by
2143the host and read/cleared exclusively by the guest.  The guest operation of
2144checking and clearing the flag must an atomic operation so
2145load-link/store-conditional, or equivalent must be used.  There are two cases
2146where the guest will clear the flag: when the soft lockup watchdog timer resets
2147itself or when a soft lockup is detected.  This ioctl can be called any time
2148after pausing the vcpu, but before it is resumed.
2149
2150
21514.71 KVM_SIGNAL_MSI
2152
2153Capability: KVM_CAP_SIGNAL_MSI
2154Architectures: x86
2155Type: vm ioctl
2156Parameters: struct kvm_msi (in)
2157Returns: >0 on delivery, 0 if guest blocked the MSI, and -1 on error
2158
2159Directly inject a MSI message. Only valid with in-kernel irqchip that handles
2160MSI messages.
2161
2162struct kvm_msi {
2163	__u32 address_lo;
2164	__u32 address_hi;
2165	__u32 data;
2166	__u32 flags;
2167	__u8  pad[16];
2168};
2169
2170No flags are defined so far. The corresponding field must be 0.
2171
2172
21734.71 KVM_CREATE_PIT2
2174
2175Capability: KVM_CAP_PIT2
2176Architectures: x86
2177Type: vm ioctl
2178Parameters: struct kvm_pit_config (in)
2179Returns: 0 on success, -1 on error
2180
2181Creates an in-kernel device model for the i8254 PIT. This call is only valid
2182after enabling in-kernel irqchip support via KVM_CREATE_IRQCHIP. The following
2183parameters have to be passed:
2184
2185struct kvm_pit_config {
2186	__u32 flags;
2187	__u32 pad[15];
2188};
2189
2190Valid flags are:
2191
2192#define KVM_PIT_SPEAKER_DUMMY     1 /* emulate speaker port stub */
2193
2194PIT timer interrupts may use a per-VM kernel thread for injection. If it
2195exists, this thread will have a name of the following pattern:
2196
2197kvm-pit/<owner-process-pid>
2198
2199When running a guest with elevated priorities, the scheduling parameters of
2200this thread may have to be adjusted accordingly.
2201
2202This IOCTL replaces the obsolete KVM_CREATE_PIT.
2203
2204
22054.72 KVM_GET_PIT2
2206
2207Capability: KVM_CAP_PIT_STATE2
2208Architectures: x86
2209Type: vm ioctl
2210Parameters: struct kvm_pit_state2 (out)
2211Returns: 0 on success, -1 on error
2212
2213Retrieves the state of the in-kernel PIT model. Only valid after
2214KVM_CREATE_PIT2. The state is returned in the following structure:
2215
2216struct kvm_pit_state2 {
2217	struct kvm_pit_channel_state channels[3];
2218	__u32 flags;
2219	__u32 reserved[9];
2220};
2221
2222Valid flags are:
2223
2224/* disable PIT in HPET legacy mode */
2225#define KVM_PIT_FLAGS_HPET_LEGACY  0x00000001
2226
2227This IOCTL replaces the obsolete KVM_GET_PIT.
2228
2229
22304.73 KVM_SET_PIT2
2231
2232Capability: KVM_CAP_PIT_STATE2
2233Architectures: x86
2234Type: vm ioctl
2235Parameters: struct kvm_pit_state2 (in)
2236Returns: 0 on success, -1 on error
2237
2238Sets the state of the in-kernel PIT model. Only valid after KVM_CREATE_PIT2.
2239See KVM_GET_PIT2 for details on struct kvm_pit_state2.
2240
2241This IOCTL replaces the obsolete KVM_SET_PIT.
2242
2243
22444.74 KVM_PPC_GET_SMMU_INFO
2245
2246Capability: KVM_CAP_PPC_GET_SMMU_INFO
2247Architectures: powerpc
2248Type: vm ioctl
2249Parameters: None
2250Returns: 0 on success, -1 on error
2251
2252This populates and returns a structure describing the features of
2253the "Server" class MMU emulation supported by KVM.
2254This can in turn be used by userspace to generate the appropriate
2255device-tree properties for the guest operating system.
2256
2257The structure contains some global information, followed by an
2258array of supported segment page sizes:
2259
2260      struct kvm_ppc_smmu_info {
2261	     __u64 flags;
2262	     __u32 slb_size;
2263	     __u32 pad;
2264	     struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
2265      };
2266
2267The supported flags are:
2268
2269    - KVM_PPC_PAGE_SIZES_REAL:
2270        When that flag is set, guest page sizes must "fit" the backing
2271        store page sizes. When not set, any page size in the list can
2272        be used regardless of how they are backed by userspace.
2273
2274    - KVM_PPC_1T_SEGMENTS
2275        The emulated MMU supports 1T segments in addition to the
2276        standard 256M ones.
2277
2278The "slb_size" field indicates how many SLB entries are supported
2279
2280The "sps" array contains 8 entries indicating the supported base
2281page sizes for a segment in increasing order. Each entry is defined
2282as follow:
2283
2284   struct kvm_ppc_one_seg_page_size {
2285	__u32 page_shift;	/* Base page shift of segment (or 0) */
2286	__u32 slb_enc;		/* SLB encoding for BookS */
2287	struct kvm_ppc_one_page_size enc[KVM_PPC_PAGE_SIZES_MAX_SZ];
2288   };
2289
2290An entry with a "page_shift" of 0 is unused. Because the array is
2291organized in increasing order, a lookup can stop when encoutering
2292such an entry.
2293
2294The "slb_enc" field provides the encoding to use in the SLB for the
2295page size. The bits are in positions such as the value can directly
2296be OR'ed into the "vsid" argument of the slbmte instruction.
2297
2298The "enc" array is a list which for each of those segment base page
2299size provides the list of supported actual page sizes (which can be
2300only larger or equal to the base page size), along with the
2301corresponding encoding in the hash PTE. Similarly, the array is
23028 entries sorted by increasing sizes and an entry with a "0" shift
2303is an empty entry and a terminator:
2304
2305   struct kvm_ppc_one_page_size {
2306	__u32 page_shift;	/* Page shift (or 0) */
2307	__u32 pte_enc;		/* Encoding in the HPTE (>>12) */
2308   };
2309
2310The "pte_enc" field provides a value that can OR'ed into the hash
2311PTE's RPN field (ie, it needs to be shifted left by 12 to OR it
2312into the hash PTE second double word).
2313
23144.75 KVM_IRQFD
2315
2316Capability: KVM_CAP_IRQFD
2317Architectures: x86 s390 arm arm64
2318Type: vm ioctl
2319Parameters: struct kvm_irqfd (in)
2320Returns: 0 on success, -1 on error
2321
2322Allows setting an eventfd to directly trigger a guest interrupt.
2323kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
2324kvm_irqfd.gsi specifies the irqchip pin toggled by this event.  When
2325an event is triggered on the eventfd, an interrupt is injected into
2326the guest using the specified gsi pin.  The irqfd is removed using
2327the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
2328and kvm_irqfd.gsi.
2329
2330With KVM_CAP_IRQFD_RESAMPLE, KVM_IRQFD supports a de-assert and notify
2331mechanism allowing emulation of level-triggered, irqfd-based
2332interrupts.  When KVM_IRQFD_FLAG_RESAMPLE is set the user must pass an
2333additional eventfd in the kvm_irqfd.resamplefd field.  When operating
2334in resample mode, posting of an interrupt through kvm_irq.fd asserts
2335the specified gsi in the irqchip.  When the irqchip is resampled, such
2336as from an EOI, the gsi is de-asserted and the user is notified via
2337kvm_irqfd.resamplefd.  It is the user's responsibility to re-queue
2338the interrupt if the device making use of it still requires service.
2339Note that closing the resamplefd is not sufficient to disable the
2340irqfd.  The KVM_IRQFD_FLAG_RESAMPLE is only necessary on assignment
2341and need not be specified with KVM_IRQFD_FLAG_DEASSIGN.
2342
2343On ARM/ARM64, the gsi field in the kvm_irqfd struct specifies the Shared
2344Peripheral Interrupt (SPI) index, such that the GIC interrupt ID is
2345given by gsi + 32.
2346
23474.76 KVM_PPC_ALLOCATE_HTAB
2348
2349Capability: KVM_CAP_PPC_ALLOC_HTAB
2350Architectures: powerpc
2351Type: vm ioctl
2352Parameters: Pointer to u32 containing hash table order (in/out)
2353Returns: 0 on success, -1 on error
2354
2355This requests the host kernel to allocate an MMU hash table for a
2356guest using the PAPR paravirtualization interface.  This only does
2357anything if the kernel is configured to use the Book 3S HV style of
2358virtualization.  Otherwise the capability doesn't exist and the ioctl
2359returns an ENOTTY error.  The rest of this description assumes Book 3S
2360HV.
2361
2362There must be no vcpus running when this ioctl is called; if there
2363are, it will do nothing and return an EBUSY error.
2364
2365The parameter is a pointer to a 32-bit unsigned integer variable
2366containing the order (log base 2) of the desired size of the hash
2367table, which must be between 18 and 46.  On successful return from the
2368ioctl, it will have been updated with the order of the hash table that
2369was allocated.
2370
2371If no hash table has been allocated when any vcpu is asked to run
2372(with the KVM_RUN ioctl), the host kernel will allocate a
2373default-sized hash table (16 MB).
2374
2375If this ioctl is called when a hash table has already been allocated,
2376the kernel will clear out the existing hash table (zero all HPTEs) and
2377return the hash table order in the parameter.  (If the guest is using
2378the virtualized real-mode area (VRMA) facility, the kernel will
2379re-create the VMRA HPTEs on the next KVM_RUN of any vcpu.)
2380
23814.77 KVM_S390_INTERRUPT
2382
2383Capability: basic
2384Architectures: s390
2385Type: vm ioctl, vcpu ioctl
2386Parameters: struct kvm_s390_interrupt (in)
2387Returns: 0 on success, -1 on error
2388
2389Allows to inject an interrupt to the guest. Interrupts can be floating
2390(vm ioctl) or per cpu (vcpu ioctl), depending on the interrupt type.
2391
2392Interrupt parameters are passed via kvm_s390_interrupt:
2393
2394struct kvm_s390_interrupt {
2395	__u32 type;
2396	__u32 parm;
2397	__u64 parm64;
2398};
2399
2400type can be one of the following:
2401
2402KVM_S390_SIGP_STOP (vcpu) - sigp stop; optional flags in parm
2403KVM_S390_PROGRAM_INT (vcpu) - program check; code in parm
2404KVM_S390_SIGP_SET_PREFIX (vcpu) - sigp set prefix; prefix address in parm
2405KVM_S390_RESTART (vcpu) - restart
2406KVM_S390_INT_CLOCK_COMP (vcpu) - clock comparator interrupt
2407KVM_S390_INT_CPU_TIMER (vcpu) - CPU timer interrupt
2408KVM_S390_INT_VIRTIO (vm) - virtio external interrupt; external interrupt
2409			   parameters in parm and parm64
2410KVM_S390_INT_SERVICE (vm) - sclp external interrupt; sclp parameter in parm
2411KVM_S390_INT_EMERGENCY (vcpu) - sigp emergency; source cpu in parm
2412KVM_S390_INT_EXTERNAL_CALL (vcpu) - sigp external call; source cpu in parm
2413KVM_S390_INT_IO(ai,cssid,ssid,schid) (vm) - compound value to indicate an
2414    I/O interrupt (ai - adapter interrupt; cssid,ssid,schid - subchannel);
2415    I/O interruption parameters in parm (subchannel) and parm64 (intparm,
2416    interruption subclass)
2417KVM_S390_MCHK (vm, vcpu) - machine check interrupt; cr 14 bits in parm,
2418                           machine check interrupt code in parm64 (note that
2419                           machine checks needing further payload are not
2420                           supported by this ioctl)
2421
2422Note that the vcpu ioctl is asynchronous to vcpu execution.
2423
24244.78 KVM_PPC_GET_HTAB_FD
2425
2426Capability: KVM_CAP_PPC_HTAB_FD
2427Architectures: powerpc
2428Type: vm ioctl
2429Parameters: Pointer to struct kvm_get_htab_fd (in)
2430Returns: file descriptor number (>= 0) on success, -1 on error
2431
2432This returns a file descriptor that can be used either to read out the
2433entries in the guest's hashed page table (HPT), or to write entries to
2434initialize the HPT.  The returned fd can only be written to if the
2435KVM_GET_HTAB_WRITE bit is set in the flags field of the argument, and
2436can only be read if that bit is clear.  The argument struct looks like
2437this:
2438
2439/* For KVM_PPC_GET_HTAB_FD */
2440struct kvm_get_htab_fd {
2441	__u64	flags;
2442	__u64	start_index;
2443	__u64	reserved[2];
2444};
2445
2446/* Values for kvm_get_htab_fd.flags */
2447#define KVM_GET_HTAB_BOLTED_ONLY	((__u64)0x1)
2448#define KVM_GET_HTAB_WRITE		((__u64)0x2)
2449
2450The `start_index' field gives the index in the HPT of the entry at
2451which to start reading.  It is ignored when writing.
2452
2453Reads on the fd will initially supply information about all
2454"interesting" HPT entries.  Interesting entries are those with the
2455bolted bit set, if the KVM_GET_HTAB_BOLTED_ONLY bit is set, otherwise
2456all entries.  When the end of the HPT is reached, the read() will
2457return.  If read() is called again on the fd, it will start again from
2458the beginning of the HPT, but will only return HPT entries that have
2459changed since they were last read.
2460
2461Data read or written is structured as a header (8 bytes) followed by a
2462series of valid HPT entries (16 bytes) each.  The header indicates how
2463many valid HPT entries there are and how many invalid entries follow
2464the valid entries.  The invalid entries are not represented explicitly
2465in the stream.  The header format is:
2466
2467struct kvm_get_htab_header {
2468	__u32	index;
2469	__u16	n_valid;
2470	__u16	n_invalid;
2471};
2472
2473Writes to the fd create HPT entries starting at the index given in the
2474header; first `n_valid' valid entries with contents from the data
2475written, then `n_invalid' invalid entries, invalidating any previously
2476valid entries found.
2477
24784.79 KVM_CREATE_DEVICE
2479
2480Capability: KVM_CAP_DEVICE_CTRL
2481Type: vm ioctl
2482Parameters: struct kvm_create_device (in/out)
2483Returns: 0 on success, -1 on error
2484Errors:
2485  ENODEV: The device type is unknown or unsupported
2486  EEXIST: Device already created, and this type of device may not
2487          be instantiated multiple times
2488
2489  Other error conditions may be defined by individual device types or
2490  have their standard meanings.
2491
2492Creates an emulated device in the kernel.  The file descriptor returned
2493in fd can be used with KVM_SET/GET/HAS_DEVICE_ATTR.
2494
2495If the KVM_CREATE_DEVICE_TEST flag is set, only test whether the
2496device type is supported (not necessarily whether it can be created
2497in the current vm).
2498
2499Individual devices should not define flags.  Attributes should be used
2500for specifying any behavior that is not implied by the device type
2501number.
2502
2503struct kvm_create_device {
2504	__u32	type;	/* in: KVM_DEV_TYPE_xxx */
2505	__u32	fd;	/* out: device handle */
2506	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
2507};
2508
25094.80 KVM_SET_DEVICE_ATTR/KVM_GET_DEVICE_ATTR
2510
2511Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device
2512Type: device ioctl, vm ioctl
2513Parameters: struct kvm_device_attr
2514Returns: 0 on success, -1 on error
2515Errors:
2516  ENXIO:  The group or attribute is unknown/unsupported for this device
2517  EPERM:  The attribute cannot (currently) be accessed this way
2518          (e.g. read-only attribute, or attribute that only makes
2519          sense when the device is in a different state)
2520
2521  Other error conditions may be defined by individual device types.
2522
2523Gets/sets a specified piece of device configuration and/or state.  The
2524semantics are device-specific.  See individual device documentation in
2525the "devices" directory.  As with ONE_REG, the size of the data
2526transferred is defined by the particular attribute.
2527
2528struct kvm_device_attr {
2529	__u32	flags;		/* no flags currently defined */
2530	__u32	group;		/* device-defined */
2531	__u64	attr;		/* group-defined */
2532	__u64	addr;		/* userspace address of attr data */
2533};
2534
25354.81 KVM_HAS_DEVICE_ATTR
2536
2537Capability: KVM_CAP_DEVICE_CTRL, KVM_CAP_VM_ATTRIBUTES for vm device
2538Type: device ioctl, vm ioctl
2539Parameters: struct kvm_device_attr
2540Returns: 0 on success, -1 on error
2541Errors:
2542  ENXIO:  The group or attribute is unknown/unsupported for this device
2543
2544Tests whether a device supports a particular attribute.  A successful
2545return indicates the attribute is implemented.  It does not necessarily
2546indicate that the attribute can be read or written in the device's
2547current state.  "addr" is ignored.
2548
25494.82 KVM_ARM_VCPU_INIT
2550
2551Capability: basic
2552Architectures: arm, arm64
2553Type: vcpu ioctl
2554Parameters: struct kvm_vcpu_init (in)
2555Returns: 0 on success; -1 on error
2556Errors:
2557  EINVAL:    the target is unknown, or the combination of features is invalid.
2558  ENOENT:    a features bit specified is unknown.
2559
2560This tells KVM what type of CPU to present to the guest, and what
2561optional features it should have.  This will cause a reset of the cpu
2562registers to their initial values.  If this is not called, KVM_RUN will
2563return ENOEXEC for that vcpu.
2564
2565Note that because some registers reflect machine topology, all vcpus
2566should be created before this ioctl is invoked.
2567
2568Userspace can call this function multiple times for a given vcpu, including
2569after the vcpu has been run. This will reset the vcpu to its initial
2570state. All calls to this function after the initial call must use the same
2571target and same set of feature flags, otherwise EINVAL will be returned.
2572
2573Possible features:
2574	- KVM_ARM_VCPU_POWER_OFF: Starts the CPU in a power-off state.
2575	  Depends on KVM_CAP_ARM_PSCI.  If not set, the CPU will be powered on
2576	  and execute guest code when KVM_RUN is called.
2577	- KVM_ARM_VCPU_EL1_32BIT: Starts the CPU in a 32bit mode.
2578	  Depends on KVM_CAP_ARM_EL1_32BIT (arm64 only).
2579	- KVM_ARM_VCPU_PSCI_0_2: Emulate PSCI v0.2 for the CPU.
2580	  Depends on KVM_CAP_ARM_PSCI_0_2.
2581
2582
25834.83 KVM_ARM_PREFERRED_TARGET
2584
2585Capability: basic
2586Architectures: arm, arm64
2587Type: vm ioctl
2588Parameters: struct struct kvm_vcpu_init (out)
2589Returns: 0 on success; -1 on error
2590Errors:
2591  ENODEV:    no preferred target available for the host
2592
2593This queries KVM for preferred CPU target type which can be emulated
2594by KVM on underlying host.
2595
2596The ioctl returns struct kvm_vcpu_init instance containing information
2597about preferred CPU target type and recommended features for it.  The
2598kvm_vcpu_init->features bitmap returned will have feature bits set if
2599the preferred target recommends setting these features, but this is
2600not mandatory.
2601
2602The information returned by this ioctl can be used to prepare an instance
2603of struct kvm_vcpu_init for KVM_ARM_VCPU_INIT ioctl which will result in
2604in VCPU matching underlying host.
2605
2606
26074.84 KVM_GET_REG_LIST
2608
2609Capability: basic
2610Architectures: arm, arm64, mips
2611Type: vcpu ioctl
2612Parameters: struct kvm_reg_list (in/out)
2613Returns: 0 on success; -1 on error
2614Errors:
2615  E2BIG:     the reg index list is too big to fit in the array specified by
2616             the user (the number required will be written into n).
2617
2618struct kvm_reg_list {
2619	__u64 n; /* number of registers in reg[] */
2620	__u64 reg[0];
2621};
2622
2623This ioctl returns the guest registers that are supported for the
2624KVM_GET_ONE_REG/KVM_SET_ONE_REG calls.
2625
2626
26274.85 KVM_ARM_SET_DEVICE_ADDR (deprecated)
2628
2629Capability: KVM_CAP_ARM_SET_DEVICE_ADDR
2630Architectures: arm, arm64
2631Type: vm ioctl
2632Parameters: struct kvm_arm_device_address (in)
2633Returns: 0 on success, -1 on error
2634Errors:
2635  ENODEV: The device id is unknown
2636  ENXIO:  Device not supported on current system
2637  EEXIST: Address already set
2638  E2BIG:  Address outside guest physical address space
2639  EBUSY:  Address overlaps with other device range
2640
2641struct kvm_arm_device_addr {
2642	__u64 id;
2643	__u64 addr;
2644};
2645
2646Specify a device address in the guest's physical address space where guests
2647can access emulated or directly exposed devices, which the host kernel needs
2648to know about. The id field is an architecture specific identifier for a
2649specific device.
2650
2651ARM/arm64 divides the id field into two parts, a device id and an
2652address type id specific to the individual device.
2653
2654  bits:  | 63        ...       32 | 31    ...    16 | 15    ...    0 |
2655  field: |        0x00000000      |     device id   |  addr type id  |
2656
2657ARM/arm64 currently only require this when using the in-kernel GIC
2658support for the hardware VGIC features, using KVM_ARM_DEVICE_VGIC_V2
2659as the device id.  When setting the base address for the guest's
2660mapping of the VGIC virtual CPU and distributor interface, the ioctl
2661must be called after calling KVM_CREATE_IRQCHIP, but before calling
2662KVM_RUN on any of the VCPUs.  Calling this ioctl twice for any of the
2663base addresses will return -EEXIST.
2664
2665Note, this IOCTL is deprecated and the more flexible SET/GET_DEVICE_ATTR API
2666should be used instead.
2667
2668
26694.86 KVM_PPC_RTAS_DEFINE_TOKEN
2670
2671Capability: KVM_CAP_PPC_RTAS
2672Architectures: ppc
2673Type: vm ioctl
2674Parameters: struct kvm_rtas_token_args
2675Returns: 0 on success, -1 on error
2676
2677Defines a token value for a RTAS (Run Time Abstraction Services)
2678service in order to allow it to be handled in the kernel.  The
2679argument struct gives the name of the service, which must be the name
2680of a service that has a kernel-side implementation.  If the token
2681value is non-zero, it will be associated with that service, and
2682subsequent RTAS calls by the guest specifying that token will be
2683handled by the kernel.  If the token value is 0, then any token
2684associated with the service will be forgotten, and subsequent RTAS
2685calls by the guest for that service will be passed to userspace to be
2686handled.
2687
26884.87 KVM_SET_GUEST_DEBUG
2689
2690Capability: KVM_CAP_SET_GUEST_DEBUG
2691Architectures: x86, s390, ppc, arm64
2692Type: vcpu ioctl
2693Parameters: struct kvm_guest_debug (in)
2694Returns: 0 on success; -1 on error
2695
2696struct kvm_guest_debug {
2697       __u32 control;
2698       __u32 pad;
2699       struct kvm_guest_debug_arch arch;
2700};
2701
2702Set up the processor specific debug registers and configure vcpu for
2703handling guest debug events. There are two parts to the structure, the
2704first a control bitfield indicates the type of debug events to handle
2705when running. Common control bits are:
2706
2707  - KVM_GUESTDBG_ENABLE:        guest debugging is enabled
2708  - KVM_GUESTDBG_SINGLESTEP:    the next run should single-step
2709
2710The top 16 bits of the control field are architecture specific control
2711flags which can include the following:
2712
2713  - KVM_GUESTDBG_USE_SW_BP:     using software breakpoints [x86, arm64]
2714  - KVM_GUESTDBG_USE_HW_BP:     using hardware breakpoints [x86, s390, arm64]
2715  - KVM_GUESTDBG_INJECT_DB:     inject DB type exception [x86]
2716  - KVM_GUESTDBG_INJECT_BP:     inject BP type exception [x86]
2717  - KVM_GUESTDBG_EXIT_PENDING:  trigger an immediate guest exit [s390]
2718
2719For example KVM_GUESTDBG_USE_SW_BP indicates that software breakpoints
2720are enabled in memory so we need to ensure breakpoint exceptions are
2721correctly trapped and the KVM run loop exits at the breakpoint and not
2722running off into the normal guest vector. For KVM_GUESTDBG_USE_HW_BP
2723we need to ensure the guest vCPUs architecture specific registers are
2724updated to the correct (supplied) values.
2725
2726The second part of the structure is architecture specific and
2727typically contains a set of debug registers.
2728
2729For arm64 the number of debug registers is implementation defined and
2730can be determined by querying the KVM_CAP_GUEST_DEBUG_HW_BPS and
2731KVM_CAP_GUEST_DEBUG_HW_WPS capabilities which return a positive number
2732indicating the number of supported registers.
2733
2734When debug events exit the main run loop with the reason
2735KVM_EXIT_DEBUG with the kvm_debug_exit_arch part of the kvm_run
2736structure containing architecture specific debug information.
2737
27384.88 KVM_GET_EMULATED_CPUID
2739
2740Capability: KVM_CAP_EXT_EMUL_CPUID
2741Architectures: x86
2742Type: system ioctl
2743Parameters: struct kvm_cpuid2 (in/out)
2744Returns: 0 on success, -1 on error
2745
2746struct kvm_cpuid2 {
2747	__u32 nent;
2748	__u32 flags;
2749	struct kvm_cpuid_entry2 entries[0];
2750};
2751
2752The member 'flags' is used for passing flags from userspace.
2753
2754#define KVM_CPUID_FLAG_SIGNIFCANT_INDEX		BIT(0)
2755#define KVM_CPUID_FLAG_STATEFUL_FUNC		BIT(1)
2756#define KVM_CPUID_FLAG_STATE_READ_NEXT		BIT(2)
2757
2758struct kvm_cpuid_entry2 {
2759	__u32 function;
2760	__u32 index;
2761	__u32 flags;
2762	__u32 eax;
2763	__u32 ebx;
2764	__u32 ecx;
2765	__u32 edx;
2766	__u32 padding[3];
2767};
2768
2769This ioctl returns x86 cpuid features which are emulated by
2770kvm.Userspace can use the information returned by this ioctl to query
2771which features are emulated by kvm instead of being present natively.
2772
2773Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
2774structure with the 'nent' field indicating the number of entries in
2775the variable-size array 'entries'. If the number of entries is too low
2776to describe the cpu capabilities, an error (E2BIG) is returned. If the
2777number is too high, the 'nent' field is adjusted and an error (ENOMEM)
2778is returned. If the number is just right, the 'nent' field is adjusted
2779to the number of valid entries in the 'entries' array, which is then
2780filled.
2781
2782The entries returned are the set CPUID bits of the respective features
2783which kvm emulates, as returned by the CPUID instruction, with unknown
2784or unsupported feature bits cleared.
2785
2786Features like x2apic, for example, may not be present in the host cpu
2787but are exposed by kvm in KVM_GET_SUPPORTED_CPUID because they can be
2788emulated efficiently and thus not included here.
2789
2790The fields in each entry are defined as follows:
2791
2792  function: the eax value used to obtain the entry
2793  index: the ecx value used to obtain the entry (for entries that are
2794         affected by ecx)
2795  flags: an OR of zero or more of the following:
2796        KVM_CPUID_FLAG_SIGNIFCANT_INDEX:
2797           if the index field is valid
2798        KVM_CPUID_FLAG_STATEFUL_FUNC:
2799           if cpuid for this function returns different values for successive
2800           invocations; there will be several entries with the same function,
2801           all with this flag set
2802        KVM_CPUID_FLAG_STATE_READ_NEXT:
2803           for KVM_CPUID_FLAG_STATEFUL_FUNC entries, set if this entry is
2804           the first entry to be read by a cpu
2805   eax, ebx, ecx, edx: the values returned by the cpuid instruction for
2806         this function/index combination
2807
28084.89 KVM_S390_MEM_OP
2809
2810Capability: KVM_CAP_S390_MEM_OP
2811Architectures: s390
2812Type: vcpu ioctl
2813Parameters: struct kvm_s390_mem_op (in)
2814Returns: = 0 on success,
2815         < 0 on generic error (e.g. -EFAULT or -ENOMEM),
2816         > 0 if an exception occurred while walking the page tables
2817
2818Read or write data from/to the logical (virtual) memory of a VCPU.
2819
2820Parameters are specified via the following structure:
2821
2822struct kvm_s390_mem_op {
2823	__u64 gaddr;		/* the guest address */
2824	__u64 flags;		/* flags */
2825	__u32 size;		/* amount of bytes */
2826	__u32 op;		/* type of operation */
2827	__u64 buf;		/* buffer in userspace */
2828	__u8 ar;		/* the access register number */
2829	__u8 reserved[31];	/* should be set to 0 */
2830};
2831
2832The type of operation is specified in the "op" field. It is either
2833KVM_S390_MEMOP_LOGICAL_READ for reading from logical memory space or
2834KVM_S390_MEMOP_LOGICAL_WRITE for writing to logical memory space. The
2835KVM_S390_MEMOP_F_CHECK_ONLY flag can be set in the "flags" field to check
2836whether the corresponding memory access would create an access exception
2837(without touching the data in the memory at the destination). In case an
2838access exception occurred while walking the MMU tables of the guest, the
2839ioctl returns a positive error number to indicate the type of exception.
2840This exception is also raised directly at the corresponding VCPU if the
2841flag KVM_S390_MEMOP_F_INJECT_EXCEPTION is set in the "flags" field.
2842
2843The start address of the memory region has to be specified in the "gaddr"
2844field, and the length of the region in the "size" field. "buf" is the buffer
2845supplied by the userspace application where the read data should be written
2846to for KVM_S390_MEMOP_LOGICAL_READ, or where the data that should be written
2847is stored for a KVM_S390_MEMOP_LOGICAL_WRITE. "buf" is unused and can be NULL
2848when KVM_S390_MEMOP_F_CHECK_ONLY is specified. "ar" designates the access
2849register number to be used.
2850
2851The "reserved" field is meant for future extensions. It is not used by
2852KVM with the currently defined set of flags.
2853
28544.90 KVM_S390_GET_SKEYS
2855
2856Capability: KVM_CAP_S390_SKEYS
2857Architectures: s390
2858Type: vm ioctl
2859Parameters: struct kvm_s390_skeys
2860Returns: 0 on success, KVM_S390_GET_KEYS_NONE if guest is not using storage
2861         keys, negative value on error
2862
2863This ioctl is used to get guest storage key values on the s390
2864architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2865
2866struct kvm_s390_skeys {
2867	__u64 start_gfn;
2868	__u64 count;
2869	__u64 skeydata_addr;
2870	__u32 flags;
2871	__u32 reserved[9];
2872};
2873
2874The start_gfn field is the number of the first guest frame whose storage keys
2875you want to get.
2876
2877The count field is the number of consecutive frames (starting from start_gfn)
2878whose storage keys to get. The count field must be at least 1 and the maximum
2879allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2880will cause the ioctl to return -EINVAL.
2881
2882The skeydata_addr field is the address to a buffer large enough to hold count
2883bytes. This buffer will be filled with storage key data by the ioctl.
2884
28854.91 KVM_S390_SET_SKEYS
2886
2887Capability: KVM_CAP_S390_SKEYS
2888Architectures: s390
2889Type: vm ioctl
2890Parameters: struct kvm_s390_skeys
2891Returns: 0 on success, negative value on error
2892
2893This ioctl is used to set guest storage key values on the s390
2894architecture. The ioctl takes parameters via the kvm_s390_skeys struct.
2895See section on KVM_S390_GET_SKEYS for struct definition.
2896
2897The start_gfn field is the number of the first guest frame whose storage keys
2898you want to set.
2899
2900The count field is the number of consecutive frames (starting from start_gfn)
2901whose storage keys to get. The count field must be at least 1 and the maximum
2902allowed value is defined as KVM_S390_SKEYS_ALLOC_MAX. Values outside this range
2903will cause the ioctl to return -EINVAL.
2904
2905The skeydata_addr field is the address to a buffer containing count bytes of
2906storage keys. Each byte in the buffer will be set as the storage key for a
2907single frame starting at start_gfn for count frames.
2908
2909Note: If any architecturally invalid key value is found in the given data then
2910the ioctl will return -EINVAL.
2911
29124.92 KVM_S390_IRQ
2913
2914Capability: KVM_CAP_S390_INJECT_IRQ
2915Architectures: s390
2916Type: vcpu ioctl
2917Parameters: struct kvm_s390_irq (in)
2918Returns: 0 on success, -1 on error
2919Errors:
2920  EINVAL: interrupt type is invalid
2921          type is KVM_S390_SIGP_STOP and flag parameter is invalid value
2922          type is KVM_S390_INT_EXTERNAL_CALL and code is bigger
2923            than the maximum of VCPUs
2924  EBUSY:  type is KVM_S390_SIGP_SET_PREFIX and vcpu is not stopped
2925          type is KVM_S390_SIGP_STOP and a stop irq is already pending
2926          type is KVM_S390_INT_EXTERNAL_CALL and an external call interrupt
2927            is already pending
2928
2929Allows to inject an interrupt to the guest.
2930
2931Using struct kvm_s390_irq as a parameter allows
2932to inject additional payload which is not
2933possible via KVM_S390_INTERRUPT.
2934
2935Interrupt parameters are passed via kvm_s390_irq:
2936
2937struct kvm_s390_irq {
2938	__u64 type;
2939	union {
2940		struct kvm_s390_io_info io;
2941		struct kvm_s390_ext_info ext;
2942		struct kvm_s390_pgm_info pgm;
2943		struct kvm_s390_emerg_info emerg;
2944		struct kvm_s390_extcall_info extcall;
2945		struct kvm_s390_prefix_info prefix;
2946		struct kvm_s390_stop_info stop;
2947		struct kvm_s390_mchk_info mchk;
2948		char reserved[64];
2949	} u;
2950};
2951
2952type can be one of the following:
2953
2954KVM_S390_SIGP_STOP - sigp stop; parameter in .stop
2955KVM_S390_PROGRAM_INT - program check; parameters in .pgm
2956KVM_S390_SIGP_SET_PREFIX - sigp set prefix; parameters in .prefix
2957KVM_S390_RESTART - restart; no parameters
2958KVM_S390_INT_CLOCK_COMP - clock comparator interrupt; no parameters
2959KVM_S390_INT_CPU_TIMER - CPU timer interrupt; no parameters
2960KVM_S390_INT_EMERGENCY - sigp emergency; parameters in .emerg
2961KVM_S390_INT_EXTERNAL_CALL - sigp external call; parameters in .extcall
2962KVM_S390_MCHK - machine check interrupt; parameters in .mchk
2963
2964
2965Note that the vcpu ioctl is asynchronous to vcpu execution.
2966
29674.94 KVM_S390_GET_IRQ_STATE
2968
2969Capability: KVM_CAP_S390_IRQ_STATE
2970Architectures: s390
2971Type: vcpu ioctl
2972Parameters: struct kvm_s390_irq_state (out)
2973Returns: >= number of bytes copied into buffer,
2974         -EINVAL if buffer size is 0,
2975         -ENOBUFS if buffer size is too small to fit all pending interrupts,
2976         -EFAULT if the buffer address was invalid
2977
2978This ioctl allows userspace to retrieve the complete state of all currently
2979pending interrupts in a single buffer. Use cases include migration
2980and introspection. The parameter structure contains the address of a
2981userspace buffer and its length:
2982
2983struct kvm_s390_irq_state {
2984	__u64 buf;
2985	__u32 flags;
2986	__u32 len;
2987	__u32 reserved[4];
2988};
2989
2990Userspace passes in the above struct and for each pending interrupt a
2991struct kvm_s390_irq is copied to the provided buffer.
2992
2993If -ENOBUFS is returned the buffer provided was too small and userspace
2994may retry with a bigger buffer.
2995
29964.95 KVM_S390_SET_IRQ_STATE
2997
2998Capability: KVM_CAP_S390_IRQ_STATE
2999Architectures: s390
3000Type: vcpu ioctl
3001Parameters: struct kvm_s390_irq_state (in)
3002Returns: 0 on success,
3003         -EFAULT if the buffer address was invalid,
3004         -EINVAL for an invalid buffer length (see below),
3005         -EBUSY if there were already interrupts pending,
3006         errors occurring when actually injecting the
3007          interrupt. See KVM_S390_IRQ.
3008
3009This ioctl allows userspace to set the complete state of all cpu-local
3010interrupts currently pending for the vcpu. It is intended for restoring
3011interrupt state after a migration. The input parameter is a userspace buffer
3012containing a struct kvm_s390_irq_state:
3013
3014struct kvm_s390_irq_state {
3015	__u64 buf;
3016	__u32 len;
3017	__u32 pad;
3018};
3019
3020The userspace memory referenced by buf contains a struct kvm_s390_irq
3021for each interrupt to be injected into the guest.
3022If one of the interrupts could not be injected for some reason the
3023ioctl aborts.
3024
3025len must be a multiple of sizeof(struct kvm_s390_irq). It must be > 0
3026and it must not exceed (max_vcpus + 32) * sizeof(struct kvm_s390_irq),
3027which is the maximum number of possibly pending cpu-local interrupts.
3028
30294.90 KVM_SMI
3030
3031Capability: KVM_CAP_X86_SMM
3032Architectures: x86
3033Type: vcpu ioctl
3034Parameters: none
3035Returns: 0 on success, -1 on error
3036
3037Queues an SMI on the thread's vcpu.
3038
30395. The kvm_run structure
3040------------------------
3041
3042Application code obtains a pointer to the kvm_run structure by
3043mmap()ing a vcpu fd.  From that point, application code can control
3044execution by changing fields in kvm_run prior to calling the KVM_RUN
3045ioctl, and obtain information about the reason KVM_RUN returned by
3046looking up structure members.
3047
3048struct kvm_run {
3049	/* in */
3050	__u8 request_interrupt_window;
3051
3052Request that KVM_RUN return when it becomes possible to inject external
3053interrupts into the guest.  Useful in conjunction with KVM_INTERRUPT.
3054
3055	__u8 padding1[7];
3056
3057	/* out */
3058	__u32 exit_reason;
3059
3060When KVM_RUN has returned successfully (return value 0), this informs
3061application code why KVM_RUN has returned.  Allowable values for this
3062field are detailed below.
3063
3064	__u8 ready_for_interrupt_injection;
3065
3066If request_interrupt_window has been specified, this field indicates
3067an interrupt can be injected now with KVM_INTERRUPT.
3068
3069	__u8 if_flag;
3070
3071The value of the current interrupt flag.  Only valid if in-kernel
3072local APIC is not used.
3073
3074	__u16 flags;
3075
3076More architecture-specific flags detailing state of the VCPU that may
3077affect the device's behavior.  The only currently defined flag is
3078KVM_RUN_X86_SMM, which is valid on x86 machines and is set if the
3079VCPU is in system management mode.
3080
3081	/* in (pre_kvm_run), out (post_kvm_run) */
3082	__u64 cr8;
3083
3084The value of the cr8 register.  Only valid if in-kernel local APIC is
3085not used.  Both input and output.
3086
3087	__u64 apic_base;
3088
3089The value of the APIC BASE msr.  Only valid if in-kernel local
3090APIC is not used.  Both input and output.
3091
3092	union {
3093		/* KVM_EXIT_UNKNOWN */
3094		struct {
3095			__u64 hardware_exit_reason;
3096		} hw;
3097
3098If exit_reason is KVM_EXIT_UNKNOWN, the vcpu has exited due to unknown
3099reasons.  Further architecture-specific information is available in
3100hardware_exit_reason.
3101
3102		/* KVM_EXIT_FAIL_ENTRY */
3103		struct {
3104			__u64 hardware_entry_failure_reason;
3105		} fail_entry;
3106
3107If exit_reason is KVM_EXIT_FAIL_ENTRY, the vcpu could not be run due
3108to unknown reasons.  Further architecture-specific information is
3109available in hardware_entry_failure_reason.
3110
3111		/* KVM_EXIT_EXCEPTION */
3112		struct {
3113			__u32 exception;
3114			__u32 error_code;
3115		} ex;
3116
3117Unused.
3118
3119		/* KVM_EXIT_IO */
3120		struct {
3121#define KVM_EXIT_IO_IN  0
3122#define KVM_EXIT_IO_OUT 1
3123			__u8 direction;
3124			__u8 size; /* bytes */
3125			__u16 port;
3126			__u32 count;
3127			__u64 data_offset; /* relative to kvm_run start */
3128		} io;
3129
3130If exit_reason is KVM_EXIT_IO, then the vcpu has
3131executed a port I/O instruction which could not be satisfied by kvm.
3132data_offset describes where the data is located (KVM_EXIT_IO_OUT) or
3133where kvm expects application code to place the data for the next
3134KVM_RUN invocation (KVM_EXIT_IO_IN).  Data format is a packed array.
3135
3136		/* KVM_EXIT_DEBUG */
3137		struct {
3138			struct kvm_debug_exit_arch arch;
3139		} debug;
3140
3141If the exit_reason is KVM_EXIT_DEBUG, then a vcpu is processing a debug event
3142for which architecture specific information is returned.
3143
3144		/* KVM_EXIT_MMIO */
3145		struct {
3146			__u64 phys_addr;
3147			__u8  data[8];
3148			__u32 len;
3149			__u8  is_write;
3150		} mmio;
3151
3152If exit_reason is KVM_EXIT_MMIO, then the vcpu has
3153executed a memory-mapped I/O instruction which could not be satisfied
3154by kvm.  The 'data' member contains the written data if 'is_write' is
3155true, and should be filled by application code otherwise.
3156
3157The 'data' member contains, in its first 'len' bytes, the value as it would
3158appear if the VCPU performed a load or store of the appropriate width directly
3159to the byte array.
3160
3161NOTE: For KVM_EXIT_IO, KVM_EXIT_MMIO, KVM_EXIT_OSI, KVM_EXIT_PAPR and
3162      KVM_EXIT_EPR the corresponding
3163operations are complete (and guest state is consistent) only after userspace
3164has re-entered the kernel with KVM_RUN.  The kernel side will first finish
3165incomplete operations and then check for pending signals.  Userspace
3166can re-enter the guest with an unmasked signal pending to complete
3167pending operations.
3168
3169		/* KVM_EXIT_HYPERCALL */
3170		struct {
3171			__u64 nr;
3172			__u64 args[6];
3173			__u64 ret;
3174			__u32 longmode;
3175			__u32 pad;
3176		} hypercall;
3177
3178Unused.  This was once used for 'hypercall to userspace'.  To implement
3179such functionality, use KVM_EXIT_IO (x86) or KVM_EXIT_MMIO (all except s390).
3180Note KVM_EXIT_IO is significantly faster than KVM_EXIT_MMIO.
3181
3182		/* KVM_EXIT_TPR_ACCESS */
3183		struct {
3184			__u64 rip;
3185			__u32 is_write;
3186			__u32 pad;
3187		} tpr_access;
3188
3189To be documented (KVM_TPR_ACCESS_REPORTING).
3190
3191		/* KVM_EXIT_S390_SIEIC */
3192		struct {
3193			__u8 icptcode;
3194			__u64 mask; /* psw upper half */
3195			__u64 addr; /* psw lower half */
3196			__u16 ipa;
3197			__u32 ipb;
3198		} s390_sieic;
3199
3200s390 specific.
3201
3202		/* KVM_EXIT_S390_RESET */
3203#define KVM_S390_RESET_POR       1
3204#define KVM_S390_RESET_CLEAR     2
3205#define KVM_S390_RESET_SUBSYSTEM 4
3206#define KVM_S390_RESET_CPU_INIT  8
3207#define KVM_S390_RESET_IPL       16
3208		__u64 s390_reset_flags;
3209
3210s390 specific.
3211
3212		/* KVM_EXIT_S390_UCONTROL */
3213		struct {
3214			__u64 trans_exc_code;
3215			__u32 pgm_code;
3216		} s390_ucontrol;
3217
3218s390 specific. A page fault has occurred for a user controlled virtual
3219machine (KVM_VM_S390_UNCONTROL) on it's host page table that cannot be
3220resolved by the kernel.
3221The program code and the translation exception code that were placed
3222in the cpu's lowcore are presented here as defined by the z Architecture
3223Principles of Operation Book in the Chapter for Dynamic Address Translation
3224(DAT)
3225
3226		/* KVM_EXIT_DCR */
3227		struct {
3228			__u32 dcrn;
3229			__u32 data;
3230			__u8  is_write;
3231		} dcr;
3232
3233Deprecated - was used for 440 KVM.
3234
3235		/* KVM_EXIT_OSI */
3236		struct {
3237			__u64 gprs[32];
3238		} osi;
3239
3240MOL uses a special hypercall interface it calls 'OSI'. To enable it, we catch
3241hypercalls and exit with this exit struct that contains all the guest gprs.
3242
3243If exit_reason is KVM_EXIT_OSI, then the vcpu has triggered such a hypercall.
3244Userspace can now handle the hypercall and when it's done modify the gprs as
3245necessary. Upon guest entry all guest GPRs will then be replaced by the values
3246in this struct.
3247
3248		/* KVM_EXIT_PAPR_HCALL */
3249		struct {
3250			__u64 nr;
3251			__u64 ret;
3252			__u64 args[9];
3253		} papr_hcall;
3254
3255This is used on 64-bit PowerPC when emulating a pSeries partition,
3256e.g. with the 'pseries' machine type in qemu.  It occurs when the
3257guest does a hypercall using the 'sc 1' instruction.  The 'nr' field
3258contains the hypercall number (from the guest R3), and 'args' contains
3259the arguments (from the guest R4 - R12).  Userspace should put the
3260return code in 'ret' and any extra returned values in args[].
3261The possible hypercalls are defined in the Power Architecture Platform
3262Requirements (PAPR) document available from www.power.org (free
3263developer registration required to access it).
3264
3265		/* KVM_EXIT_S390_TSCH */
3266		struct {
3267			__u16 subchannel_id;
3268			__u16 subchannel_nr;
3269			__u32 io_int_parm;
3270			__u32 io_int_word;
3271			__u32 ipb;
3272			__u8 dequeued;
3273		} s390_tsch;
3274
3275s390 specific. This exit occurs when KVM_CAP_S390_CSS_SUPPORT has been enabled
3276and TEST SUBCHANNEL was intercepted. If dequeued is set, a pending I/O
3277interrupt for the target subchannel has been dequeued and subchannel_id,
3278subchannel_nr, io_int_parm and io_int_word contain the parameters for that
3279interrupt. ipb is needed for instruction parameter decoding.
3280
3281		/* KVM_EXIT_EPR */
3282		struct {
3283			__u32 epr;
3284		} epr;
3285
3286On FSL BookE PowerPC chips, the interrupt controller has a fast patch
3287interrupt acknowledge path to the core. When the core successfully
3288delivers an interrupt, it automatically populates the EPR register with
3289the interrupt vector number and acknowledges the interrupt inside
3290the interrupt controller.
3291
3292In case the interrupt controller lives in user space, we need to do
3293the interrupt acknowledge cycle through it to fetch the next to be
3294delivered interrupt vector using this exit.
3295
3296It gets triggered whenever both KVM_CAP_PPC_EPR are enabled and an
3297external interrupt has just been delivered into the guest. User space
3298should put the acknowledged interrupt vector into the 'epr' field.
3299
3300		/* KVM_EXIT_SYSTEM_EVENT */
3301		struct {
3302#define KVM_SYSTEM_EVENT_SHUTDOWN       1
3303#define KVM_SYSTEM_EVENT_RESET          2
3304#define KVM_SYSTEM_EVENT_CRASH          3
3305			__u32 type;
3306			__u64 flags;
3307		} system_event;
3308
3309If exit_reason is KVM_EXIT_SYSTEM_EVENT then the vcpu has triggered
3310a system-level event using some architecture specific mechanism (hypercall
3311or some special instruction). In case of ARM/ARM64, this is triggered using
3312HVC instruction based PSCI call from the vcpu. The 'type' field describes
3313the system-level event type. The 'flags' field describes architecture
3314specific flags for the system-level event.
3315
3316Valid values for 'type' are:
3317  KVM_SYSTEM_EVENT_SHUTDOWN -- the guest has requested a shutdown of the
3318   VM. Userspace is not obliged to honour this, and if it does honour
3319   this does not need to destroy the VM synchronously (ie it may call
3320   KVM_RUN again before shutdown finally occurs).
3321  KVM_SYSTEM_EVENT_RESET -- the guest has requested a reset of the VM.
3322   As with SHUTDOWN, userspace can choose to ignore the request, or
3323   to schedule the reset to occur in the future and may call KVM_RUN again.
3324  KVM_SYSTEM_EVENT_CRASH -- the guest crash occurred and the guest
3325   has requested a crash condition maintenance. Userspace can choose
3326   to ignore the request, or to gather VM memory core dump and/or
3327   reset/shutdown of the VM.
3328
3329		/* KVM_EXIT_IOAPIC_EOI */
3330		struct {
3331			__u8 vector;
3332		} eoi;
3333
3334Indicates that the VCPU's in-kernel local APIC received an EOI for a
3335level-triggered IOAPIC interrupt.  This exit only triggers when the
3336IOAPIC is implemented in userspace (i.e. KVM_CAP_SPLIT_IRQCHIP is enabled);
3337the userspace IOAPIC should process the EOI and retrigger the interrupt if
3338it is still asserted.  Vector is the LAPIC interrupt vector for which the
3339EOI was received.
3340
3341		/* Fix the size of the union. */
3342		char padding[256];
3343	};
3344
3345	/*
3346	 * shared registers between kvm and userspace.
3347	 * kvm_valid_regs specifies the register classes set by the host
3348	 * kvm_dirty_regs specified the register classes dirtied by userspace
3349	 * struct kvm_sync_regs is architecture specific, as well as the
3350	 * bits for kvm_valid_regs and kvm_dirty_regs
3351	 */
3352	__u64 kvm_valid_regs;
3353	__u64 kvm_dirty_regs;
3354	union {
3355		struct kvm_sync_regs regs;
3356		char padding[1024];
3357	} s;
3358
3359If KVM_CAP_SYNC_REGS is defined, these fields allow userspace to access
3360certain guest registers without having to call SET/GET_*REGS. Thus we can
3361avoid some system call overhead if userspace has to handle the exit.
3362Userspace can query the validity of the structure by checking
3363kvm_valid_regs for specific bits. These bits are architecture specific
3364and usually define the validity of a groups of registers. (e.g. one bit
3365 for general purpose registers)
3366
3367Please note that the kernel is allowed to use the kvm_run structure as the
3368primary storage for certain register types. Therefore, the kernel may use the
3369values in kvm_run even if the corresponding bit in kvm_dirty_regs is not set.
3370
3371};
3372
3373
3374
33756. Capabilities that can be enabled on vCPUs
3376--------------------------------------------
3377
3378There are certain capabilities that change the behavior of the virtual CPU or
3379the virtual machine when enabled. To enable them, please see section 4.37.
3380Below you can find a list of capabilities and what their effect on the vCPU or
3381the virtual machine is when enabling them.
3382
3383The following information is provided along with the description:
3384
3385  Architectures: which instruction set architectures provide this ioctl.
3386      x86 includes both i386 and x86_64.
3387
3388  Target: whether this is a per-vcpu or per-vm capability.
3389
3390  Parameters: what parameters are accepted by the capability.
3391
3392  Returns: the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
3393      are not detailed, but errors with specific meanings are.
3394
3395
33966.1 KVM_CAP_PPC_OSI
3397
3398Architectures: ppc
3399Target: vcpu
3400Parameters: none
3401Returns: 0 on success; -1 on error
3402
3403This capability enables interception of OSI hypercalls that otherwise would
3404be treated as normal system calls to be injected into the guest. OSI hypercalls
3405were invented by Mac-on-Linux to have a standardized communication mechanism
3406between the guest and the host.
3407
3408When this capability is enabled, KVM_EXIT_OSI can occur.
3409
3410
34116.2 KVM_CAP_PPC_PAPR
3412
3413Architectures: ppc
3414Target: vcpu
3415Parameters: none
3416Returns: 0 on success; -1 on error
3417
3418This capability enables interception of PAPR hypercalls. PAPR hypercalls are
3419done using the hypercall instruction "sc 1".
3420
3421It also sets the guest privilege level to "supervisor" mode. Usually the guest
3422runs in "hypervisor" privilege mode with a few missing features.
3423
3424In addition to the above, it changes the semantics of SDR1. In this mode, the
3425HTAB address part of SDR1 contains an HVA instead of a GPA, as PAPR keeps the
3426HTAB invisible to the guest.
3427
3428When this capability is enabled, KVM_EXIT_PAPR_HCALL can occur.
3429
3430
34316.3 KVM_CAP_SW_TLB
3432
3433Architectures: ppc
3434Target: vcpu
3435Parameters: args[0] is the address of a struct kvm_config_tlb
3436Returns: 0 on success; -1 on error
3437
3438struct kvm_config_tlb {
3439	__u64 params;
3440	__u64 array;
3441	__u32 mmu_type;
3442	__u32 array_len;
3443};
3444
3445Configures the virtual CPU's TLB array, establishing a shared memory area
3446between userspace and KVM.  The "params" and "array" fields are userspace
3447addresses of mmu-type-specific data structures.  The "array_len" field is an
3448safety mechanism, and should be set to the size in bytes of the memory that
3449userspace has reserved for the array.  It must be at least the size dictated
3450by "mmu_type" and "params".
3451
3452While KVM_RUN is active, the shared region is under control of KVM.  Its
3453contents are undefined, and any modification by userspace results in
3454boundedly undefined behavior.
3455
3456On return from KVM_RUN, the shared region will reflect the current state of
3457the guest's TLB.  If userspace makes any changes, it must call KVM_DIRTY_TLB
3458to tell KVM which entries have been changed, prior to calling KVM_RUN again
3459on this vcpu.
3460
3461For mmu types KVM_MMU_FSL_BOOKE_NOHV and KVM_MMU_FSL_BOOKE_HV:
3462 - The "params" field is of type "struct kvm_book3e_206_tlb_params".
3463 - The "array" field points to an array of type "struct
3464   kvm_book3e_206_tlb_entry".
3465 - The array consists of all entries in the first TLB, followed by all
3466   entries in the second TLB.
3467 - Within a TLB, entries are ordered first by increasing set number.  Within a
3468   set, entries are ordered by way (increasing ESEL).
3469 - The hash for determining set number in TLB0 is: (MAS2 >> 12) & (num_sets - 1)
3470   where "num_sets" is the tlb_sizes[] value divided by the tlb_ways[] value.
3471 - The tsize field of mas1 shall be set to 4K on TLB0, even though the
3472   hardware ignores this value for TLB0.
3473
34746.4 KVM_CAP_S390_CSS_SUPPORT
3475
3476Architectures: s390
3477Target: vcpu
3478Parameters: none
3479Returns: 0 on success; -1 on error
3480
3481This capability enables support for handling of channel I/O instructions.
3482
3483TEST PENDING INTERRUPTION and the interrupt portion of TEST SUBCHANNEL are
3484handled in-kernel, while the other I/O instructions are passed to userspace.
3485
3486When this capability is enabled, KVM_EXIT_S390_TSCH will occur on TEST
3487SUBCHANNEL intercepts.
3488
3489Note that even though this capability is enabled per-vcpu, the complete
3490virtual machine is affected.
3491
34926.5 KVM_CAP_PPC_EPR
3493
3494Architectures: ppc
3495Target: vcpu
3496Parameters: args[0] defines whether the proxy facility is active
3497Returns: 0 on success; -1 on error
3498
3499This capability enables or disables the delivery of interrupts through the
3500external proxy facility.
3501
3502When enabled (args[0] != 0), every time the guest gets an external interrupt
3503delivered, it automatically exits into user space with a KVM_EXIT_EPR exit
3504to receive the topmost interrupt vector.
3505
3506When disabled (args[0] == 0), behavior is as if this facility is unsupported.
3507
3508When this capability is enabled, KVM_EXIT_EPR can occur.
3509
35106.6 KVM_CAP_IRQ_MPIC
3511
3512Architectures: ppc
3513Parameters: args[0] is the MPIC device fd
3514            args[1] is the MPIC CPU number for this vcpu
3515
3516This capability connects the vcpu to an in-kernel MPIC device.
3517
35186.7 KVM_CAP_IRQ_XICS
3519
3520Architectures: ppc
3521Target: vcpu
3522Parameters: args[0] is the XICS device fd
3523            args[1] is the XICS CPU number (server ID) for this vcpu
3524
3525This capability connects the vcpu to an in-kernel XICS device.
3526
35276.8 KVM_CAP_S390_IRQCHIP
3528
3529Architectures: s390
3530Target: vm
3531Parameters: none
3532
3533This capability enables the in-kernel irqchip for s390. Please refer to
3534"4.24 KVM_CREATE_IRQCHIP" for details.
3535
35366.9 KVM_CAP_MIPS_FPU
3537
3538Architectures: mips
3539Target: vcpu
3540Parameters: args[0] is reserved for future use (should be 0).
3541
3542This capability allows the use of the host Floating Point Unit by the guest. It
3543allows the Config1.FP bit to be set to enable the FPU in the guest. Once this is
3544done the KVM_REG_MIPS_FPR_* and KVM_REG_MIPS_FCR_* registers can be accessed
3545(depending on the current guest FPU register mode), and the Status.FR,
3546Config5.FRE bits are accessible via the KVM API and also from the guest,
3547depending on them being supported by the FPU.
3548
35496.10 KVM_CAP_MIPS_MSA
3550
3551Architectures: mips
3552Target: vcpu
3553Parameters: args[0] is reserved for future use (should be 0).
3554
3555This capability allows the use of the MIPS SIMD Architecture (MSA) by the guest.
3556It allows the Config3.MSAP bit to be set to enable the use of MSA by the guest.
3557Once this is done the KVM_REG_MIPS_VEC_* and KVM_REG_MIPS_MSA_* registers can be
3558accessed, and the Config5.MSAEn bit is accessible via the KVM API and also from
3559the guest.
3560
35617. Capabilities that can be enabled on VMs
3562------------------------------------------
3563
3564There are certain capabilities that change the behavior of the virtual
3565machine when enabled. To enable them, please see section 4.37. Below
3566you can find a list of capabilities and what their effect on the VM
3567is when enabling them.
3568
3569The following information is provided along with the description:
3570
3571  Architectures: which instruction set architectures provide this ioctl.
3572      x86 includes both i386 and x86_64.
3573
3574  Parameters: what parameters are accepted by the capability.
3575
3576  Returns: the return value.  General error numbers (EBADF, ENOMEM, EINVAL)
3577      are not detailed, but errors with specific meanings are.
3578
3579
35807.1 KVM_CAP_PPC_ENABLE_HCALL
3581
3582Architectures: ppc
3583Parameters: args[0] is the sPAPR hcall number
3584	    args[1] is 0 to disable, 1 to enable in-kernel handling
3585
3586This capability controls whether individual sPAPR hypercalls (hcalls)
3587get handled by the kernel or not.  Enabling or disabling in-kernel
3588handling of an hcall is effective across the VM.  On creation, an
3589initial set of hcalls are enabled for in-kernel handling, which
3590consists of those hcalls for which in-kernel handlers were implemented
3591before this capability was implemented.  If disabled, the kernel will
3592not to attempt to handle the hcall, but will always exit to userspace
3593to handle it.  Note that it may not make sense to enable some and
3594disable others of a group of related hcalls, but KVM does not prevent
3595userspace from doing that.
3596
3597If the hcall number specified is not one that has an in-kernel
3598implementation, the KVM_ENABLE_CAP ioctl will fail with an EINVAL
3599error.
3600
36017.2 KVM_CAP_S390_USER_SIGP
3602
3603Architectures: s390
3604Parameters: none
3605
3606This capability controls which SIGP orders will be handled completely in user
3607space. With this capability enabled, all fast orders will be handled completely
3608in the kernel:
3609- SENSE
3610- SENSE RUNNING
3611- EXTERNAL CALL
3612- EMERGENCY SIGNAL
3613- CONDITIONAL EMERGENCY SIGNAL
3614
3615All other orders will be handled completely in user space.
3616
3617Only privileged operation exceptions will be checked for in the kernel (or even
3618in the hardware prior to interception). If this capability is not enabled, the
3619old way of handling SIGP orders is used (partially in kernel and user space).
3620
36217.3 KVM_CAP_S390_VECTOR_REGISTERS
3622
3623Architectures: s390
3624Parameters: none
3625Returns: 0 on success, negative value on error
3626
3627Allows use of the vector registers introduced with z13 processor, and
3628provides for the synchronization between host and user space.  Will
3629return -EINVAL if the machine does not support vectors.
3630
36317.4 KVM_CAP_S390_USER_STSI
3632
3633Architectures: s390
3634Parameters: none
3635
3636This capability allows post-handlers for the STSI instruction. After
3637initial handling in the kernel, KVM exits to user space with
3638KVM_EXIT_S390_STSI to allow user space to insert further data.
3639
3640Before exiting to userspace, kvm handlers should fill in s390_stsi field of
3641vcpu->run:
3642struct {
3643	__u64 addr;
3644	__u8 ar;
3645	__u8 reserved;
3646	__u8 fc;
3647	__u8 sel1;
3648	__u16 sel2;
3649} s390_stsi;
3650
3651@addr - guest address of STSI SYSIB
3652@fc   - function code
3653@sel1 - selector 1
3654@sel2 - selector 2
3655@ar   - access register number
3656
3657KVM handlers should exit to userspace with rc = -EREMOTE.
3658
36597.5 KVM_CAP_SPLIT_IRQCHIP
3660
3661Architectures: x86
3662Parameters: args[0] - number of routes reserved for userspace IOAPICs
3663Returns: 0 on success, -1 on error
3664
3665Create a local apic for each processor in the kernel. This can be used
3666instead of KVM_CREATE_IRQCHIP if the userspace VMM wishes to emulate the
3667IOAPIC and PIC (and also the PIT, even though this has to be enabled
3668separately).
3669
3670This capability also enables in kernel routing of interrupt requests;
3671when KVM_CAP_SPLIT_IRQCHIP only routes of KVM_IRQ_ROUTING_MSI type are
3672used in the IRQ routing table.  The first args[0] MSI routes are reserved
3673for the IOAPIC pins.  Whenever the LAPIC receives an EOI for these routes,
3674a KVM_EXIT_IOAPIC_EOI vmexit will be reported to userspace.
3675
3676Fails if VCPU has already been created, or if the irqchip is already in the
3677kernel (i.e. KVM_CREATE_IRQCHIP has already been called).
3678
3679
36808. Other capabilities.
3681----------------------
3682
3683This section lists capabilities that give information about other
3684features of the KVM implementation.
3685
36868.1 KVM_CAP_PPC_HWRNG
3687
3688Architectures: ppc
3689
3690This capability, if KVM_CHECK_EXTENSION indicates that it is
3691available, means that that the kernel has an implementation of the
3692H_RANDOM hypercall backed by a hardware random-number generator.
3693If present, the kernel H_RANDOM handler can be enabled for guest use
3694with the KVM_CAP_PPC_ENABLE_HCALL capability.
3695