vmm.h revision 288943
1/*-
2 * Copyright (c) 2011 NetApp, Inc.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 *
14 * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED.  IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24 * SUCH DAMAGE.
25 *
26 * $FreeBSD: head/sys/amd64/include/vmm.h 288943 2015-10-06 17:53:29Z dim $
27 */
28
29#ifndef _VMM_H_
30#define	_VMM_H_
31
32#include <x86/segments.h>
33
34enum vm_suspend_how {
35	VM_SUSPEND_NONE,
36	VM_SUSPEND_RESET,
37	VM_SUSPEND_POWEROFF,
38	VM_SUSPEND_HALT,
39	VM_SUSPEND_TRIPLEFAULT,
40	VM_SUSPEND_LAST
41};
42
43/*
44 * Identifiers for architecturally defined registers.
45 */
46enum vm_reg_name {
47	VM_REG_GUEST_RAX,
48	VM_REG_GUEST_RBX,
49	VM_REG_GUEST_RCX,
50	VM_REG_GUEST_RDX,
51	VM_REG_GUEST_RSI,
52	VM_REG_GUEST_RDI,
53	VM_REG_GUEST_RBP,
54	VM_REG_GUEST_R8,
55	VM_REG_GUEST_R9,
56	VM_REG_GUEST_R10,
57	VM_REG_GUEST_R11,
58	VM_REG_GUEST_R12,
59	VM_REG_GUEST_R13,
60	VM_REG_GUEST_R14,
61	VM_REG_GUEST_R15,
62	VM_REG_GUEST_CR0,
63	VM_REG_GUEST_CR3,
64	VM_REG_GUEST_CR4,
65	VM_REG_GUEST_DR7,
66	VM_REG_GUEST_RSP,
67	VM_REG_GUEST_RIP,
68	VM_REG_GUEST_RFLAGS,
69	VM_REG_GUEST_ES,
70	VM_REG_GUEST_CS,
71	VM_REG_GUEST_SS,
72	VM_REG_GUEST_DS,
73	VM_REG_GUEST_FS,
74	VM_REG_GUEST_GS,
75	VM_REG_GUEST_LDTR,
76	VM_REG_GUEST_TR,
77	VM_REG_GUEST_IDTR,
78	VM_REG_GUEST_GDTR,
79	VM_REG_GUEST_EFER,
80	VM_REG_GUEST_CR2,
81	VM_REG_GUEST_PDPTE0,
82	VM_REG_GUEST_PDPTE1,
83	VM_REG_GUEST_PDPTE2,
84	VM_REG_GUEST_PDPTE3,
85	VM_REG_GUEST_INTR_SHADOW,
86	VM_REG_LAST
87};
88
89enum x2apic_state {
90	X2APIC_DISABLED,
91	X2APIC_ENABLED,
92	X2APIC_STATE_LAST
93};
94
95#define	VM_INTINFO_VECTOR(info)	((info) & 0xff)
96#define	VM_INTINFO_DEL_ERRCODE	0x800
97#define	VM_INTINFO_RSVD		0x7ffff000
98#define	VM_INTINFO_VALID	0x80000000
99#define	VM_INTINFO_TYPE		0x700
100#define	VM_INTINFO_HWINTR	(0 << 8)
101#define	VM_INTINFO_NMI		(2 << 8)
102#define	VM_INTINFO_HWEXCEPTION	(3 << 8)
103#define	VM_INTINFO_SWINTR	(4 << 8)
104
105#ifdef _KERNEL
106
107#define	VM_MAX_NAMELEN	32
108
109struct vm;
110struct vm_exception;
111struct seg_desc;
112struct vm_exit;
113struct vm_run;
114struct vhpet;
115struct vioapic;
116struct vlapic;
117struct vmspace;
118struct vm_object;
119struct vm_guest_paging;
120struct pmap;
121
122struct vm_eventinfo {
123	void	*rptr;		/* rendezvous cookie */
124	int	*sptr;		/* suspend cookie */
125	int	*iptr;		/* reqidle cookie */
126};
127
128typedef int	(*vmm_init_func_t)(int ipinum);
129typedef int	(*vmm_cleanup_func_t)(void);
130typedef void	(*vmm_resume_func_t)(void);
131typedef void *	(*vmi_init_func_t)(struct vm *vm, struct pmap *pmap);
132typedef int	(*vmi_run_func_t)(void *vmi, int vcpu, register_t rip,
133		    struct pmap *pmap, struct vm_eventinfo *info);
134typedef void	(*vmi_cleanup_func_t)(void *vmi);
135typedef int	(*vmi_get_register_t)(void *vmi, int vcpu, int num,
136				      uint64_t *retval);
137typedef int	(*vmi_set_register_t)(void *vmi, int vcpu, int num,
138				      uint64_t val);
139typedef int	(*vmi_get_desc_t)(void *vmi, int vcpu, int num,
140				  struct seg_desc *desc);
141typedef int	(*vmi_set_desc_t)(void *vmi, int vcpu, int num,
142				  struct seg_desc *desc);
143typedef int	(*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval);
144typedef int	(*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val);
145typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max);
146typedef void	(*vmi_vmspace_free)(struct vmspace *vmspace);
147typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu);
148typedef void	(*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic);
149
150struct vmm_ops {
151	vmm_init_func_t		init;		/* module wide initialization */
152	vmm_cleanup_func_t	cleanup;
153	vmm_resume_func_t	resume;
154
155	vmi_init_func_t		vminit;		/* vm-specific initialization */
156	vmi_run_func_t		vmrun;
157	vmi_cleanup_func_t	vmcleanup;
158	vmi_get_register_t	vmgetreg;
159	vmi_set_register_t	vmsetreg;
160	vmi_get_desc_t		vmgetdesc;
161	vmi_set_desc_t		vmsetdesc;
162	vmi_get_cap_t		vmgetcap;
163	vmi_set_cap_t		vmsetcap;
164	vmi_vmspace_alloc	vmspace_alloc;
165	vmi_vmspace_free	vmspace_free;
166	vmi_vlapic_init		vlapic_init;
167	vmi_vlapic_cleanup	vlapic_cleanup;
168};
169
170extern struct vmm_ops vmm_ops_intel;
171extern struct vmm_ops vmm_ops_amd;
172
173int vm_create(const char *name, struct vm **retvm);
174void vm_destroy(struct vm *vm);
175int vm_reinit(struct vm *vm);
176const char *vm_name(struct vm *vm);
177
178/*
179 * APIs that modify the guest memory map require all vcpus to be frozen.
180 */
181int vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t off,
182    size_t len, int prot, int flags);
183int vm_alloc_memseg(struct vm *vm, int ident, size_t len, bool sysmem);
184void vm_free_memseg(struct vm *vm, int ident);
185int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
186int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
187int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
188int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
189
190/*
191 * APIs that inspect the guest memory map require only a *single* vcpu to
192 * be frozen. This acts like a read lock on the guest memory map since any
193 * modification requires *all* vcpus to be frozen.
194 */
195int vm_mmap_getnext(struct vm *vm, vm_paddr_t *gpa, int *segid,
196    vm_ooffset_t *segoff, size_t *len, int *prot, int *flags);
197int vm_get_memseg(struct vm *vm, int ident, size_t *len, bool *sysmem,
198    struct vm_object **objptr);
199void *vm_gpa_hold(struct vm *, int vcpuid, vm_paddr_t gpa, size_t len,
200    int prot, void **cookie);
201void vm_gpa_release(void *cookie);
202bool vm_mem_allocated(struct vm *vm, int vcpuid, vm_paddr_t gpa);
203
204int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval);
205int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val);
206int vm_get_seg_desc(struct vm *vm, int vcpu, int reg,
207		    struct seg_desc *ret_desc);
208int vm_set_seg_desc(struct vm *vm, int vcpu, int reg,
209		    struct seg_desc *desc);
210int vm_run(struct vm *vm, struct vm_run *vmrun);
211int vm_suspend(struct vm *vm, enum vm_suspend_how how);
212int vm_inject_nmi(struct vm *vm, int vcpu);
213int vm_nmi_pending(struct vm *vm, int vcpuid);
214void vm_nmi_clear(struct vm *vm, int vcpuid);
215int vm_inject_extint(struct vm *vm, int vcpu);
216int vm_extint_pending(struct vm *vm, int vcpuid);
217void vm_extint_clear(struct vm *vm, int vcpuid);
218struct vlapic *vm_lapic(struct vm *vm, int cpu);
219struct vioapic *vm_ioapic(struct vm *vm);
220struct vhpet *vm_hpet(struct vm *vm);
221int vm_get_capability(struct vm *vm, int vcpu, int type, int *val);
222int vm_set_capability(struct vm *vm, int vcpu, int type, int val);
223int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state);
224int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state);
225int vm_apicid2vcpuid(struct vm *vm, int apicid);
226int vm_activate_cpu(struct vm *vm, int vcpu);
227struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid);
228void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip);
229void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip);
230void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip);
231void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip);
232
233#ifdef _SYS__CPUSET_H_
234/*
235 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'.
236 * The rendezvous 'func(arg)' is not allowed to do anything that will
237 * cause the thread to be put to sleep.
238 *
239 * If the rendezvous is being initiated from a vcpu context then the
240 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1.
241 *
242 * The caller cannot hold any locks when initiating the rendezvous.
243 *
244 * The implementation of this API may cause vcpus other than those specified
245 * by 'dest' to be stalled. The caller should not rely on any vcpus making
246 * forward progress when the rendezvous is in progress.
247 */
248typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg);
249void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
250    vm_rendezvous_func_t func, void *arg);
251cpuset_t vm_active_cpus(struct vm *vm);
252cpuset_t vm_suspended_cpus(struct vm *vm);
253#endif	/* _SYS__CPUSET_H_ */
254
255static __inline int
256vcpu_rendezvous_pending(struct vm_eventinfo *info)
257{
258
259	return (*((uintptr_t *)(info->rptr)) != 0);
260}
261
262static __inline int
263vcpu_suspended(struct vm_eventinfo *info)
264{
265
266	return (*info->sptr);
267}
268
269static __inline int
270vcpu_reqidle(struct vm_eventinfo *info)
271{
272
273	return (*info->iptr);
274}
275
276/*
277 * Return 1 if device indicated by bus/slot/func is supposed to be a
278 * pci passthrough device.
279 *
280 * Return 0 otherwise.
281 */
282int vmm_is_pptdev(int bus, int slot, int func);
283
284void *vm_iommu_domain(struct vm *vm);
285
286enum vcpu_state {
287	VCPU_IDLE,
288	VCPU_FROZEN,
289	VCPU_RUNNING,
290	VCPU_SLEEPING,
291};
292
293int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
294    bool from_idle);
295enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
296
297static int __inline
298vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
299{
300	return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
301}
302
303#ifdef _SYS_PROC_H_
304static int __inline
305vcpu_should_yield(struct vm *vm, int vcpu)
306{
307
308	if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED))
309		return (1);
310	else if (curthread->td_owepreempt)
311		return (1);
312	else
313		return (0);
314}
315#endif
316
317void *vcpu_stats(struct vm *vm, int vcpu);
318void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
319struct vmspace *vm_get_vmspace(struct vm *vm);
320struct vatpic *vm_atpic(struct vm *vm);
321struct vatpit *vm_atpit(struct vm *vm);
322struct vpmtmr *vm_pmtmr(struct vm *vm);
323struct vrtc *vm_rtc(struct vm *vm);
324
325/*
326 * Inject exception 'vector' into the guest vcpu. This function returns 0 on
327 * success and non-zero on failure.
328 *
329 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
330 * this function directly because they enforce the trap-like or fault-like
331 * behavior of an exception.
332 *
333 * This function should only be called in the context of the thread that is
334 * executing this vcpu.
335 */
336int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid,
337    uint32_t errcode, int restart_instruction);
338
339/*
340 * This function is called after a VM-exit that occurred during exception or
341 * interrupt delivery through the IDT. The format of 'intinfo' is described
342 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
343 *
344 * If a VM-exit handler completes the event delivery successfully then it
345 * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
346 * if the task switch emulation is triggered via a task gate then it should
347 * call this function with 'intinfo=0' to indicate that the external event
348 * is not pending anymore.
349 *
350 * Return value is 0 on success and non-zero on failure.
351 */
352int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo);
353
354/*
355 * This function is called before every VM-entry to retrieve a pending
356 * event that should be injected into the guest. This function combines
357 * nested events into a double or triple fault.
358 *
359 * Returns 0 if there are no events that need to be injected into the guest
360 * and non-zero otherwise.
361 */
362int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info);
363
364int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2);
365
366enum vm_reg_name vm_segment_name(int seg_encoding);
367
368struct vm_copyinfo {
369	uint64_t	gpa;
370	size_t		len;
371	void		*hva;
372	void		*cookie;
373};
374
375/*
376 * Set up 'copyinfo[]' to copy to/from guest linear address space starting
377 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
378 * a copyin or PROT_WRITE for a copyout.
379 *
380 * retval	is_fault	Intepretation
381 *   0		   0		Success
382 *   0		   1		An exception was injected into the guest
383 * EFAULT	  N/A		Unrecoverable error
384 *
385 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
386 * the return value is 0. The 'copyinfo[]' resources should be freed by calling
387 * 'vm_copy_teardown()' after the copy is done.
388 */
389int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging,
390    uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
391    int num_copyinfo, int *is_fault);
392void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
393    int num_copyinfo);
394void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
395    void *kaddr, size_t len);
396void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr,
397    struct vm_copyinfo *copyinfo, size_t len);
398
399int vcpu_trace_exceptions(struct vm *vm, int vcpuid);
400#endif	/* KERNEL */
401
402#define	VM_MAXCPU	16			/* maximum virtual cpus */
403
404/*
405 * Identifiers for optional vmm capabilities
406 */
407enum vm_cap_type {
408	VM_CAP_HALT_EXIT,
409	VM_CAP_MTRAP_EXIT,
410	VM_CAP_PAUSE_EXIT,
411	VM_CAP_UNRESTRICTED_GUEST,
412	VM_CAP_ENABLE_INVPCID,
413	VM_CAP_MAX
414};
415
416enum vm_intr_trigger {
417	EDGE_TRIGGER,
418	LEVEL_TRIGGER
419};
420
421/*
422 * The 'access' field has the format specified in Table 21-2 of the Intel
423 * Architecture Manual vol 3b.
424 *
425 * XXX The contents of the 'access' field are architecturally defined except
426 * bit 16 - Segment Unusable.
427 */
428struct seg_desc {
429	uint64_t	base;
430	uint32_t	limit;
431	uint32_t	access;
432};
433#define	SEG_DESC_TYPE(access)		((access) & 0x001f)
434#define	SEG_DESC_DPL(access)		(((access) >> 5) & 0x3)
435#define	SEG_DESC_PRESENT(access)	(((access) & 0x0080) ? 1 : 0)
436#define	SEG_DESC_DEF32(access)		(((access) & 0x4000) ? 1 : 0)
437#define	SEG_DESC_GRANULARITY(access)	(((access) & 0x8000) ? 1 : 0)
438#define	SEG_DESC_UNUSABLE(access)	(((access) & 0x10000) ? 1 : 0)
439
440enum vm_cpu_mode {
441	CPU_MODE_REAL,
442	CPU_MODE_PROTECTED,
443	CPU_MODE_COMPATIBILITY,		/* IA-32E mode (CS.L = 0) */
444	CPU_MODE_64BIT,			/* IA-32E mode (CS.L = 1) */
445};
446
447enum vm_paging_mode {
448	PAGING_MODE_FLAT,
449	PAGING_MODE_32,
450	PAGING_MODE_PAE,
451	PAGING_MODE_64,
452};
453
454struct vm_guest_paging {
455	uint64_t	cr3;
456	int		cpl;
457	enum vm_cpu_mode cpu_mode;
458	enum vm_paging_mode paging_mode;
459};
460
461/*
462 * The data structures 'vie' and 'vie_op' are meant to be opaque to the
463 * consumers of instruction decoding. The only reason why their contents
464 * need to be exposed is because they are part of the 'vm_exit' structure.
465 */
466struct vie_op {
467	uint8_t		op_byte;	/* actual opcode byte */
468	uint8_t		op_type;	/* type of operation (e.g. MOV) */
469	uint16_t	op_flags;
470};
471
472#define	VIE_INST_SIZE	15
473struct vie {
474	uint8_t		inst[VIE_INST_SIZE];	/* instruction bytes */
475	uint8_t		num_valid;		/* size of the instruction */
476	uint8_t		num_processed;
477
478	uint8_t		addrsize:4, opsize:4;	/* address and operand sizes */
479	uint8_t		rex_w:1,		/* REX prefix */
480			rex_r:1,
481			rex_x:1,
482			rex_b:1,
483			rex_present:1,
484			repz_present:1,		/* REP/REPE/REPZ prefix */
485			repnz_present:1,	/* REPNE/REPNZ prefix */
486			opsize_override:1,	/* Operand size override */
487			addrsize_override:1,	/* Address size override */
488			segment_override:1;	/* Segment override */
489
490	uint8_t		mod:2,			/* ModRM byte */
491			reg:4,
492			rm:4;
493
494	uint8_t		ss:2,			/* SIB byte */
495			index:4,
496			base:4;
497
498	uint8_t		disp_bytes;
499	uint8_t		imm_bytes;
500
501	uint8_t		scale;
502	int		base_register;		/* VM_REG_GUEST_xyz */
503	int		index_register;		/* VM_REG_GUEST_xyz */
504	int		segment_register;	/* VM_REG_GUEST_xyz */
505
506	int64_t		displacement;		/* optional addr displacement */
507	int64_t		immediate;		/* optional immediate operand */
508
509	uint8_t		decoded;	/* set to 1 if successfully decoded */
510
511	struct vie_op	op;			/* opcode description */
512};
513
514enum vm_exitcode {
515	VM_EXITCODE_INOUT,
516	VM_EXITCODE_VMX,
517	VM_EXITCODE_BOGUS,
518	VM_EXITCODE_RDMSR,
519	VM_EXITCODE_WRMSR,
520	VM_EXITCODE_HLT,
521	VM_EXITCODE_MTRAP,
522	VM_EXITCODE_PAUSE,
523	VM_EXITCODE_PAGING,
524	VM_EXITCODE_INST_EMUL,
525	VM_EXITCODE_SPINUP_AP,
526	VM_EXITCODE_DEPRECATED1,	/* used to be SPINDOWN_CPU */
527	VM_EXITCODE_RENDEZVOUS,
528	VM_EXITCODE_IOAPIC_EOI,
529	VM_EXITCODE_SUSPENDED,
530	VM_EXITCODE_INOUT_STR,
531	VM_EXITCODE_TASK_SWITCH,
532	VM_EXITCODE_MONITOR,
533	VM_EXITCODE_MWAIT,
534	VM_EXITCODE_SVM,
535	VM_EXITCODE_REQIDLE,
536	VM_EXITCODE_MAX
537};
538
539struct vm_inout {
540	uint16_t	bytes:3;	/* 1 or 2 or 4 */
541	uint16_t	in:1;
542	uint16_t	string:1;
543	uint16_t	rep:1;
544	uint16_t	port;
545	uint32_t	eax;		/* valid for out */
546};
547
548struct vm_inout_str {
549	struct vm_inout	inout;		/* must be the first element */
550	struct vm_guest_paging paging;
551	uint64_t	rflags;
552	uint64_t	cr0;
553	uint64_t	index;
554	uint64_t	count;		/* rep=1 (%rcx), rep=0 (1) */
555	int		addrsize;
556	enum vm_reg_name seg_name;
557	struct seg_desc seg_desc;
558};
559
560enum task_switch_reason {
561	TSR_CALL,
562	TSR_IRET,
563	TSR_JMP,
564	TSR_IDT_GATE,	/* task gate in IDT */
565};
566
567struct vm_task_switch {
568	uint16_t	tsssel;		/* new TSS selector */
569	int		ext;		/* task switch due to external event */
570	uint32_t	errcode;
571	int		errcode_valid;	/* push 'errcode' on the new stack */
572	enum task_switch_reason reason;
573	struct vm_guest_paging paging;
574};
575
576struct vm_exit {
577	enum vm_exitcode	exitcode;
578	int			inst_length;	/* 0 means unknown */
579	uint64_t		rip;
580	union {
581		struct vm_inout	inout;
582		struct vm_inout_str inout_str;
583		struct {
584			uint64_t	gpa;
585			int		fault_type;
586		} paging;
587		struct {
588			uint64_t	gpa;
589			uint64_t	gla;
590			uint64_t	cs_base;
591			int		cs_d;		/* CS.D */
592			struct vm_guest_paging paging;
593			struct vie	vie;
594		} inst_emul;
595		/*
596		 * VMX specific payload. Used when there is no "better"
597		 * exitcode to represent the VM-exit.
598		 */
599		struct {
600			int		status;		/* vmx inst status */
601			/*
602			 * 'exit_reason' and 'exit_qualification' are valid
603			 * only if 'status' is zero.
604			 */
605			uint32_t	exit_reason;
606			uint64_t	exit_qualification;
607			/*
608			 * 'inst_error' and 'inst_type' are valid
609			 * only if 'status' is non-zero.
610			 */
611			int		inst_type;
612			int		inst_error;
613		} vmx;
614		/*
615		 * SVM specific payload.
616		 */
617		struct {
618			uint64_t	exitcode;
619			uint64_t	exitinfo1;
620			uint64_t	exitinfo2;
621		} svm;
622		struct {
623			uint32_t	code;		/* ecx value */
624			uint64_t	wval;
625		} msr;
626		struct {
627			int		vcpu;
628			uint64_t	rip;
629		} spinup_ap;
630		struct {
631			uint64_t	rflags;
632		} hlt;
633		struct {
634			int		vector;
635		} ioapic_eoi;
636		struct {
637			enum vm_suspend_how how;
638		} suspended;
639		struct vm_task_switch task_switch;
640	} u;
641};
642
643/* APIs to inject faults into the guest */
644void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid,
645    int errcode);
646
647static __inline void
648vm_inject_ud(void *vm, int vcpuid)
649{
650	vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0);
651}
652
653static __inline void
654vm_inject_gp(void *vm, int vcpuid)
655{
656	vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0);
657}
658
659static __inline void
660vm_inject_ac(void *vm, int vcpuid, int errcode)
661{
662	vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode);
663}
664
665static __inline void
666vm_inject_ss(void *vm, int vcpuid, int errcode)
667{
668	vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode);
669}
670
671void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2);
672
673int vm_restart_instruction(void *vm, int vcpuid);
674
675#endif	/* _VMM_H_ */
676