xref: /freebsd/sys/amd64/include/vmm.h (revision 5f0216bd883edee71bf81051e3c20505e4820903)
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  * $FreeBSD$
27  */
28 
29 #ifndef _VMM_H_
30 #define	_VMM_H_
31 
32 #include <x86/segments.h>
33 
34 enum 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  */
46 enum 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 
89 enum 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 
109 struct vm;
110 struct vm_exception;
111 struct vm_memory_segment;
112 struct seg_desc;
113 struct vm_exit;
114 struct vm_run;
115 struct vhpet;
116 struct vioapic;
117 struct vlapic;
118 struct vmspace;
119 struct vm_object;
120 struct vm_guest_paging;
121 struct pmap;
122 
123 struct vm_eventinfo {
124 	void	*rptr;		/* rendezvous cookie */
125 	int	*sptr;		/* suspend cookie */
126 	int	*iptr;		/* reqidle cookie */
127 };
128 
129 typedef int	(*vmm_init_func_t)(int ipinum);
130 typedef int	(*vmm_cleanup_func_t)(void);
131 typedef void	(*vmm_resume_func_t)(void);
132 typedef void *	(*vmi_init_func_t)(struct vm *vm, struct pmap *pmap);
133 typedef int	(*vmi_run_func_t)(void *vmi, int vcpu, register_t rip,
134 		    struct pmap *pmap, struct vm_eventinfo *info);
135 typedef void	(*vmi_cleanup_func_t)(void *vmi);
136 typedef int	(*vmi_get_register_t)(void *vmi, int vcpu, int num,
137 				      uint64_t *retval);
138 typedef int	(*vmi_set_register_t)(void *vmi, int vcpu, int num,
139 				      uint64_t val);
140 typedef int	(*vmi_get_desc_t)(void *vmi, int vcpu, int num,
141 				  struct seg_desc *desc);
142 typedef int	(*vmi_set_desc_t)(void *vmi, int vcpu, int num,
143 				  struct seg_desc *desc);
144 typedef int	(*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval);
145 typedef int	(*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val);
146 typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max);
147 typedef void	(*vmi_vmspace_free)(struct vmspace *vmspace);
148 typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu);
149 typedef void	(*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic);
150 
151 struct vmm_ops {
152 	vmm_init_func_t		init;		/* module wide initialization */
153 	vmm_cleanup_func_t	cleanup;
154 	vmm_resume_func_t	resume;
155 
156 	vmi_init_func_t		vminit;		/* vm-specific initialization */
157 	vmi_run_func_t		vmrun;
158 	vmi_cleanup_func_t	vmcleanup;
159 	vmi_get_register_t	vmgetreg;
160 	vmi_set_register_t	vmsetreg;
161 	vmi_get_desc_t		vmgetdesc;
162 	vmi_set_desc_t		vmsetdesc;
163 	vmi_get_cap_t		vmgetcap;
164 	vmi_set_cap_t		vmsetcap;
165 	vmi_vmspace_alloc	vmspace_alloc;
166 	vmi_vmspace_free	vmspace_free;
167 	vmi_vlapic_init		vlapic_init;
168 	vmi_vlapic_cleanup	vlapic_cleanup;
169 };
170 
171 extern struct vmm_ops vmm_ops_intel;
172 extern struct vmm_ops vmm_ops_amd;
173 
174 int vm_create(const char *name, struct vm **retvm);
175 void vm_destroy(struct vm *vm);
176 int vm_reinit(struct vm *vm);
177 const char *vm_name(struct vm *vm);
178 int vm_malloc(struct vm *vm, vm_paddr_t gpa, size_t len);
179 int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
180 int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
181 void *vm_gpa_hold(struct vm *, vm_paddr_t gpa, size_t len, int prot,
182 		  void **cookie);
183 void vm_gpa_release(void *cookie);
184 int vm_gpabase2memseg(struct vm *vm, vm_paddr_t gpabase,
185 	      struct vm_memory_segment *seg);
186 int vm_get_memobj(struct vm *vm, vm_paddr_t gpa, size_t len,
187 		  vm_offset_t *offset, struct vm_object **object);
188 boolean_t vm_mem_allocated(struct vm *vm, vm_paddr_t gpa);
189 int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval);
190 int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val);
191 int vm_get_seg_desc(struct vm *vm, int vcpu, int reg,
192 		    struct seg_desc *ret_desc);
193 int vm_set_seg_desc(struct vm *vm, int vcpu, int reg,
194 		    struct seg_desc *desc);
195 int vm_run(struct vm *vm, struct vm_run *vmrun);
196 int vm_suspend(struct vm *vm, enum vm_suspend_how how);
197 int vm_inject_nmi(struct vm *vm, int vcpu);
198 int vm_nmi_pending(struct vm *vm, int vcpuid);
199 void vm_nmi_clear(struct vm *vm, int vcpuid);
200 int vm_inject_extint(struct vm *vm, int vcpu);
201 int vm_extint_pending(struct vm *vm, int vcpuid);
202 void vm_extint_clear(struct vm *vm, int vcpuid);
203 struct vlapic *vm_lapic(struct vm *vm, int cpu);
204 struct vioapic *vm_ioapic(struct vm *vm);
205 struct vhpet *vm_hpet(struct vm *vm);
206 int vm_get_capability(struct vm *vm, int vcpu, int type, int *val);
207 int vm_set_capability(struct vm *vm, int vcpu, int type, int val);
208 int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state);
209 int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state);
210 int vm_apicid2vcpuid(struct vm *vm, int apicid);
211 int vm_activate_cpu(struct vm *vm, int vcpu);
212 struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid);
213 void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip);
214 void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip);
215 void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip);
216 void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip);
217 
218 #ifdef _SYS__CPUSET_H_
219 /*
220  * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'.
221  * The rendezvous 'func(arg)' is not allowed to do anything that will
222  * cause the thread to be put to sleep.
223  *
224  * If the rendezvous is being initiated from a vcpu context then the
225  * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1.
226  *
227  * The caller cannot hold any locks when initiating the rendezvous.
228  *
229  * The implementation of this API may cause vcpus other than those specified
230  * by 'dest' to be stalled. The caller should not rely on any vcpus making
231  * forward progress when the rendezvous is in progress.
232  */
233 typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg);
234 void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
235     vm_rendezvous_func_t func, void *arg);
236 cpuset_t vm_active_cpus(struct vm *vm);
237 cpuset_t vm_suspended_cpus(struct vm *vm);
238 #endif	/* _SYS__CPUSET_H_ */
239 
240 static __inline int
241 vcpu_rendezvous_pending(struct vm_eventinfo *info)
242 {
243 
244 	return (*((uintptr_t *)(info->rptr)) != 0);
245 }
246 
247 static __inline int
248 vcpu_suspended(struct vm_eventinfo *info)
249 {
250 
251 	return (*info->sptr);
252 }
253 
254 static __inline int
255 vcpu_reqidle(struct vm_eventinfo *info)
256 {
257 
258 	return (*info->iptr);
259 }
260 
261 /*
262  * Return 1 if device indicated by bus/slot/func is supposed to be a
263  * pci passthrough device.
264  *
265  * Return 0 otherwise.
266  */
267 int vmm_is_pptdev(int bus, int slot, int func);
268 
269 void *vm_iommu_domain(struct vm *vm);
270 
271 enum vcpu_state {
272 	VCPU_IDLE,
273 	VCPU_FROZEN,
274 	VCPU_RUNNING,
275 	VCPU_SLEEPING,
276 };
277 
278 int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
279     bool from_idle);
280 enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
281 
282 static int __inline
283 vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
284 {
285 	return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
286 }
287 
288 #ifdef _SYS_PROC_H_
289 static int __inline
290 vcpu_should_yield(struct vm *vm, int vcpu)
291 {
292 
293 	if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED))
294 		return (1);
295 	else if (curthread->td_owepreempt)
296 		return (1);
297 	else
298 		return (0);
299 }
300 #endif
301 
302 void *vcpu_stats(struct vm *vm, int vcpu);
303 void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
304 struct vmspace *vm_get_vmspace(struct vm *vm);
305 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
306 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
307 struct vatpic *vm_atpic(struct vm *vm);
308 struct vatpit *vm_atpit(struct vm *vm);
309 struct vpmtmr *vm_pmtmr(struct vm *vm);
310 struct vrtc *vm_rtc(struct vm *vm);
311 
312 /*
313  * Inject exception 'vector' into the guest vcpu. This function returns 0 on
314  * success and non-zero on failure.
315  *
316  * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
317  * this function directly because they enforce the trap-like or fault-like
318  * behavior of an exception.
319  *
320  * This function should only be called in the context of the thread that is
321  * executing this vcpu.
322  */
323 int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid,
324     uint32_t errcode, int restart_instruction);
325 
326 /*
327  * This function is called after a VM-exit that occurred during exception or
328  * interrupt delivery through the IDT. The format of 'intinfo' is described
329  * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
330  *
331  * If a VM-exit handler completes the event delivery successfully then it
332  * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
333  * if the task switch emulation is triggered via a task gate then it should
334  * call this function with 'intinfo=0' to indicate that the external event
335  * is not pending anymore.
336  *
337  * Return value is 0 on success and non-zero on failure.
338  */
339 int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo);
340 
341 /*
342  * This function is called before every VM-entry to retrieve a pending
343  * event that should be injected into the guest. This function combines
344  * nested events into a double or triple fault.
345  *
346  * Returns 0 if there are no events that need to be injected into the guest
347  * and non-zero otherwise.
348  */
349 int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info);
350 
351 int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2);
352 
353 enum vm_reg_name vm_segment_name(int seg_encoding);
354 
355 struct vm_copyinfo {
356 	uint64_t	gpa;
357 	size_t		len;
358 	void		*hva;
359 	void		*cookie;
360 };
361 
362 /*
363  * Set up 'copyinfo[]' to copy to/from guest linear address space starting
364  * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
365  * a copyin or PROT_WRITE for a copyout.
366  *
367  * retval	is_fault	Intepretation
368  *   0		   0		Success
369  *   0		   1		An exception was injected into the guest
370  * EFAULT	  N/A		Unrecoverable error
371  *
372  * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
373  * the return value is 0. The 'copyinfo[]' resources should be freed by calling
374  * 'vm_copy_teardown()' after the copy is done.
375  */
376 int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging,
377     uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
378     int num_copyinfo, int *is_fault);
379 void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
380     int num_copyinfo);
381 void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
382     void *kaddr, size_t len);
383 void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr,
384     struct vm_copyinfo *copyinfo, size_t len);
385 
386 int vcpu_trace_exceptions(struct vm *vm, int vcpuid);
387 #endif	/* KERNEL */
388 
389 #define	VM_MAXCPU	16			/* maximum virtual cpus */
390 
391 /*
392  * Identifiers for optional vmm capabilities
393  */
394 enum vm_cap_type {
395 	VM_CAP_HALT_EXIT,
396 	VM_CAP_MTRAP_EXIT,
397 	VM_CAP_PAUSE_EXIT,
398 	VM_CAP_UNRESTRICTED_GUEST,
399 	VM_CAP_ENABLE_INVPCID,
400 	VM_CAP_MAX
401 };
402 
403 enum vm_intr_trigger {
404 	EDGE_TRIGGER,
405 	LEVEL_TRIGGER
406 };
407 
408 /*
409  * The 'access' field has the format specified in Table 21-2 of the Intel
410  * Architecture Manual vol 3b.
411  *
412  * XXX The contents of the 'access' field are architecturally defined except
413  * bit 16 - Segment Unusable.
414  */
415 struct seg_desc {
416 	uint64_t	base;
417 	uint32_t	limit;
418 	uint32_t	access;
419 };
420 #define	SEG_DESC_TYPE(access)		((access) & 0x001f)
421 #define	SEG_DESC_DPL(access)		(((access) >> 5) & 0x3)
422 #define	SEG_DESC_PRESENT(access)	(((access) & 0x0080) ? 1 : 0)
423 #define	SEG_DESC_DEF32(access)		(((access) & 0x4000) ? 1 : 0)
424 #define	SEG_DESC_GRANULARITY(access)	(((access) & 0x8000) ? 1 : 0)
425 #define	SEG_DESC_UNUSABLE(access)	(((access) & 0x10000) ? 1 : 0)
426 
427 enum vm_cpu_mode {
428 	CPU_MODE_REAL,
429 	CPU_MODE_PROTECTED,
430 	CPU_MODE_COMPATIBILITY,		/* IA-32E mode (CS.L = 0) */
431 	CPU_MODE_64BIT,			/* IA-32E mode (CS.L = 1) */
432 };
433 
434 enum vm_paging_mode {
435 	PAGING_MODE_FLAT,
436 	PAGING_MODE_32,
437 	PAGING_MODE_PAE,
438 	PAGING_MODE_64,
439 };
440 
441 struct vm_guest_paging {
442 	uint64_t	cr3;
443 	int		cpl;
444 	enum vm_cpu_mode cpu_mode;
445 	enum vm_paging_mode paging_mode;
446 };
447 
448 /*
449  * The data structures 'vie' and 'vie_op' are meant to be opaque to the
450  * consumers of instruction decoding. The only reason why their contents
451  * need to be exposed is because they are part of the 'vm_exit' structure.
452  */
453 struct vie_op {
454 	uint8_t		op_byte;	/* actual opcode byte */
455 	uint8_t		op_type;	/* type of operation (e.g. MOV) */
456 	uint16_t	op_flags;
457 };
458 
459 #define	VIE_INST_SIZE	15
460 struct vie {
461 	uint8_t		inst[VIE_INST_SIZE];	/* instruction bytes */
462 	uint8_t		num_valid;		/* size of the instruction */
463 	uint8_t		num_processed;
464 
465 	uint8_t		addrsize:4, opsize:4;	/* address and operand sizes */
466 	uint8_t		rex_w:1,		/* REX prefix */
467 			rex_r:1,
468 			rex_x:1,
469 			rex_b:1,
470 			rex_present:1,
471 			repz_present:1,		/* REP/REPE/REPZ prefix */
472 			repnz_present:1,	/* REPNE/REPNZ prefix */
473 			opsize_override:1,	/* Operand size override */
474 			addrsize_override:1,	/* Address size override */
475 			segment_override:1;	/* Segment override */
476 
477 	uint8_t		mod:2,			/* ModRM byte */
478 			reg:4,
479 			rm:4;
480 
481 	uint8_t		ss:2,			/* SIB byte */
482 			index:4,
483 			base:4;
484 
485 	uint8_t		disp_bytes;
486 	uint8_t		imm_bytes;
487 
488 	uint8_t		scale;
489 	int		base_register;		/* VM_REG_GUEST_xyz */
490 	int		index_register;		/* VM_REG_GUEST_xyz */
491 	int		segment_register;	/* VM_REG_GUEST_xyz */
492 
493 	int64_t		displacement;		/* optional addr displacement */
494 	int64_t		immediate;		/* optional immediate operand */
495 
496 	uint8_t		decoded;	/* set to 1 if successfully decoded */
497 
498 	struct vie_op	op;			/* opcode description */
499 };
500 
501 enum vm_exitcode {
502 	VM_EXITCODE_INOUT,
503 	VM_EXITCODE_VMX,
504 	VM_EXITCODE_BOGUS,
505 	VM_EXITCODE_RDMSR,
506 	VM_EXITCODE_WRMSR,
507 	VM_EXITCODE_HLT,
508 	VM_EXITCODE_MTRAP,
509 	VM_EXITCODE_PAUSE,
510 	VM_EXITCODE_PAGING,
511 	VM_EXITCODE_INST_EMUL,
512 	VM_EXITCODE_SPINUP_AP,
513 	VM_EXITCODE_DEPRECATED1,	/* used to be SPINDOWN_CPU */
514 	VM_EXITCODE_RENDEZVOUS,
515 	VM_EXITCODE_IOAPIC_EOI,
516 	VM_EXITCODE_SUSPENDED,
517 	VM_EXITCODE_INOUT_STR,
518 	VM_EXITCODE_TASK_SWITCH,
519 	VM_EXITCODE_MONITOR,
520 	VM_EXITCODE_MWAIT,
521 	VM_EXITCODE_SVM,
522 	VM_EXITCODE_REQIDLE,
523 	VM_EXITCODE_MAX
524 };
525 
526 struct vm_inout {
527 	uint16_t	bytes:3;	/* 1 or 2 or 4 */
528 	uint16_t	in:1;
529 	uint16_t	string:1;
530 	uint16_t	rep:1;
531 	uint16_t	port;
532 	uint32_t	eax;		/* valid for out */
533 };
534 
535 struct vm_inout_str {
536 	struct vm_inout	inout;		/* must be the first element */
537 	struct vm_guest_paging paging;
538 	uint64_t	rflags;
539 	uint64_t	cr0;
540 	uint64_t	index;
541 	uint64_t	count;		/* rep=1 (%rcx), rep=0 (1) */
542 	int		addrsize;
543 	enum vm_reg_name seg_name;
544 	struct seg_desc seg_desc;
545 };
546 
547 enum task_switch_reason {
548 	TSR_CALL,
549 	TSR_IRET,
550 	TSR_JMP,
551 	TSR_IDT_GATE,	/* task gate in IDT */
552 };
553 
554 struct vm_task_switch {
555 	uint16_t	tsssel;		/* new TSS selector */
556 	int		ext;		/* task switch due to external event */
557 	uint32_t	errcode;
558 	int		errcode_valid;	/* push 'errcode' on the new stack */
559 	enum task_switch_reason reason;
560 	struct vm_guest_paging paging;
561 };
562 
563 struct vm_exit {
564 	enum vm_exitcode	exitcode;
565 	int			inst_length;	/* 0 means unknown */
566 	uint64_t		rip;
567 	union {
568 		struct vm_inout	inout;
569 		struct vm_inout_str inout_str;
570 		struct {
571 			uint64_t	gpa;
572 			int		fault_type;
573 		} paging;
574 		struct {
575 			uint64_t	gpa;
576 			uint64_t	gla;
577 			uint64_t	cs_base;
578 			int		cs_d;		/* CS.D */
579 			struct vm_guest_paging paging;
580 			struct vie	vie;
581 		} inst_emul;
582 		/*
583 		 * VMX specific payload. Used when there is no "better"
584 		 * exitcode to represent the VM-exit.
585 		 */
586 		struct {
587 			int		status;		/* vmx inst status */
588 			/*
589 			 * 'exit_reason' and 'exit_qualification' are valid
590 			 * only if 'status' is zero.
591 			 */
592 			uint32_t	exit_reason;
593 			uint64_t	exit_qualification;
594 			/*
595 			 * 'inst_error' and 'inst_type' are valid
596 			 * only if 'status' is non-zero.
597 			 */
598 			int		inst_type;
599 			int		inst_error;
600 		} vmx;
601 		/*
602 		 * SVM specific payload.
603 		 */
604 		struct {
605 			uint64_t	exitcode;
606 			uint64_t	exitinfo1;
607 			uint64_t	exitinfo2;
608 		} svm;
609 		struct {
610 			uint32_t	code;		/* ecx value */
611 			uint64_t	wval;
612 		} msr;
613 		struct {
614 			int		vcpu;
615 			uint64_t	rip;
616 		} spinup_ap;
617 		struct {
618 			uint64_t	rflags;
619 		} hlt;
620 		struct {
621 			int		vector;
622 		} ioapic_eoi;
623 		struct {
624 			enum vm_suspend_how how;
625 		} suspended;
626 		struct vm_task_switch task_switch;
627 	} u;
628 };
629 
630 /* APIs to inject faults into the guest */
631 void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid,
632     int errcode);
633 
634 static __inline void
635 vm_inject_ud(void *vm, int vcpuid)
636 {
637 	vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0);
638 }
639 
640 static __inline void
641 vm_inject_gp(void *vm, int vcpuid)
642 {
643 	vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0);
644 }
645 
646 static __inline void
647 vm_inject_ac(void *vm, int vcpuid, int errcode)
648 {
649 	vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode);
650 }
651 
652 static __inline void
653 vm_inject_ss(void *vm, int vcpuid, int errcode)
654 {
655 	vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode);
656 }
657 
658 void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2);
659 
660 int vm_restart_instruction(void *vm, int vcpuid);
661 
662 #endif	/* _VMM_H_ */
663