1 /*-
2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3 *
4 * Copyright (c) 2011 NetApp, Inc.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 * SUCH DAMAGE.
27 *
28 * $FreeBSD$
29 */
30
31 #ifndef _VMM_H_
32 #define _VMM_H_
33
34 #include <sys/sdt.h>
35 #include <x86/segments.h>
36
37 #ifdef _KERNEL
38 SDT_PROVIDER_DECLARE(vmm);
39 #endif
40
41 enum vm_suspend_how {
42 VM_SUSPEND_NONE,
43 VM_SUSPEND_RESET,
44 VM_SUSPEND_POWEROFF,
45 VM_SUSPEND_HALT,
46 VM_SUSPEND_TRIPLEFAULT,
47 VM_SUSPEND_LAST
48 };
49
50 /*
51 * Identifiers for architecturally defined registers.
52 */
53 enum vm_reg_name {
54 VM_REG_GUEST_RAX,
55 VM_REG_GUEST_RBX,
56 VM_REG_GUEST_RCX,
57 VM_REG_GUEST_RDX,
58 VM_REG_GUEST_RSI,
59 VM_REG_GUEST_RDI,
60 VM_REG_GUEST_RBP,
61 VM_REG_GUEST_R8,
62 VM_REG_GUEST_R9,
63 VM_REG_GUEST_R10,
64 VM_REG_GUEST_R11,
65 VM_REG_GUEST_R12,
66 VM_REG_GUEST_R13,
67 VM_REG_GUEST_R14,
68 VM_REG_GUEST_R15,
69 VM_REG_GUEST_CR0,
70 VM_REG_GUEST_CR3,
71 VM_REG_GUEST_CR4,
72 VM_REG_GUEST_DR7,
73 VM_REG_GUEST_RSP,
74 VM_REG_GUEST_RIP,
75 VM_REG_GUEST_RFLAGS,
76 VM_REG_GUEST_ES,
77 VM_REG_GUEST_CS,
78 VM_REG_GUEST_SS,
79 VM_REG_GUEST_DS,
80 VM_REG_GUEST_FS,
81 VM_REG_GUEST_GS,
82 VM_REG_GUEST_LDTR,
83 VM_REG_GUEST_TR,
84 VM_REG_GUEST_IDTR,
85 VM_REG_GUEST_GDTR,
86 VM_REG_GUEST_EFER,
87 VM_REG_GUEST_CR2,
88 VM_REG_GUEST_PDPTE0,
89 VM_REG_GUEST_PDPTE1,
90 VM_REG_GUEST_PDPTE2,
91 VM_REG_GUEST_PDPTE3,
92 VM_REG_GUEST_INTR_SHADOW,
93 VM_REG_GUEST_DR0,
94 VM_REG_GUEST_DR1,
95 VM_REG_GUEST_DR2,
96 VM_REG_GUEST_DR3,
97 VM_REG_GUEST_DR6,
98 VM_REG_LAST
99 };
100
101 enum x2apic_state {
102 X2APIC_DISABLED,
103 X2APIC_ENABLED,
104 X2APIC_STATE_LAST
105 };
106
107 #define VM_INTINFO_VECTOR(info) ((info) & 0xff)
108 #define VM_INTINFO_DEL_ERRCODE 0x800
109 #define VM_INTINFO_RSVD 0x7ffff000
110 #define VM_INTINFO_VALID 0x80000000
111 #define VM_INTINFO_TYPE 0x700
112 #define VM_INTINFO_HWINTR (0 << 8)
113 #define VM_INTINFO_NMI (2 << 8)
114 #define VM_INTINFO_HWEXCEPTION (3 << 8)
115 #define VM_INTINFO_SWINTR (4 << 8)
116
117 #ifdef _KERNEL
118
119 #define VM_MAX_NAMELEN 32
120
121 struct vm;
122 struct vm_exception;
123 struct seg_desc;
124 struct vm_exit;
125 struct vm_run;
126 struct vhpet;
127 struct vioapic;
128 struct vlapic;
129 struct vmspace;
130 struct vm_object;
131 struct vm_guest_paging;
132 struct pmap;
133
134 struct vm_eventinfo {
135 void *rptr; /* rendezvous cookie */
136 int *sptr; /* suspend cookie */
137 int *iptr; /* reqidle cookie */
138 };
139
140 typedef int (*vmm_init_func_t)(int ipinum);
141 typedef int (*vmm_cleanup_func_t)(void);
142 typedef void (*vmm_resume_func_t)(void);
143 typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap);
144 typedef int (*vmi_run_func_t)(void *vmi, int vcpu, register_t rip,
145 struct pmap *pmap, struct vm_eventinfo *info);
146 typedef void (*vmi_cleanup_func_t)(void *vmi);
147 typedef int (*vmi_get_register_t)(void *vmi, int vcpu, int num,
148 uint64_t *retval);
149 typedef int (*vmi_set_register_t)(void *vmi, int vcpu, int num,
150 uint64_t val);
151 typedef int (*vmi_get_desc_t)(void *vmi, int vcpu, int num,
152 struct seg_desc *desc);
153 typedef int (*vmi_set_desc_t)(void *vmi, int vcpu, int num,
154 struct seg_desc *desc);
155 typedef int (*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval);
156 typedef int (*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val);
157 typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max);
158 typedef void (*vmi_vmspace_free)(struct vmspace *vmspace);
159 typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu);
160 typedef void (*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic);
161
162 struct vmm_ops {
163 vmm_init_func_t init; /* module wide initialization */
164 vmm_cleanup_func_t cleanup;
165 vmm_resume_func_t resume;
166
167 vmi_init_func_t vminit; /* vm-specific initialization */
168 vmi_run_func_t vmrun;
169 vmi_cleanup_func_t vmcleanup;
170 vmi_get_register_t vmgetreg;
171 vmi_set_register_t vmsetreg;
172 vmi_get_desc_t vmgetdesc;
173 vmi_set_desc_t vmsetdesc;
174 vmi_get_cap_t vmgetcap;
175 vmi_set_cap_t vmsetcap;
176 vmi_vmspace_alloc vmspace_alloc;
177 vmi_vmspace_free vmspace_free;
178 vmi_vlapic_init vlapic_init;
179 vmi_vlapic_cleanup vlapic_cleanup;
180 };
181
182 extern struct vmm_ops vmm_ops_intel;
183 extern struct vmm_ops vmm_ops_amd;
184
185 int vm_create(const char *name, struct vm **retvm);
186 void vm_destroy(struct vm *vm);
187 int vm_reinit(struct vm *vm);
188 const char *vm_name(struct vm *vm);
189 uint16_t vm_get_maxcpus(struct vm *vm);
190 void vm_get_topology(struct vm *vm, uint16_t *sockets, uint16_t *cores,
191 uint16_t *threads, uint16_t *maxcpus);
192 int vm_set_topology(struct vm *vm, uint16_t sockets, uint16_t cores,
193 uint16_t threads, uint16_t maxcpus);
194
195 /*
196 * APIs that modify the guest memory map require all vcpus to be frozen.
197 */
198 int vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t off,
199 size_t len, int prot, int flags);
200 int vm_alloc_memseg(struct vm *vm, int ident, size_t len, bool sysmem);
201 void vm_free_memseg(struct vm *vm, int ident);
202 int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
203 int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
204 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
205 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
206
207 /*
208 * APIs that inspect the guest memory map require only a *single* vcpu to
209 * be frozen. This acts like a read lock on the guest memory map since any
210 * modification requires *all* vcpus to be frozen.
211 */
212 int vm_mmap_getnext(struct vm *vm, vm_paddr_t *gpa, int *segid,
213 vm_ooffset_t *segoff, size_t *len, int *prot, int *flags);
214 int vm_get_memseg(struct vm *vm, int ident, size_t *len, bool *sysmem,
215 struct vm_object **objptr);
216 vm_paddr_t vmm_sysmem_maxaddr(struct vm *vm);
217 void *vm_gpa_hold(struct vm *, int vcpuid, vm_paddr_t gpa, size_t len,
218 int prot, void **cookie);
219 void vm_gpa_release(void *cookie);
220 bool vm_mem_allocated(struct vm *vm, int vcpuid, vm_paddr_t gpa);
221
222 int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval);
223 int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val);
224 int vm_get_seg_desc(struct vm *vm, int vcpu, int reg,
225 struct seg_desc *ret_desc);
226 int vm_set_seg_desc(struct vm *vm, int vcpu, int reg,
227 struct seg_desc *desc);
228 int vm_run(struct vm *vm, struct vm_run *vmrun);
229 int vm_suspend(struct vm *vm, enum vm_suspend_how how);
230 int vm_inject_nmi(struct vm *vm, int vcpu);
231 int vm_nmi_pending(struct vm *vm, int vcpuid);
232 void vm_nmi_clear(struct vm *vm, int vcpuid);
233 int vm_inject_extint(struct vm *vm, int vcpu);
234 int vm_extint_pending(struct vm *vm, int vcpuid);
235 void vm_extint_clear(struct vm *vm, int vcpuid);
236 struct vlapic *vm_lapic(struct vm *vm, int cpu);
237 struct vioapic *vm_ioapic(struct vm *vm);
238 struct vhpet *vm_hpet(struct vm *vm);
239 int vm_get_capability(struct vm *vm, int vcpu, int type, int *val);
240 int vm_set_capability(struct vm *vm, int vcpu, int type, int val);
241 int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state);
242 int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state);
243 int vm_apicid2vcpuid(struct vm *vm, int apicid);
244 int vm_activate_cpu(struct vm *vm, int vcpu);
245 int vm_suspend_cpu(struct vm *vm, int vcpu);
246 int vm_resume_cpu(struct vm *vm, int vcpu);
247 struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid);
248 void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip);
249 void vm_exit_debug(struct vm *vm, int vcpuid, uint64_t rip);
250 void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip);
251 void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip);
252 void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip);
253
254 #ifdef _SYS__CPUSET_H_
255 /*
256 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'.
257 * The rendezvous 'func(arg)' is not allowed to do anything that will
258 * cause the thread to be put to sleep.
259 *
260 * If the rendezvous is being initiated from a vcpu context then the
261 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1.
262 *
263 * The caller cannot hold any locks when initiating the rendezvous.
264 *
265 * The implementation of this API may cause vcpus other than those specified
266 * by 'dest' to be stalled. The caller should not rely on any vcpus making
267 * forward progress when the rendezvous is in progress.
268 */
269 typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg);
270 void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest,
271 vm_rendezvous_func_t func, void *arg);
272 cpuset_t vm_active_cpus(struct vm *vm);
273 cpuset_t vm_debug_cpus(struct vm *vm);
274 cpuset_t vm_suspended_cpus(struct vm *vm);
275 #endif /* _SYS__CPUSET_H_ */
276
277 static __inline int
vcpu_rendezvous_pending(struct vm_eventinfo * info)278 vcpu_rendezvous_pending(struct vm_eventinfo *info)
279 {
280
281 return (*((uintptr_t *)(info->rptr)) != 0);
282 }
283
284 static __inline int
vcpu_suspended(struct vm_eventinfo * info)285 vcpu_suspended(struct vm_eventinfo *info)
286 {
287
288 return (*info->sptr);
289 }
290
291 static __inline int
vcpu_reqidle(struct vm_eventinfo * info)292 vcpu_reqidle(struct vm_eventinfo *info)
293 {
294
295 return (*info->iptr);
296 }
297
298 int vcpu_debugged(struct vm *vm, int vcpuid);
299
300 /*
301 * Return true if device indicated by bus/slot/func is supposed to be a
302 * pci passthrough device.
303 *
304 * Return false otherwise.
305 */
306 bool vmm_is_pptdev(int bus, int slot, int func);
307
308 void *vm_iommu_domain(struct vm *vm);
309
310 enum vcpu_state {
311 VCPU_IDLE,
312 VCPU_FROZEN,
313 VCPU_RUNNING,
314 VCPU_SLEEPING,
315 };
316
317 int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state,
318 bool from_idle);
319 enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu);
320
321 static int __inline
vcpu_is_running(struct vm * vm,int vcpu,int * hostcpu)322 vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu)
323 {
324 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING);
325 }
326
327 #ifdef _SYS_PROC_H_
328 static int __inline
vcpu_should_yield(struct vm * vm,int vcpu)329 vcpu_should_yield(struct vm *vm, int vcpu)
330 {
331
332 if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED))
333 return (1);
334 else if (curthread->td_owepreempt)
335 return (1);
336 else
337 return (0);
338 }
339 #endif
340
341 void *vcpu_stats(struct vm *vm, int vcpu);
342 void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr);
343 struct vmspace *vm_get_vmspace(struct vm *vm);
344 struct vatpic *vm_atpic(struct vm *vm);
345 struct vatpit *vm_atpit(struct vm *vm);
346 struct vpmtmr *vm_pmtmr(struct vm *vm);
347 struct vrtc *vm_rtc(struct vm *vm);
348
349 /*
350 * Inject exception 'vector' into the guest vcpu. This function returns 0 on
351 * success and non-zero on failure.
352 *
353 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
354 * this function directly because they enforce the trap-like or fault-like
355 * behavior of an exception.
356 *
357 * This function should only be called in the context of the thread that is
358 * executing this vcpu.
359 */
360 int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid,
361 uint32_t errcode, int restart_instruction);
362
363 /*
364 * This function is called after a VM-exit that occurred during exception or
365 * interrupt delivery through the IDT. The format of 'intinfo' is described
366 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
367 *
368 * If a VM-exit handler completes the event delivery successfully then it
369 * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
370 * if the task switch emulation is triggered via a task gate then it should
371 * call this function with 'intinfo=0' to indicate that the external event
372 * is not pending anymore.
373 *
374 * Return value is 0 on success and non-zero on failure.
375 */
376 int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo);
377
378 /*
379 * This function is called before every VM-entry to retrieve a pending
380 * event that should be injected into the guest. This function combines
381 * nested events into a double or triple fault.
382 *
383 * Returns 0 if there are no events that need to be injected into the guest
384 * and non-zero otherwise.
385 */
386 int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info);
387
388 int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2);
389
390 enum vm_reg_name vm_segment_name(int seg_encoding);
391
392 struct vm_copyinfo {
393 uint64_t gpa;
394 size_t len;
395 void *hva;
396 void *cookie;
397 };
398
399 /*
400 * Set up 'copyinfo[]' to copy to/from guest linear address space starting
401 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
402 * a copyin or PROT_WRITE for a copyout.
403 *
404 * retval is_fault Interpretation
405 * 0 0 Success
406 * 0 1 An exception was injected into the guest
407 * EFAULT N/A Unrecoverable error
408 *
409 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
410 * the return value is 0. The 'copyinfo[]' resources should be freed by calling
411 * 'vm_copy_teardown()' after the copy is done.
412 */
413 int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging,
414 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
415 int num_copyinfo, int *is_fault);
416 void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
417 int num_copyinfo);
418 void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo,
419 void *kaddr, size_t len);
420 void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr,
421 struct vm_copyinfo *copyinfo, size_t len);
422
423 int vcpu_trace_exceptions(struct vm *vm, int vcpuid);
424 #endif /* KERNEL */
425
426 #define VM_MAXCPU 16 /* maximum virtual cpus */
427
428 /*
429 * Identifiers for optional vmm capabilities
430 */
431 enum vm_cap_type {
432 VM_CAP_HALT_EXIT,
433 VM_CAP_MTRAP_EXIT,
434 VM_CAP_PAUSE_EXIT,
435 VM_CAP_UNRESTRICTED_GUEST,
436 VM_CAP_ENABLE_INVPCID,
437 VM_CAP_MAX
438 };
439
440 enum vm_intr_trigger {
441 EDGE_TRIGGER,
442 LEVEL_TRIGGER
443 };
444
445 /*
446 * The 'access' field has the format specified in Table 21-2 of the Intel
447 * Architecture Manual vol 3b.
448 *
449 * XXX The contents of the 'access' field are architecturally defined except
450 * bit 16 - Segment Unusable.
451 */
452 struct seg_desc {
453 uint64_t base;
454 uint32_t limit;
455 uint32_t access;
456 };
457 #define SEG_DESC_TYPE(access) ((access) & 0x001f)
458 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3)
459 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0)
460 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0)
461 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0)
462 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0)
463
464 enum vm_cpu_mode {
465 CPU_MODE_REAL,
466 CPU_MODE_PROTECTED,
467 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */
468 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */
469 };
470
471 enum vm_paging_mode {
472 PAGING_MODE_FLAT,
473 PAGING_MODE_32,
474 PAGING_MODE_PAE,
475 PAGING_MODE_64,
476 };
477
478 struct vm_guest_paging {
479 uint64_t cr3;
480 int cpl;
481 enum vm_cpu_mode cpu_mode;
482 enum vm_paging_mode paging_mode;
483 };
484
485 /*
486 * The data structures 'vie' and 'vie_op' are meant to be opaque to the
487 * consumers of instruction decoding. The only reason why their contents
488 * need to be exposed is because they are part of the 'vm_exit' structure.
489 */
490 struct vie_op {
491 uint8_t op_byte; /* actual opcode byte */
492 uint8_t op_type; /* type of operation (e.g. MOV) */
493 uint16_t op_flags;
494 };
495
496 #define VIE_INST_SIZE 15
497 struct vie {
498 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */
499 uint8_t num_valid; /* size of the instruction */
500 uint8_t num_processed;
501
502 uint8_t addrsize:4, opsize:4; /* address and operand sizes */
503 uint8_t rex_w:1, /* REX prefix */
504 rex_r:1,
505 rex_x:1,
506 rex_b:1,
507 rex_present:1,
508 repz_present:1, /* REP/REPE/REPZ prefix */
509 repnz_present:1, /* REPNE/REPNZ prefix */
510 opsize_override:1, /* Operand size override */
511 addrsize_override:1, /* Address size override */
512 segment_override:1; /* Segment override */
513
514 uint8_t mod:2, /* ModRM byte */
515 reg:4,
516 rm:4;
517
518 uint8_t ss:2, /* SIB byte */
519 index:4,
520 base:4;
521
522 uint8_t disp_bytes;
523 uint8_t imm_bytes;
524
525 uint8_t scale;
526 int base_register; /* VM_REG_GUEST_xyz */
527 int index_register; /* VM_REG_GUEST_xyz */
528 int segment_register; /* VM_REG_GUEST_xyz */
529
530 int64_t displacement; /* optional addr displacement */
531 int64_t immediate; /* optional immediate operand */
532
533 uint8_t decoded; /* set to 1 if successfully decoded */
534
535 struct vie_op op; /* opcode description */
536 };
537
538 enum vm_exitcode {
539 VM_EXITCODE_INOUT,
540 VM_EXITCODE_VMX,
541 VM_EXITCODE_BOGUS,
542 VM_EXITCODE_RDMSR,
543 VM_EXITCODE_WRMSR,
544 VM_EXITCODE_HLT,
545 VM_EXITCODE_MTRAP,
546 VM_EXITCODE_PAUSE,
547 VM_EXITCODE_PAGING,
548 VM_EXITCODE_INST_EMUL,
549 VM_EXITCODE_SPINUP_AP,
550 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */
551 VM_EXITCODE_RENDEZVOUS,
552 VM_EXITCODE_IOAPIC_EOI,
553 VM_EXITCODE_SUSPENDED,
554 VM_EXITCODE_INOUT_STR,
555 VM_EXITCODE_TASK_SWITCH,
556 VM_EXITCODE_MONITOR,
557 VM_EXITCODE_MWAIT,
558 VM_EXITCODE_SVM,
559 VM_EXITCODE_REQIDLE,
560 VM_EXITCODE_DEBUG,
561 VM_EXITCODE_VMINSN,
562 VM_EXITCODE_MAX
563 };
564
565 struct vm_inout {
566 uint16_t bytes:3; /* 1 or 2 or 4 */
567 uint16_t in:1;
568 uint16_t string:1;
569 uint16_t rep:1;
570 uint16_t port;
571 uint32_t eax; /* valid for out */
572 };
573
574 struct vm_inout_str {
575 struct vm_inout inout; /* must be the first element */
576 struct vm_guest_paging paging;
577 uint64_t rflags;
578 uint64_t cr0;
579 uint64_t index;
580 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */
581 int addrsize;
582 enum vm_reg_name seg_name;
583 struct seg_desc seg_desc;
584 };
585
586 enum task_switch_reason {
587 TSR_CALL,
588 TSR_IRET,
589 TSR_JMP,
590 TSR_IDT_GATE, /* task gate in IDT */
591 };
592
593 struct vm_task_switch {
594 uint16_t tsssel; /* new TSS selector */
595 int ext; /* task switch due to external event */
596 uint32_t errcode;
597 int errcode_valid; /* push 'errcode' on the new stack */
598 enum task_switch_reason reason;
599 struct vm_guest_paging paging;
600 };
601
602 struct vm_exit {
603 enum vm_exitcode exitcode;
604 int inst_length; /* 0 means unknown */
605 uint64_t rip;
606 union {
607 struct vm_inout inout;
608 struct vm_inout_str inout_str;
609 struct {
610 uint64_t gpa;
611 int fault_type;
612 } paging;
613 struct {
614 uint64_t gpa;
615 uint64_t gla;
616 uint64_t cs_base;
617 int cs_d; /* CS.D */
618 struct vm_guest_paging paging;
619 struct vie vie;
620 } inst_emul;
621 /*
622 * VMX specific payload. Used when there is no "better"
623 * exitcode to represent the VM-exit.
624 */
625 struct {
626 int status; /* vmx inst status */
627 /*
628 * 'exit_reason' and 'exit_qualification' are valid
629 * only if 'status' is zero.
630 */
631 uint32_t exit_reason;
632 uint64_t exit_qualification;
633 /*
634 * 'inst_error' and 'inst_type' are valid
635 * only if 'status' is non-zero.
636 */
637 int inst_type;
638 int inst_error;
639 } vmx;
640 /*
641 * SVM specific payload.
642 */
643 struct {
644 uint64_t exitcode;
645 uint64_t exitinfo1;
646 uint64_t exitinfo2;
647 } svm;
648 struct {
649 uint32_t code; /* ecx value */
650 uint64_t wval;
651 } msr;
652 struct {
653 int vcpu;
654 uint64_t rip;
655 } spinup_ap;
656 struct {
657 uint64_t rflags;
658 uint64_t intr_status;
659 } hlt;
660 struct {
661 int vector;
662 } ioapic_eoi;
663 struct {
664 enum vm_suspend_how how;
665 } suspended;
666 struct vm_task_switch task_switch;
667 } u;
668 };
669
670 /* APIs to inject faults into the guest */
671 void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid,
672 int errcode);
673
674 static __inline void
vm_inject_ud(void * vm,int vcpuid)675 vm_inject_ud(void *vm, int vcpuid)
676 {
677 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0);
678 }
679
680 static __inline void
vm_inject_gp(void * vm,int vcpuid)681 vm_inject_gp(void *vm, int vcpuid)
682 {
683 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0);
684 }
685
686 static __inline void
vm_inject_ac(void * vm,int vcpuid,int errcode)687 vm_inject_ac(void *vm, int vcpuid, int errcode)
688 {
689 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode);
690 }
691
692 static __inline void
vm_inject_ss(void * vm,int vcpuid,int errcode)693 vm_inject_ss(void *vm, int vcpuid, int errcode)
694 {
695 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode);
696 }
697
698 void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2);
699
700 int vm_restart_instruction(void *vm, int vcpuid);
701
702 #endif /* _VMM_H_ */
703