1 /*-
2 * SPDX-License-Identifier: BSD-2-Clause
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
29 #ifndef _VMM_H_
30 #define _VMM_H_
31
32 #include <sys/cpuset.h>
33 #include <sys/sdt.h>
34 #include <x86/segments.h>
35
36 struct vcpu;
37 struct vm_snapshot_meta;
38
39 #ifdef _KERNEL
40 SDT_PROVIDER_DECLARE(vmm);
41 #endif
42
43 enum vm_suspend_how {
44 VM_SUSPEND_NONE,
45 VM_SUSPEND_RESET,
46 VM_SUSPEND_POWEROFF,
47 VM_SUSPEND_HALT,
48 VM_SUSPEND_TRIPLEFAULT,
49 VM_SUSPEND_DESTROY,
50 VM_SUSPEND_LAST
51 };
52
53 /*
54 * Identifiers for architecturally defined registers.
55 */
56 enum vm_reg_name {
57 VM_REG_GUEST_RAX,
58 VM_REG_GUEST_RBX,
59 VM_REG_GUEST_RCX,
60 VM_REG_GUEST_RDX,
61 VM_REG_GUEST_RSI,
62 VM_REG_GUEST_RDI,
63 VM_REG_GUEST_RBP,
64 VM_REG_GUEST_R8,
65 VM_REG_GUEST_R9,
66 VM_REG_GUEST_R10,
67 VM_REG_GUEST_R11,
68 VM_REG_GUEST_R12,
69 VM_REG_GUEST_R13,
70 VM_REG_GUEST_R14,
71 VM_REG_GUEST_R15,
72 VM_REG_GUEST_CR0,
73 VM_REG_GUEST_CR3,
74 VM_REG_GUEST_CR4,
75 VM_REG_GUEST_DR7,
76 VM_REG_GUEST_RSP,
77 VM_REG_GUEST_RIP,
78 VM_REG_GUEST_RFLAGS,
79 VM_REG_GUEST_ES,
80 VM_REG_GUEST_CS,
81 VM_REG_GUEST_SS,
82 VM_REG_GUEST_DS,
83 VM_REG_GUEST_FS,
84 VM_REG_GUEST_GS,
85 VM_REG_GUEST_LDTR,
86 VM_REG_GUEST_TR,
87 VM_REG_GUEST_IDTR,
88 VM_REG_GUEST_GDTR,
89 VM_REG_GUEST_EFER,
90 VM_REG_GUEST_CR2,
91 VM_REG_GUEST_PDPTE0,
92 VM_REG_GUEST_PDPTE1,
93 VM_REG_GUEST_PDPTE2,
94 VM_REG_GUEST_PDPTE3,
95 VM_REG_GUEST_INTR_SHADOW,
96 VM_REG_GUEST_DR0,
97 VM_REG_GUEST_DR1,
98 VM_REG_GUEST_DR2,
99 VM_REG_GUEST_DR3,
100 VM_REG_GUEST_DR6,
101 VM_REG_GUEST_ENTRY_INST_LENGTH,
102 VM_REG_GUEST_FS_BASE,
103 VM_REG_GUEST_GS_BASE,
104 VM_REG_GUEST_KGS_BASE,
105 VM_REG_GUEST_TPR,
106 VM_REG_LAST
107 };
108
109 enum x2apic_state {
110 X2APIC_DISABLED,
111 X2APIC_ENABLED,
112 X2APIC_STATE_LAST
113 };
114
115 #define VM_INTINFO_VECTOR(info) ((info) & 0xff)
116 #define VM_INTINFO_DEL_ERRCODE 0x800
117 #define VM_INTINFO_RSVD 0x7ffff000
118 #define VM_INTINFO_VALID 0x80000000
119 #define VM_INTINFO_TYPE 0x700
120 #define VM_INTINFO_HWINTR (0 << 8)
121 #define VM_INTINFO_NMI (2 << 8)
122 #define VM_INTINFO_HWEXCEPTION (3 << 8)
123 #define VM_INTINFO_SWINTR (4 << 8)
124
125 #ifdef _KERNEL
126 #define VMM_VCPU_MD_FIELDS \
127 struct vlapic *vlapic; /* (i) APIC device model */ \
128 enum x2apic_state x2apic_state; /* (i) APIC mode */ \
129 uint64_t exitintinfo; /* (i) events pending at VM exit */ \
130 int nmi_pending; /* (i) NMI pending */ \
131 int extint_pending; /* (i) INTR pending */ \
132 int exception_pending; /* (i) exception pending */ \
133 int exc_vector; /* (x) exception collateral */ \
134 int exc_errcode_valid; \
135 uint32_t exc_errcode; \
136 struct savefpu *guestfpu; /* (a,i) guest fpu state */ \
137 uint64_t guest_xcr0; /* (i) guest %xcr0 register */ \
138 struct vm_exit exitinfo; /* (x) exit reason and collateral */ \
139 cpuset_t exitinfo_cpuset; /* (x) storage for vmexit handlers */ \
140 uint64_t nextrip; /* (x) next instruction to execute */ \
141 uint64_t tsc_offset /* (o) TSC offsetting */
142
143 #define VMM_VM_MD_FIELDS \
144 cpuset_t startup_cpus; /* (i) [r] waiting for startup */ \
145 void *iommu; /* (x) iommu-specific data */ \
146 struct vioapic *vioapic; /* (i) virtual ioapic */ \
147 struct vatpic *vatpic; /* (i) virtual atpic */ \
148 struct vatpit *vatpit; /* (i) virtual atpit */ \
149 struct vpmtmr *vpmtmr; /* (i) virtual ACPI PM timer */ \
150 struct vrtc *vrtc; /* (o) virtual RTC */ \
151 struct vhpet *vhpet /* (i) virtual HPET */
152
153 struct vm;
154 struct vm_exception;
155 struct vm_mem;
156 struct seg_desc;
157 struct vm_exit;
158 struct vm_run;
159 struct vhpet;
160 struct vioapic;
161 struct vlapic;
162 struct vmspace;
163 struct vm_eventinfo;
164 struct vm_object;
165 struct vm_guest_paging;
166 struct pmap;
167 enum snapshot_req;
168
169 #define DECLARE_VMMOPS_FUNC(ret_type, opname, args) \
170 typedef ret_type (*vmmops_##opname##_t) args; \
171 ret_type vmmops_##opname args
172
173 DECLARE_VMMOPS_FUNC(int, modinit, (int ipinum));
174 DECLARE_VMMOPS_FUNC(int, modcleanup, (void));
175 DECLARE_VMMOPS_FUNC(void, modresume, (void));
176 DECLARE_VMMOPS_FUNC(void, modsuspend, (void));
177 DECLARE_VMMOPS_FUNC(void *, init, (struct vm *vm, struct pmap *pmap));
178 DECLARE_VMMOPS_FUNC(int, run, (void *vcpui, register_t pc,
179 struct pmap *pmap, struct vm_eventinfo *info));
180 DECLARE_VMMOPS_FUNC(void, cleanup, (void *vmi));
181 DECLARE_VMMOPS_FUNC(void *, vcpu_init, (void *vmi, struct vcpu *vcpu,
182 int vcpu_id));
183 DECLARE_VMMOPS_FUNC(void, vcpu_cleanup, (void *vcpui));
184 DECLARE_VMMOPS_FUNC(int, getreg, (void *vcpui, int num, uint64_t *retval));
185 DECLARE_VMMOPS_FUNC(int, setreg, (void *vcpui, int num, uint64_t val));
186 DECLARE_VMMOPS_FUNC(int, getdesc, (void *vcpui, int num,
187 struct seg_desc *desc));
188 DECLARE_VMMOPS_FUNC(int, setdesc, (void *vcpui, int num,
189 struct seg_desc *desc));
190 DECLARE_VMMOPS_FUNC(int, getcap, (void *vcpui, int num, int *retval));
191 DECLARE_VMMOPS_FUNC(int, setcap, (void *vcpui, int num, int val));
192 DECLARE_VMMOPS_FUNC(struct vmspace *, vmspace_alloc,
193 (vm_offset_t min, vm_offset_t max));
194 DECLARE_VMMOPS_FUNC(void, vmspace_free, (struct vmspace *vmspace));
195 DECLARE_VMMOPS_FUNC(struct vlapic *, vlapic_init, (void *vcpui));
196 DECLARE_VMMOPS_FUNC(void, vlapic_cleanup, (struct vlapic *vlapic));
197 DECLARE_VMMOPS_FUNC(int, vcpu_snapshot, (void *vcpui,
198 struct vm_snapshot_meta *meta));
199 DECLARE_VMMOPS_FUNC(int, restore_tsc, (void *vcpui, uint64_t now));
200
201 struct vmm_ops {
202 vmmops_modinit_t modinit; /* module wide initialization */
203 vmmops_modcleanup_t modcleanup;
204 vmmops_modresume_t modsuspend;
205 vmmops_modresume_t modresume;
206
207 vmmops_init_t init; /* vm-specific initialization */
208 vmmops_run_t run;
209 vmmops_cleanup_t cleanup;
210 vmmops_vcpu_init_t vcpu_init;
211 vmmops_vcpu_cleanup_t vcpu_cleanup;
212 vmmops_getreg_t getreg;
213 vmmops_setreg_t setreg;
214 vmmops_getdesc_t getdesc;
215 vmmops_setdesc_t setdesc;
216 vmmops_getcap_t getcap;
217 vmmops_setcap_t setcap;
218 vmmops_vmspace_alloc_t vmspace_alloc;
219 vmmops_vmspace_free_t vmspace_free;
220 vmmops_vlapic_init_t vlapic_init;
221 vmmops_vlapic_cleanup_t vlapic_cleanup;
222
223 /* checkpoint operations */
224 vmmops_vcpu_snapshot_t vcpu_snapshot;
225 vmmops_restore_tsc_t restore_tsc;
226 };
227
228 extern const struct vmm_ops vmm_ops_intel;
229 extern const struct vmm_ops vmm_ops_amd;
230
231 int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa);
232 int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len);
233 int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func);
234 int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func);
235
236 int vm_get_register(struct vcpu *vcpu, int reg, uint64_t *retval);
237 int vm_set_register(struct vcpu *vcpu, int reg, uint64_t val);
238 int vm_get_seg_desc(struct vcpu *vcpu, int reg,
239 struct seg_desc *ret_desc);
240 int vm_set_seg_desc(struct vcpu *vcpu, int reg,
241 struct seg_desc *desc);
242 int vm_run(struct vcpu *vcpu);
243 int vm_inject_nmi(struct vcpu *vcpu);
244 int vm_nmi_pending(struct vcpu *vcpu);
245 void vm_nmi_clear(struct vcpu *vcpu);
246 int vm_inject_extint(struct vcpu *vcpu);
247 int vm_extint_pending(struct vcpu *vcpu);
248 void vm_extint_clear(struct vcpu *vcpu);
249 struct vlapic *vm_lapic(struct vcpu *vcpu);
250 struct vioapic *vm_ioapic(struct vm *vm);
251 struct vhpet *vm_hpet(struct vm *vm);
252 int vm_get_capability(struct vcpu *vcpu, int type, int *val);
253 int vm_set_capability(struct vcpu *vcpu, int type, int val);
254 int vm_get_x2apic_state(struct vcpu *vcpu, enum x2apic_state *state);
255 int vm_set_x2apic_state(struct vcpu *vcpu, enum x2apic_state state);
256 int vm_apicid2vcpuid(struct vm *vm, int apicid);
257 int vm_restart_instruction(struct vcpu *vcpu);
258 struct vm_exit *vm_exitinfo(struct vcpu *vcpu);
259 cpuset_t *vm_exitinfo_cpuset(struct vcpu *vcpu);
260 void vm_exit_suspended(struct vcpu *vcpu, uint64_t rip);
261 void vm_exit_debug(struct vcpu *vcpu, uint64_t rip);
262 void vm_exit_rendezvous(struct vcpu *vcpu, uint64_t rip);
263 void vm_exit_astpending(struct vcpu *vcpu, uint64_t rip);
264 void vm_exit_reqidle(struct vcpu *vcpu, uint64_t rip);
265 int vm_snapshot_req(struct vm *vm, struct vm_snapshot_meta *meta);
266 int vm_restore_time(struct vm *vm);
267
268 #ifdef _SYS__CPUSET_H_
269 cpuset_t vm_start_cpus(struct vm *vm, const cpuset_t *tostart);
270 void vm_await_start(struct vm *vm, const cpuset_t *waiting);
271 #endif /* _SYS__CPUSET_H_ */
272
273 /*
274 * Return true if device indicated by bus/slot/func is supposed to be a
275 * pci passthrough device.
276 *
277 * Return false otherwise.
278 */
279 bool vmm_is_pptdev(int bus, int slot, int func);
280
281 void *vm_iommu_domain(struct vm *vm);
282
283 void vcpu_notify_lapic(struct vcpu *vcpu);
284 struct vatpic *vm_atpic(struct vm *vm);
285 struct vatpit *vm_atpit(struct vm *vm);
286 struct vpmtmr *vm_pmtmr(struct vm *vm);
287 struct vrtc *vm_rtc(struct vm *vm);
288
289 /*
290 * Inject exception 'vector' into the guest vcpu. This function returns 0 on
291 * success and non-zero on failure.
292 *
293 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
294 * this function directly because they enforce the trap-like or fault-like
295 * behavior of an exception.
296 *
297 * This function should only be called in the context of the thread that is
298 * executing this vcpu.
299 */
300 int vm_inject_exception(struct vcpu *vcpu, int vector, int err_valid,
301 uint32_t errcode, int restart_instruction);
302
303 /*
304 * This function is called after a VM-exit that occurred during exception or
305 * interrupt delivery through the IDT. The format of 'intinfo' is described
306 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
307 *
308 * If a VM-exit handler completes the event delivery successfully then it
309 * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
310 * if the task switch emulation is triggered via a task gate then it should
311 * call this function with 'intinfo=0' to indicate that the external event
312 * is not pending anymore.
313 *
314 * Return value is 0 on success and non-zero on failure.
315 */
316 int vm_exit_intinfo(struct vcpu *vcpu, uint64_t intinfo);
317
318 /*
319 * This function is called before every VM-entry to retrieve a pending
320 * event that should be injected into the guest. This function combines
321 * nested events into a double or triple fault.
322 *
323 * Returns 0 if there are no events that need to be injected into the guest
324 * and non-zero otherwise.
325 */
326 int vm_entry_intinfo(struct vcpu *vcpu, uint64_t *info);
327
328 int vm_get_intinfo(struct vcpu *vcpu, uint64_t *info1, uint64_t *info2);
329
330 /*
331 * Function used to keep track of the guest's TSC offset. The
332 * offset is used by the virtualization extensions to provide a consistent
333 * value for the Time Stamp Counter to the guest.
334 */
335 void vm_set_tsc_offset(struct vcpu *vcpu, uint64_t offset);
336
337 enum vm_reg_name vm_segment_name(int seg_encoding);
338
339 struct vm_copyinfo {
340 uint64_t gpa;
341 size_t len;
342 void *hva;
343 void *cookie;
344 };
345
346 /*
347 * Set up 'copyinfo[]' to copy to/from guest linear address space starting
348 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
349 * a copyin or PROT_WRITE for a copyout.
350 *
351 * retval is_fault Interpretation
352 * 0 0 Success
353 * 0 1 An exception was injected into the guest
354 * EFAULT N/A Unrecoverable error
355 *
356 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
357 * the return value is 0. The 'copyinfo[]' resources should be freed by calling
358 * 'vm_copy_teardown()' after the copy is done.
359 */
360 int vm_copy_setup(struct vcpu *vcpu, struct vm_guest_paging *paging,
361 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
362 int num_copyinfo, int *is_fault);
363 void vm_copy_teardown(struct vm_copyinfo *copyinfo, int num_copyinfo);
364 void vm_copyin(struct vm_copyinfo *copyinfo, void *kaddr, size_t len);
365 void vm_copyout(const void *kaddr, struct vm_copyinfo *copyinfo, size_t len);
366
367 int vcpu_trace_exceptions(struct vcpu *vcpu);
368 int vcpu_trap_wbinvd(struct vcpu *vcpu);
369 #endif /* KERNEL */
370
371 /*
372 * Identifiers for optional vmm capabilities
373 */
374 enum vm_cap_type {
375 VM_CAP_HALT_EXIT,
376 VM_CAP_MTRAP_EXIT,
377 VM_CAP_PAUSE_EXIT,
378 VM_CAP_UNRESTRICTED_GUEST,
379 VM_CAP_ENABLE_INVPCID,
380 VM_CAP_BPT_EXIT,
381 VM_CAP_RDPID,
382 VM_CAP_RDTSCP,
383 VM_CAP_IPI_EXIT,
384 VM_CAP_MASK_HWINTR,
385 VM_CAP_RFLAGS_TF,
386 VM_CAP_MAX
387 };
388
389 enum vm_intr_trigger {
390 EDGE_TRIGGER,
391 LEVEL_TRIGGER
392 };
393
394 /*
395 * The 'access' field has the format specified in Table 21-2 of the Intel
396 * Architecture Manual vol 3b.
397 *
398 * XXX The contents of the 'access' field are architecturally defined except
399 * bit 16 - Segment Unusable.
400 */
401 struct seg_desc {
402 uint64_t base;
403 uint32_t limit;
404 uint32_t access;
405 };
406 #define SEG_DESC_TYPE(access) ((access) & 0x001f)
407 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3)
408 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0)
409 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0)
410 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0)
411 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0)
412
413 enum vm_cpu_mode {
414 CPU_MODE_REAL,
415 CPU_MODE_PROTECTED,
416 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */
417 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */
418 };
419
420 enum vm_paging_mode {
421 PAGING_MODE_FLAT,
422 PAGING_MODE_32,
423 PAGING_MODE_PAE,
424 PAGING_MODE_64,
425 PAGING_MODE_64_LA57,
426 };
427
428 struct vm_guest_paging {
429 uint64_t cr3;
430 int cpl;
431 enum vm_cpu_mode cpu_mode;
432 enum vm_paging_mode paging_mode;
433 };
434
435 /*
436 * The data structures 'vie' and 'vie_op' are meant to be opaque to the
437 * consumers of instruction decoding. The only reason why their contents
438 * need to be exposed is because they are part of the 'vm_exit' structure.
439 */
440 struct vie_op {
441 uint8_t op_byte; /* actual opcode byte */
442 uint8_t op_type; /* type of operation (e.g. MOV) */
443 uint16_t op_flags;
444 };
445 _Static_assert(sizeof(struct vie_op) == 4, "ABI");
446 _Static_assert(_Alignof(struct vie_op) == 2, "ABI");
447
448 #define VIE_INST_SIZE 15
449 struct vie {
450 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */
451 uint8_t num_valid; /* size of the instruction */
452
453 /* The following fields are all zeroed upon restart. */
454 #define vie_startzero num_processed
455 uint8_t num_processed;
456
457 uint8_t addrsize:4, opsize:4; /* address and operand sizes */
458 uint8_t rex_w:1, /* REX prefix */
459 rex_r:1,
460 rex_x:1,
461 rex_b:1,
462 rex_present:1,
463 repz_present:1, /* REP/REPE/REPZ prefix */
464 repnz_present:1, /* REPNE/REPNZ prefix */
465 opsize_override:1, /* Operand size override */
466 addrsize_override:1, /* Address size override */
467 segment_override:1; /* Segment override */
468
469 uint8_t mod:2, /* ModRM byte */
470 reg:4,
471 rm:4;
472
473 uint8_t ss:2, /* SIB byte */
474 vex_present:1, /* VEX prefixed */
475 vex_l:1, /* L bit */
476 index:4, /* SIB byte */
477 base:4; /* SIB byte */
478
479 uint8_t disp_bytes;
480 uint8_t imm_bytes;
481
482 uint8_t scale;
483
484 uint8_t vex_reg:4, /* vvvv: first source register specifier */
485 vex_pp:2, /* pp */
486 _sparebits:2;
487
488 uint8_t _sparebytes[2];
489
490 int base_register; /* VM_REG_GUEST_xyz */
491 int index_register; /* VM_REG_GUEST_xyz */
492 int segment_register; /* VM_REG_GUEST_xyz */
493
494 int64_t displacement; /* optional addr displacement */
495 int64_t immediate; /* optional immediate operand */
496
497 uint8_t decoded; /* set to 1 if successfully decoded */
498
499 uint8_t _sparebyte;
500
501 struct vie_op op; /* opcode description */
502 };
503 _Static_assert(sizeof(struct vie) == 64, "ABI");
504 _Static_assert(__offsetof(struct vie, disp_bytes) == 22, "ABI");
505 _Static_assert(__offsetof(struct vie, scale) == 24, "ABI");
506 _Static_assert(__offsetof(struct vie, base_register) == 28, "ABI");
507
508 enum vm_exitcode {
509 VM_EXITCODE_INOUT,
510 VM_EXITCODE_VMX,
511 VM_EXITCODE_BOGUS,
512 VM_EXITCODE_RDMSR,
513 VM_EXITCODE_WRMSR,
514 VM_EXITCODE_HLT,
515 VM_EXITCODE_MTRAP,
516 VM_EXITCODE_PAUSE,
517 VM_EXITCODE_PAGING,
518 VM_EXITCODE_INST_EMUL,
519 VM_EXITCODE_SPINUP_AP,
520 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */
521 VM_EXITCODE_RENDEZVOUS,
522 VM_EXITCODE_IOAPIC_EOI,
523 VM_EXITCODE_SUSPENDED,
524 VM_EXITCODE_INOUT_STR,
525 VM_EXITCODE_TASK_SWITCH,
526 VM_EXITCODE_MONITOR,
527 VM_EXITCODE_MWAIT,
528 VM_EXITCODE_SVM,
529 VM_EXITCODE_REQIDLE,
530 VM_EXITCODE_DEBUG,
531 VM_EXITCODE_VMINSN,
532 VM_EXITCODE_BPT,
533 VM_EXITCODE_IPI,
534 VM_EXITCODE_DB,
535 VM_EXITCODE_MAX
536 };
537
538 struct vm_inout {
539 uint16_t bytes:3; /* 1 or 2 or 4 */
540 uint16_t in:1;
541 uint16_t string:1;
542 uint16_t rep:1;
543 uint16_t port;
544 uint32_t eax; /* valid for out */
545 };
546
547 struct vm_inout_str {
548 struct vm_inout inout; /* must be the first element */
549 struct vm_guest_paging paging;
550 uint64_t rflags;
551 uint64_t cr0;
552 uint64_t index;
553 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */
554 int addrsize;
555 enum vm_reg_name seg_name;
556 struct seg_desc seg_desc;
557 int cs_d;
558 uint64_t cs_base;
559 };
560
561 enum task_switch_reason {
562 TSR_CALL,
563 TSR_IRET,
564 TSR_JMP,
565 TSR_IDT_GATE, /* task gate in IDT */
566 };
567
568 struct vm_task_switch {
569 uint16_t tsssel; /* new TSS selector */
570 int ext; /* task switch due to external event */
571 uint32_t errcode;
572 int errcode_valid; /* push 'errcode' on the new stack */
573 enum task_switch_reason reason;
574 struct vm_guest_paging paging;
575 };
576
577 struct vm_exit {
578 enum vm_exitcode exitcode;
579 int inst_length; /* 0 means unknown */
580 uint64_t rip;
581 union {
582 struct vm_inout inout;
583 struct vm_inout_str inout_str;
584 struct {
585 uint64_t gpa;
586 int fault_type;
587 } paging;
588 struct {
589 uint64_t gpa;
590 uint64_t gla;
591 uint64_t cs_base;
592 int cs_d; /* CS.D */
593 struct vm_guest_paging paging;
594 struct vie vie;
595 } inst_emul;
596 /*
597 * VMX specific payload. Used when there is no "better"
598 * exitcode to represent the VM-exit.
599 */
600 struct {
601 int status; /* vmx inst status */
602 /*
603 * 'exit_reason' and 'exit_qualification' are valid
604 * only if 'status' is zero.
605 */
606 uint32_t exit_reason;
607 uint64_t exit_qualification;
608 /*
609 * 'inst_error' and 'inst_type' are valid
610 * only if 'status' is non-zero.
611 */
612 int inst_type;
613 int inst_error;
614 } vmx;
615 /*
616 * SVM specific payload.
617 */
618 struct {
619 uint64_t exitcode;
620 uint64_t exitinfo1;
621 uint64_t exitinfo2;
622 } svm;
623 struct {
624 int inst_length;
625 } bpt;
626 struct {
627 int trace_trap;
628 int pushf_intercept;
629 int tf_shadow_val;
630 struct vm_guest_paging paging;
631 } dbg;
632 struct {
633 uint32_t code; /* ecx value */
634 uint64_t wval;
635 } msr;
636 struct {
637 int vcpu;
638 uint64_t rip;
639 } spinup_ap;
640 struct {
641 uint64_t rflags;
642 uint64_t intr_status;
643 } hlt;
644 struct {
645 int vector;
646 } ioapic_eoi;
647 struct {
648 enum vm_suspend_how how;
649 } suspended;
650 struct {
651 /*
652 * The destination vCPU mask is saved in vcpu->cpuset
653 * and is copied out to userspace separately to avoid
654 * ABI concerns.
655 */
656 uint32_t mode;
657 uint8_t vector;
658 } ipi;
659 struct vm_task_switch task_switch;
660 } u;
661 };
662
663 /* APIs to inject faults into the guest */
664 void vm_inject_fault(struct vcpu *vcpu, int vector, int errcode_valid,
665 int errcode);
666
667 static __inline void
vm_inject_ud(struct vcpu * vcpu)668 vm_inject_ud(struct vcpu *vcpu)
669 {
670 vm_inject_fault(vcpu, IDT_UD, 0, 0);
671 }
672
673 static __inline void
vm_inject_gp(struct vcpu * vcpu)674 vm_inject_gp(struct vcpu *vcpu)
675 {
676 vm_inject_fault(vcpu, IDT_GP, 1, 0);
677 }
678
679 static __inline void
vm_inject_ac(struct vcpu * vcpu,int errcode)680 vm_inject_ac(struct vcpu *vcpu, int errcode)
681 {
682 vm_inject_fault(vcpu, IDT_AC, 1, errcode);
683 }
684
685 static __inline void
vm_inject_ss(struct vcpu * vcpu,int errcode)686 vm_inject_ss(struct vcpu *vcpu, int errcode)
687 {
688 vm_inject_fault(vcpu, IDT_SS, 1, errcode);
689 }
690
691 void vm_inject_pf(struct vcpu *vcpu, int error_code, uint64_t cr2);
692
693 #endif /* _VMM_H_ */
694