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 int vm_reset_pptdev(struct vm *vm, int bus, int slot, int func);
236
237 int vm_get_register(struct vcpu *vcpu, int reg, uint64_t *retval);
238 int vm_set_register(struct vcpu *vcpu, int reg, uint64_t val);
239 int vm_get_seg_desc(struct vcpu *vcpu, int reg,
240 struct seg_desc *ret_desc);
241 int vm_set_seg_desc(struct vcpu *vcpu, int reg,
242 struct seg_desc *desc);
243 int vm_run(struct vcpu *vcpu);
244 int vm_inject_nmi(struct vcpu *vcpu);
245 int vm_nmi_pending(struct vcpu *vcpu);
246 void vm_nmi_clear(struct vcpu *vcpu);
247 int vm_inject_extint(struct vcpu *vcpu);
248 int vm_extint_pending(struct vcpu *vcpu);
249 void vm_extint_clear(struct vcpu *vcpu);
250 struct vlapic *vm_lapic(struct vcpu *vcpu);
251 struct vioapic *vm_ioapic(struct vm *vm);
252 struct vhpet *vm_hpet(struct vm *vm);
253 int vm_get_capability(struct vcpu *vcpu, int type, int *val);
254 int vm_set_capability(struct vcpu *vcpu, int type, int val);
255 int vm_get_x2apic_state(struct vcpu *vcpu, enum x2apic_state *state);
256 int vm_set_x2apic_state(struct vcpu *vcpu, enum x2apic_state state);
257 int vm_apicid2vcpuid(struct vm *vm, int apicid);
258 int vm_restart_instruction(struct vcpu *vcpu);
259 struct vm_exit *vm_exitinfo(struct vcpu *vcpu);
260 cpuset_t *vm_exitinfo_cpuset(struct vcpu *vcpu);
261 void vm_exit_suspended(struct vcpu *vcpu, uint64_t rip);
262 void vm_exit_debug(struct vcpu *vcpu, uint64_t rip);
263 void vm_exit_rendezvous(struct vcpu *vcpu, uint64_t rip);
264 void vm_exit_astpending(struct vcpu *vcpu, uint64_t rip);
265 void vm_exit_reqidle(struct vcpu *vcpu, uint64_t rip);
266 int vm_snapshot_req(struct vm *vm, struct vm_snapshot_meta *meta);
267 int vm_restore_time(struct vm *vm);
268
269 #ifdef _SYS__CPUSET_H_
270 cpuset_t vm_start_cpus(struct vm *vm, const cpuset_t *tostart);
271 void vm_await_start(struct vm *vm, const cpuset_t *waiting);
272 #endif /* _SYS__CPUSET_H_ */
273
274 /*
275 * Return true if device indicated by bus/slot/func is supposed to be a
276 * pci passthrough device.
277 *
278 * Return false otherwise.
279 */
280 bool vmm_is_pptdev(int bus, int slot, int func);
281
282 void *vm_iommu_domain(struct vm *vm);
283
284 void vcpu_notify_lapic(struct vcpu *vcpu);
285 struct vatpic *vm_atpic(struct vm *vm);
286 struct vatpit *vm_atpit(struct vm *vm);
287 struct vpmtmr *vm_pmtmr(struct vm *vm);
288 struct vrtc *vm_rtc(struct vm *vm);
289
290 /*
291 * Inject exception 'vector' into the guest vcpu. This function returns 0 on
292 * success and non-zero on failure.
293 *
294 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling
295 * this function directly because they enforce the trap-like or fault-like
296 * behavior of an exception.
297 *
298 * This function should only be called in the context of the thread that is
299 * executing this vcpu.
300 */
301 int vm_inject_exception(struct vcpu *vcpu, int vector, int err_valid,
302 uint32_t errcode, int restart_instruction);
303
304 /*
305 * This function is called after a VM-exit that occurred during exception or
306 * interrupt delivery through the IDT. The format of 'intinfo' is described
307 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2.
308 *
309 * If a VM-exit handler completes the event delivery successfully then it
310 * should call vm_exit_intinfo() to extinguish the pending event. For e.g.,
311 * if the task switch emulation is triggered via a task gate then it should
312 * call this function with 'intinfo=0' to indicate that the external event
313 * is not pending anymore.
314 *
315 * Return value is 0 on success and non-zero on failure.
316 */
317 int vm_exit_intinfo(struct vcpu *vcpu, uint64_t intinfo);
318
319 /*
320 * This function is called before every VM-entry to retrieve a pending
321 * event that should be injected into the guest. This function combines
322 * nested events into a double or triple fault.
323 *
324 * Returns 0 if there are no events that need to be injected into the guest
325 * and non-zero otherwise.
326 */
327 int vm_entry_intinfo(struct vcpu *vcpu, uint64_t *info);
328
329 int vm_get_intinfo(struct vcpu *vcpu, uint64_t *info1, uint64_t *info2);
330
331 /*
332 * Function used to keep track of the guest's TSC offset. The
333 * offset is used by the virtualization extensions to provide a consistent
334 * value for the Time Stamp Counter to the guest.
335 */
336 void vm_set_tsc_offset(struct vcpu *vcpu, uint64_t offset);
337
338 enum vm_reg_name vm_segment_name(int seg_encoding);
339
340 struct vm_copyinfo {
341 uint64_t gpa;
342 size_t len;
343 void *hva;
344 void *cookie;
345 };
346
347 /*
348 * Set up 'copyinfo[]' to copy to/from guest linear address space starting
349 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for
350 * a copyin or PROT_WRITE for a copyout.
351 *
352 * retval is_fault Interpretation
353 * 0 0 Success
354 * 0 1 An exception was injected into the guest
355 * EFAULT N/A Unrecoverable error
356 *
357 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if
358 * the return value is 0. The 'copyinfo[]' resources should be freed by calling
359 * 'vm_copy_teardown()' after the copy is done.
360 */
361 int vm_copy_setup(struct vcpu *vcpu, struct vm_guest_paging *paging,
362 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo,
363 int num_copyinfo, int *is_fault);
364 void vm_copy_teardown(struct vm_copyinfo *copyinfo, int num_copyinfo);
365 void vm_copyin(struct vm_copyinfo *copyinfo, void *kaddr, size_t len);
366 void vm_copyout(const void *kaddr, struct vm_copyinfo *copyinfo, size_t len);
367
368 int vcpu_trace_exceptions(struct vcpu *vcpu);
369 int vcpu_trap_wbinvd(struct vcpu *vcpu);
370 #endif /* KERNEL */
371
372 /*
373 * Identifiers for optional vmm capabilities
374 */
375 enum vm_cap_type {
376 VM_CAP_HALT_EXIT,
377 VM_CAP_MTRAP_EXIT,
378 VM_CAP_PAUSE_EXIT,
379 VM_CAP_UNRESTRICTED_GUEST,
380 VM_CAP_ENABLE_INVPCID,
381 VM_CAP_BPT_EXIT,
382 VM_CAP_RDPID,
383 VM_CAP_RDTSCP,
384 VM_CAP_IPI_EXIT,
385 VM_CAP_MASK_HWINTR,
386 VM_CAP_RFLAGS_TF,
387 VM_CAP_MAX
388 };
389
390 enum vm_intr_trigger {
391 EDGE_TRIGGER,
392 LEVEL_TRIGGER
393 };
394
395 /*
396 * The 'access' field has the format specified in Table 21-2 of the Intel
397 * Architecture Manual vol 3b.
398 *
399 * XXX The contents of the 'access' field are architecturally defined except
400 * bit 16 - Segment Unusable.
401 */
402 struct seg_desc {
403 uint64_t base;
404 uint32_t limit;
405 uint32_t access;
406 };
407 #define SEG_DESC_TYPE(access) ((access) & 0x001f)
408 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3)
409 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0)
410 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0)
411 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0)
412 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0)
413
414 enum vm_cpu_mode {
415 CPU_MODE_REAL,
416 CPU_MODE_PROTECTED,
417 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */
418 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */
419 };
420
421 enum vm_paging_mode {
422 PAGING_MODE_FLAT,
423 PAGING_MODE_32,
424 PAGING_MODE_PAE,
425 PAGING_MODE_64,
426 PAGING_MODE_64_LA57,
427 };
428
429 struct vm_guest_paging {
430 uint64_t cr3;
431 int cpl;
432 enum vm_cpu_mode cpu_mode;
433 enum vm_paging_mode paging_mode;
434 };
435
436 /*
437 * The data structures 'vie' and 'vie_op' are meant to be opaque to the
438 * consumers of instruction decoding. The only reason why their contents
439 * need to be exposed is because they are part of the 'vm_exit' structure.
440 */
441 struct vie_op {
442 uint8_t op_byte; /* actual opcode byte */
443 uint8_t op_type; /* type of operation (e.g. MOV) */
444 uint16_t op_flags;
445 };
446 _Static_assert(sizeof(struct vie_op) == 4, "ABI");
447 _Static_assert(_Alignof(struct vie_op) == 2, "ABI");
448
449 #define VIE_INST_SIZE 15
450 struct vie {
451 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */
452 uint8_t num_valid; /* size of the instruction */
453
454 /* The following fields are all zeroed upon restart. */
455 #define vie_startzero num_processed
456 uint8_t num_processed;
457
458 uint8_t addrsize:4, opsize:4; /* address and operand sizes */
459 uint8_t rex_w:1, /* REX prefix */
460 rex_r:1,
461 rex_x:1,
462 rex_b:1,
463 rex_present:1,
464 repz_present:1, /* REP/REPE/REPZ prefix */
465 repnz_present:1, /* REPNE/REPNZ prefix */
466 opsize_override:1, /* Operand size override */
467 addrsize_override:1, /* Address size override */
468 segment_override:1; /* Segment override */
469
470 uint8_t mod:2, /* ModRM byte */
471 reg:4,
472 rm:4;
473
474 uint8_t ss:2, /* SIB byte */
475 vex_present:1, /* VEX prefixed */
476 vex_l:1, /* L bit */
477 index:4, /* SIB byte */
478 base:4; /* SIB byte */
479
480 uint8_t disp_bytes;
481 uint8_t imm_bytes;
482
483 uint8_t scale;
484
485 uint8_t vex_reg:4, /* vvvv: first source register specifier */
486 vex_pp:2, /* pp */
487 _sparebits:2;
488
489 uint8_t _sparebytes[2];
490
491 int base_register; /* VM_REG_GUEST_xyz */
492 int index_register; /* VM_REG_GUEST_xyz */
493 int segment_register; /* VM_REG_GUEST_xyz */
494
495 int64_t displacement; /* optional addr displacement */
496 int64_t immediate; /* optional immediate operand */
497
498 uint8_t decoded; /* set to 1 if successfully decoded */
499
500 uint8_t _sparebyte;
501
502 struct vie_op op; /* opcode description */
503 };
504 _Static_assert(sizeof(struct vie) == 64, "ABI");
505 _Static_assert(__offsetof(struct vie, disp_bytes) == 22, "ABI");
506 _Static_assert(__offsetof(struct vie, scale) == 24, "ABI");
507 _Static_assert(__offsetof(struct vie, base_register) == 28, "ABI");
508
509 enum vm_exitcode {
510 VM_EXITCODE_INOUT,
511 VM_EXITCODE_VMX,
512 VM_EXITCODE_BOGUS,
513 VM_EXITCODE_RDMSR,
514 VM_EXITCODE_WRMSR,
515 VM_EXITCODE_HLT,
516 VM_EXITCODE_MTRAP,
517 VM_EXITCODE_PAUSE,
518 VM_EXITCODE_PAGING,
519 VM_EXITCODE_INST_EMUL,
520 VM_EXITCODE_SPINUP_AP,
521 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */
522 VM_EXITCODE_RENDEZVOUS,
523 VM_EXITCODE_IOAPIC_EOI,
524 VM_EXITCODE_SUSPENDED,
525 VM_EXITCODE_INOUT_STR,
526 VM_EXITCODE_TASK_SWITCH,
527 VM_EXITCODE_MONITOR,
528 VM_EXITCODE_MWAIT,
529 VM_EXITCODE_SVM,
530 VM_EXITCODE_REQIDLE,
531 VM_EXITCODE_DEBUG,
532 VM_EXITCODE_VMINSN,
533 VM_EXITCODE_BPT,
534 VM_EXITCODE_IPI,
535 VM_EXITCODE_DB,
536 VM_EXITCODE_MAX
537 };
538
539 struct 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
548 struct 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 int cs_d;
559 uint64_t cs_base;
560 };
561
562 enum task_switch_reason {
563 TSR_CALL,
564 TSR_IRET,
565 TSR_JMP,
566 TSR_IDT_GATE, /* task gate in IDT */
567 };
568
569 struct vm_task_switch {
570 uint16_t tsssel; /* new TSS selector */
571 int ext; /* task switch due to external event */
572 uint32_t errcode;
573 int errcode_valid; /* push 'errcode' on the new stack */
574 enum task_switch_reason reason;
575 struct vm_guest_paging paging;
576 };
577
578 struct vm_exit {
579 enum vm_exitcode exitcode;
580 int inst_length; /* 0 means unknown */
581 uint64_t rip;
582 union {
583 struct vm_inout inout;
584 struct vm_inout_str inout_str;
585 struct {
586 uint64_t gpa;
587 int fault_type;
588 } paging;
589 struct {
590 uint64_t gpa;
591 uint64_t gla;
592 uint64_t cs_base;
593 int cs_d; /* CS.D */
594 struct vm_guest_paging paging;
595 struct vie vie;
596 } inst_emul;
597 /*
598 * VMX specific payload. Used when there is no "better"
599 * exitcode to represent the VM-exit.
600 */
601 struct {
602 int status; /* vmx inst status */
603 /*
604 * 'exit_reason' and 'exit_qualification' are valid
605 * only if 'status' is zero.
606 */
607 uint32_t exit_reason;
608 uint64_t exit_qualification;
609 /*
610 * 'inst_error' and 'inst_type' are valid
611 * only if 'status' is non-zero.
612 */
613 int inst_type;
614 int inst_error;
615 } vmx;
616 /*
617 * SVM specific payload.
618 */
619 struct {
620 uint64_t exitcode;
621 uint64_t exitinfo1;
622 uint64_t exitinfo2;
623 } svm;
624 struct {
625 int inst_length;
626 } bpt;
627 struct {
628 int trace_trap;
629 int pushf_intercept;
630 int tf_shadow_val;
631 struct vm_guest_paging paging;
632 } dbg;
633 struct {
634 uint32_t code; /* ecx value */
635 uint64_t wval;
636 } msr;
637 struct {
638 int vcpu;
639 uint64_t rip;
640 } spinup_ap;
641 struct {
642 uint64_t rflags;
643 uint64_t intr_status;
644 } hlt;
645 struct {
646 int vector;
647 } ioapic_eoi;
648 struct {
649 enum vm_suspend_how how;
650 } suspended;
651 struct {
652 /*
653 * The destination vCPU mask is saved in vcpu->cpuset
654 * and is copied out to userspace separately to avoid
655 * ABI concerns.
656 */
657 uint32_t mode;
658 uint8_t vector;
659 } ipi;
660 struct vm_task_switch task_switch;
661 } u;
662 };
663
664 /* APIs to inject faults into the guest */
665 void vm_inject_fault(struct vcpu *vcpu, int vector, int errcode_valid,
666 int errcode);
667
668 static __inline void
vm_inject_ud(struct vcpu * vcpu)669 vm_inject_ud(struct vcpu *vcpu)
670 {
671 vm_inject_fault(vcpu, IDT_UD, 0, 0);
672 }
673
674 static __inline void
vm_inject_gp(struct vcpu * vcpu)675 vm_inject_gp(struct vcpu *vcpu)
676 {
677 vm_inject_fault(vcpu, IDT_GP, 1, 0);
678 }
679
680 static __inline void
vm_inject_ac(struct vcpu * vcpu,int errcode)681 vm_inject_ac(struct vcpu *vcpu, int errcode)
682 {
683 vm_inject_fault(vcpu, IDT_AC, 1, errcode);
684 }
685
686 static __inline void
vm_inject_ss(struct vcpu * vcpu,int errcode)687 vm_inject_ss(struct vcpu *vcpu, int errcode)
688 {
689 vm_inject_fault(vcpu, IDT_SS, 1, errcode);
690 }
691
692 void vm_inject_pf(struct vcpu *vcpu, int error_code, uint64_t cr2);
693
694 #endif /* _VMM_H_ */
695