vmm.h revision 347074
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: stable/11/sys/amd64/include/vmm.h 347074 2019-05-04 00:59:11Z jhb $ 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_GUEST_DR0, 87 VM_REG_GUEST_DR1, 88 VM_REG_GUEST_DR2, 89 VM_REG_GUEST_DR3, 90 VM_REG_GUEST_DR6, 91 VM_REG_LAST 92}; 93 94enum x2apic_state { 95 X2APIC_DISABLED, 96 X2APIC_ENABLED, 97 X2APIC_STATE_LAST 98}; 99 100#define VM_INTINFO_VECTOR(info) ((info) & 0xff) 101#define VM_INTINFO_DEL_ERRCODE 0x800 102#define VM_INTINFO_RSVD 0x7ffff000 103#define VM_INTINFO_VALID 0x80000000 104#define VM_INTINFO_TYPE 0x700 105#define VM_INTINFO_HWINTR (0 << 8) 106#define VM_INTINFO_NMI (2 << 8) 107#define VM_INTINFO_HWEXCEPTION (3 << 8) 108#define VM_INTINFO_SWINTR (4 << 8) 109 110#ifdef _KERNEL 111 112#define VM_MAX_NAMELEN 32 113 114struct vm; 115struct vm_exception; 116struct seg_desc; 117struct vm_exit; 118struct vm_run; 119struct vhpet; 120struct vioapic; 121struct vlapic; 122struct vmspace; 123struct vm_object; 124struct vm_guest_paging; 125struct pmap; 126 127struct vm_eventinfo { 128 void *rptr; /* rendezvous cookie */ 129 int *sptr; /* suspend cookie */ 130 int *iptr; /* reqidle cookie */ 131}; 132 133typedef int (*vmm_init_func_t)(int ipinum); 134typedef int (*vmm_cleanup_func_t)(void); 135typedef void (*vmm_resume_func_t)(void); 136typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap); 137typedef int (*vmi_run_func_t)(void *vmi, int vcpu, register_t rip, 138 struct pmap *pmap, struct vm_eventinfo *info); 139typedef void (*vmi_cleanup_func_t)(void *vmi); 140typedef int (*vmi_get_register_t)(void *vmi, int vcpu, int num, 141 uint64_t *retval); 142typedef int (*vmi_set_register_t)(void *vmi, int vcpu, int num, 143 uint64_t val); 144typedef int (*vmi_get_desc_t)(void *vmi, int vcpu, int num, 145 struct seg_desc *desc); 146typedef int (*vmi_set_desc_t)(void *vmi, int vcpu, int num, 147 struct seg_desc *desc); 148typedef int (*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval); 149typedef int (*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val); 150typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max); 151typedef void (*vmi_vmspace_free)(struct vmspace *vmspace); 152typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu); 153typedef void (*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic); 154 155struct vmm_ops { 156 vmm_init_func_t init; /* module wide initialization */ 157 vmm_cleanup_func_t cleanup; 158 vmm_resume_func_t resume; 159 160 vmi_init_func_t vminit; /* vm-specific initialization */ 161 vmi_run_func_t vmrun; 162 vmi_cleanup_func_t vmcleanup; 163 vmi_get_register_t vmgetreg; 164 vmi_set_register_t vmsetreg; 165 vmi_get_desc_t vmgetdesc; 166 vmi_set_desc_t vmsetdesc; 167 vmi_get_cap_t vmgetcap; 168 vmi_set_cap_t vmsetcap; 169 vmi_vmspace_alloc vmspace_alloc; 170 vmi_vmspace_free vmspace_free; 171 vmi_vlapic_init vlapic_init; 172 vmi_vlapic_cleanup vlapic_cleanup; 173}; 174 175extern struct vmm_ops vmm_ops_intel; 176extern struct vmm_ops vmm_ops_amd; 177 178int vm_create(const char *name, struct vm **retvm); 179void vm_destroy(struct vm *vm); 180int vm_reinit(struct vm *vm); 181const char *vm_name(struct vm *vm); 182 183/* 184 * APIs that modify the guest memory map require all vcpus to be frozen. 185 */ 186int vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t off, 187 size_t len, int prot, int flags); 188int vm_alloc_memseg(struct vm *vm, int ident, size_t len, bool sysmem); 189void vm_free_memseg(struct vm *vm, int ident); 190int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa); 191int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len); 192int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func); 193int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func); 194 195/* 196 * APIs that inspect the guest memory map require only a *single* vcpu to 197 * be frozen. This acts like a read lock on the guest memory map since any 198 * modification requires *all* vcpus to be frozen. 199 */ 200int vm_mmap_getnext(struct vm *vm, vm_paddr_t *gpa, int *segid, 201 vm_ooffset_t *segoff, size_t *len, int *prot, int *flags); 202int vm_get_memseg(struct vm *vm, int ident, size_t *len, bool *sysmem, 203 struct vm_object **objptr); 204void *vm_gpa_hold(struct vm *, int vcpuid, vm_paddr_t gpa, size_t len, 205 int prot, void **cookie); 206void vm_gpa_release(void *cookie); 207bool vm_mem_allocated(struct vm *vm, int vcpuid, vm_paddr_t gpa); 208 209int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval); 210int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val); 211int vm_get_seg_desc(struct vm *vm, int vcpu, int reg, 212 struct seg_desc *ret_desc); 213int vm_set_seg_desc(struct vm *vm, int vcpu, int reg, 214 struct seg_desc *desc); 215int vm_run(struct vm *vm, struct vm_run *vmrun); 216int vm_suspend(struct vm *vm, enum vm_suspend_how how); 217int vm_inject_nmi(struct vm *vm, int vcpu); 218int vm_nmi_pending(struct vm *vm, int vcpuid); 219void vm_nmi_clear(struct vm *vm, int vcpuid); 220int vm_inject_extint(struct vm *vm, int vcpu); 221int vm_extint_pending(struct vm *vm, int vcpuid); 222void vm_extint_clear(struct vm *vm, int vcpuid); 223struct vlapic *vm_lapic(struct vm *vm, int cpu); 224struct vioapic *vm_ioapic(struct vm *vm); 225struct vhpet *vm_hpet(struct vm *vm); 226int vm_get_capability(struct vm *vm, int vcpu, int type, int *val); 227int vm_set_capability(struct vm *vm, int vcpu, int type, int val); 228int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state); 229int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state); 230int vm_apicid2vcpuid(struct vm *vm, int apicid); 231int vm_activate_cpu(struct vm *vm, int vcpu); 232struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid); 233void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip); 234void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip); 235void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip); 236void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip); 237 238#ifdef _SYS__CPUSET_H_ 239/* 240 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'. 241 * The rendezvous 'func(arg)' is not allowed to do anything that will 242 * cause the thread to be put to sleep. 243 * 244 * If the rendezvous is being initiated from a vcpu context then the 245 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1. 246 * 247 * The caller cannot hold any locks when initiating the rendezvous. 248 * 249 * The implementation of this API may cause vcpus other than those specified 250 * by 'dest' to be stalled. The caller should not rely on any vcpus making 251 * forward progress when the rendezvous is in progress. 252 */ 253typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg); 254void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest, 255 vm_rendezvous_func_t func, void *arg); 256cpuset_t vm_active_cpus(struct vm *vm); 257cpuset_t vm_suspended_cpus(struct vm *vm); 258#endif /* _SYS__CPUSET_H_ */ 259 260static __inline int 261vcpu_rendezvous_pending(struct vm_eventinfo *info) 262{ 263 264 return (*((uintptr_t *)(info->rptr)) != 0); 265} 266 267static __inline int 268vcpu_suspended(struct vm_eventinfo *info) 269{ 270 271 return (*info->sptr); 272} 273 274static __inline int 275vcpu_reqidle(struct vm_eventinfo *info) 276{ 277 278 return (*info->iptr); 279} 280 281/* 282 * Return 1 if device indicated by bus/slot/func is supposed to be a 283 * pci passthrough device. 284 * 285 * Return 0 otherwise. 286 */ 287int vmm_is_pptdev(int bus, int slot, int func); 288 289void *vm_iommu_domain(struct vm *vm); 290 291enum vcpu_state { 292 VCPU_IDLE, 293 VCPU_FROZEN, 294 VCPU_RUNNING, 295 VCPU_SLEEPING, 296}; 297 298int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state, 299 bool from_idle); 300enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu); 301 302static int __inline 303vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu) 304{ 305 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING); 306} 307 308#ifdef _SYS_PROC_H_ 309static int __inline 310vcpu_should_yield(struct vm *vm, int vcpu) 311{ 312 313 if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED)) 314 return (1); 315 else if (curthread->td_owepreempt) 316 return (1); 317 else 318 return (0); 319} 320#endif 321 322void *vcpu_stats(struct vm *vm, int vcpu); 323void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr); 324struct vmspace *vm_get_vmspace(struct vm *vm); 325struct vatpic *vm_atpic(struct vm *vm); 326struct vatpit *vm_atpit(struct vm *vm); 327struct vpmtmr *vm_pmtmr(struct vm *vm); 328struct vrtc *vm_rtc(struct vm *vm); 329 330/* 331 * Inject exception 'vector' into the guest vcpu. This function returns 0 on 332 * success and non-zero on failure. 333 * 334 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling 335 * this function directly because they enforce the trap-like or fault-like 336 * behavior of an exception. 337 * 338 * This function should only be called in the context of the thread that is 339 * executing this vcpu. 340 */ 341int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid, 342 uint32_t errcode, int restart_instruction); 343 344/* 345 * This function is called after a VM-exit that occurred during exception or 346 * interrupt delivery through the IDT. The format of 'intinfo' is described 347 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2. 348 * 349 * If a VM-exit handler completes the event delivery successfully then it 350 * should call vm_exit_intinfo() to extinguish the pending event. For e.g., 351 * if the task switch emulation is triggered via a task gate then it should 352 * call this function with 'intinfo=0' to indicate that the external event 353 * is not pending anymore. 354 * 355 * Return value is 0 on success and non-zero on failure. 356 */ 357int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo); 358 359/* 360 * This function is called before every VM-entry to retrieve a pending 361 * event that should be injected into the guest. This function combines 362 * nested events into a double or triple fault. 363 * 364 * Returns 0 if there are no events that need to be injected into the guest 365 * and non-zero otherwise. 366 */ 367int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info); 368 369int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2); 370 371enum vm_reg_name vm_segment_name(int seg_encoding); 372 373struct vm_copyinfo { 374 uint64_t gpa; 375 size_t len; 376 void *hva; 377 void *cookie; 378}; 379 380/* 381 * Set up 'copyinfo[]' to copy to/from guest linear address space starting 382 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for 383 * a copyin or PROT_WRITE for a copyout. 384 * 385 * retval is_fault Interpretation 386 * 0 0 Success 387 * 0 1 An exception was injected into the guest 388 * EFAULT N/A Unrecoverable error 389 * 390 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if 391 * the return value is 0. The 'copyinfo[]' resources should be freed by calling 392 * 'vm_copy_teardown()' after the copy is done. 393 */ 394int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging, 395 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo, 396 int num_copyinfo, int *is_fault); 397void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 398 int num_copyinfo); 399void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 400 void *kaddr, size_t len); 401void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr, 402 struct vm_copyinfo *copyinfo, size_t len); 403 404int vcpu_trace_exceptions(struct vm *vm, int vcpuid); 405#endif /* KERNEL */ 406 407#define VM_MAXCPU 16 /* maximum virtual cpus */ 408 409/* 410 * Identifiers for optional vmm capabilities 411 */ 412enum vm_cap_type { 413 VM_CAP_HALT_EXIT, 414 VM_CAP_MTRAP_EXIT, 415 VM_CAP_PAUSE_EXIT, 416 VM_CAP_UNRESTRICTED_GUEST, 417 VM_CAP_ENABLE_INVPCID, 418 VM_CAP_MAX 419}; 420 421enum vm_intr_trigger { 422 EDGE_TRIGGER, 423 LEVEL_TRIGGER 424}; 425 426/* 427 * The 'access' field has the format specified in Table 21-2 of the Intel 428 * Architecture Manual vol 3b. 429 * 430 * XXX The contents of the 'access' field are architecturally defined except 431 * bit 16 - Segment Unusable. 432 */ 433struct seg_desc { 434 uint64_t base; 435 uint32_t limit; 436 uint32_t access; 437}; 438#define SEG_DESC_TYPE(access) ((access) & 0x001f) 439#define SEG_DESC_DPL(access) (((access) >> 5) & 0x3) 440#define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0) 441#define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0) 442#define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0) 443#define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0) 444 445enum vm_cpu_mode { 446 CPU_MODE_REAL, 447 CPU_MODE_PROTECTED, 448 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 449 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 450}; 451 452enum vm_paging_mode { 453 PAGING_MODE_FLAT, 454 PAGING_MODE_32, 455 PAGING_MODE_PAE, 456 PAGING_MODE_64, 457}; 458 459struct vm_guest_paging { 460 uint64_t cr3; 461 int cpl; 462 enum vm_cpu_mode cpu_mode; 463 enum vm_paging_mode paging_mode; 464}; 465 466/* 467 * The data structures 'vie' and 'vie_op' are meant to be opaque to the 468 * consumers of instruction decoding. The only reason why their contents 469 * need to be exposed is because they are part of the 'vm_exit' structure. 470 */ 471struct vie_op { 472 uint8_t op_byte; /* actual opcode byte */ 473 uint8_t op_type; /* type of operation (e.g. MOV) */ 474 uint16_t op_flags; 475}; 476 477#define VIE_INST_SIZE 15 478struct vie { 479 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */ 480 uint8_t num_valid; /* size of the instruction */ 481 uint8_t num_processed; 482 483 uint8_t addrsize:4, opsize:4; /* address and operand sizes */ 484 uint8_t rex_w:1, /* REX prefix */ 485 rex_r:1, 486 rex_x:1, 487 rex_b:1, 488 rex_present:1, 489 repz_present:1, /* REP/REPE/REPZ prefix */ 490 repnz_present:1, /* REPNE/REPNZ prefix */ 491 opsize_override:1, /* Operand size override */ 492 addrsize_override:1, /* Address size override */ 493 segment_override:1; /* Segment override */ 494 495 uint8_t mod:2, /* ModRM byte */ 496 reg:4, 497 rm:4; 498 499 uint8_t ss:2, /* SIB byte */ 500 index:4, 501 base:4; 502 503 uint8_t disp_bytes; 504 uint8_t imm_bytes; 505 506 uint8_t scale; 507 int base_register; /* VM_REG_GUEST_xyz */ 508 int index_register; /* VM_REG_GUEST_xyz */ 509 int segment_register; /* VM_REG_GUEST_xyz */ 510 511 int64_t displacement; /* optional addr displacement */ 512 int64_t immediate; /* optional immediate operand */ 513 514 uint8_t decoded; /* set to 1 if successfully decoded */ 515 516 struct vie_op op; /* opcode description */ 517}; 518 519enum vm_exitcode { 520 VM_EXITCODE_INOUT, 521 VM_EXITCODE_VMX, 522 VM_EXITCODE_BOGUS, 523 VM_EXITCODE_RDMSR, 524 VM_EXITCODE_WRMSR, 525 VM_EXITCODE_HLT, 526 VM_EXITCODE_MTRAP, 527 VM_EXITCODE_PAUSE, 528 VM_EXITCODE_PAGING, 529 VM_EXITCODE_INST_EMUL, 530 VM_EXITCODE_SPINUP_AP, 531 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */ 532 VM_EXITCODE_RENDEZVOUS, 533 VM_EXITCODE_IOAPIC_EOI, 534 VM_EXITCODE_SUSPENDED, 535 VM_EXITCODE_INOUT_STR, 536 VM_EXITCODE_TASK_SWITCH, 537 VM_EXITCODE_MONITOR, 538 VM_EXITCODE_MWAIT, 539 VM_EXITCODE_SVM, 540 VM_EXITCODE_REQIDLE, 541 VM_EXITCODE_MAX 542}; 543 544struct vm_inout { 545 uint16_t bytes:3; /* 1 or 2 or 4 */ 546 uint16_t in:1; 547 uint16_t string:1; 548 uint16_t rep:1; 549 uint16_t port; 550 uint32_t eax; /* valid for out */ 551}; 552 553struct vm_inout_str { 554 struct vm_inout inout; /* must be the first element */ 555 struct vm_guest_paging paging; 556 uint64_t rflags; 557 uint64_t cr0; 558 uint64_t index; 559 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */ 560 int addrsize; 561 enum vm_reg_name seg_name; 562 struct seg_desc seg_desc; 563}; 564 565enum task_switch_reason { 566 TSR_CALL, 567 TSR_IRET, 568 TSR_JMP, 569 TSR_IDT_GATE, /* task gate in IDT */ 570}; 571 572struct vm_task_switch { 573 uint16_t tsssel; /* new TSS selector */ 574 int ext; /* task switch due to external event */ 575 uint32_t errcode; 576 int errcode_valid; /* push 'errcode' on the new stack */ 577 enum task_switch_reason reason; 578 struct vm_guest_paging paging; 579}; 580 581struct vm_exit { 582 enum vm_exitcode exitcode; 583 int inst_length; /* 0 means unknown */ 584 uint64_t rip; 585 union { 586 struct vm_inout inout; 587 struct vm_inout_str inout_str; 588 struct { 589 uint64_t gpa; 590 int fault_type; 591 } paging; 592 struct { 593 uint64_t gpa; 594 uint64_t gla; 595 uint64_t cs_base; 596 int cs_d; /* CS.D */ 597 struct vm_guest_paging paging; 598 struct vie vie; 599 } inst_emul; 600 /* 601 * VMX specific payload. Used when there is no "better" 602 * exitcode to represent the VM-exit. 603 */ 604 struct { 605 int status; /* vmx inst status */ 606 /* 607 * 'exit_reason' and 'exit_qualification' are valid 608 * only if 'status' is zero. 609 */ 610 uint32_t exit_reason; 611 uint64_t exit_qualification; 612 /* 613 * 'inst_error' and 'inst_type' are valid 614 * only if 'status' is non-zero. 615 */ 616 int inst_type; 617 int inst_error; 618 } vmx; 619 /* 620 * SVM specific payload. 621 */ 622 struct { 623 uint64_t exitcode; 624 uint64_t exitinfo1; 625 uint64_t exitinfo2; 626 } svm; 627 struct { 628 uint32_t code; /* ecx value */ 629 uint64_t wval; 630 } msr; 631 struct { 632 int vcpu; 633 uint64_t rip; 634 } spinup_ap; 635 struct { 636 uint64_t rflags; 637 uint64_t intr_status; 638 } hlt; 639 struct { 640 int vector; 641 } ioapic_eoi; 642 struct { 643 enum vm_suspend_how how; 644 } suspended; 645 struct vm_task_switch task_switch; 646 } u; 647}; 648 649/* APIs to inject faults into the guest */ 650void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid, 651 int errcode); 652 653static __inline void 654vm_inject_ud(void *vm, int vcpuid) 655{ 656 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0); 657} 658 659static __inline void 660vm_inject_gp(void *vm, int vcpuid) 661{ 662 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0); 663} 664 665static __inline void 666vm_inject_ac(void *vm, int vcpuid, int errcode) 667{ 668 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode); 669} 670 671static __inline void 672vm_inject_ss(void *vm, int vcpuid, int errcode) 673{ 674 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode); 675} 676 677void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2); 678 679int vm_restart_instruction(void *vm, int vcpuid); 680 681#endif /* _VMM_H_ */ 682