vmm.h revision 347409
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 347409 2019-05-09 22:31:47Z 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); 204vm_paddr_t vmm_sysmem_maxaddr(struct vm *vm); 205void *vm_gpa_hold(struct vm *, int vcpuid, vm_paddr_t gpa, size_t len, 206 int prot, void **cookie); 207void vm_gpa_release(void *cookie); 208bool vm_mem_allocated(struct vm *vm, int vcpuid, vm_paddr_t gpa); 209 210int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval); 211int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val); 212int vm_get_seg_desc(struct vm *vm, int vcpu, int reg, 213 struct seg_desc *ret_desc); 214int vm_set_seg_desc(struct vm *vm, int vcpu, int reg, 215 struct seg_desc *desc); 216int vm_run(struct vm *vm, struct vm_run *vmrun); 217int vm_suspend(struct vm *vm, enum vm_suspend_how how); 218int vm_inject_nmi(struct vm *vm, int vcpu); 219int vm_nmi_pending(struct vm *vm, int vcpuid); 220void vm_nmi_clear(struct vm *vm, int vcpuid); 221int vm_inject_extint(struct vm *vm, int vcpu); 222int vm_extint_pending(struct vm *vm, int vcpuid); 223void vm_extint_clear(struct vm *vm, int vcpuid); 224struct vlapic *vm_lapic(struct vm *vm, int cpu); 225struct vioapic *vm_ioapic(struct vm *vm); 226struct vhpet *vm_hpet(struct vm *vm); 227int vm_get_capability(struct vm *vm, int vcpu, int type, int *val); 228int vm_set_capability(struct vm *vm, int vcpu, int type, int val); 229int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state); 230int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state); 231int vm_apicid2vcpuid(struct vm *vm, int apicid); 232int vm_activate_cpu(struct vm *vm, int vcpu); 233struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid); 234void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip); 235void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip); 236void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip); 237void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip); 238 239#ifdef _SYS__CPUSET_H_ 240/* 241 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'. 242 * The rendezvous 'func(arg)' is not allowed to do anything that will 243 * cause the thread to be put to sleep. 244 * 245 * If the rendezvous is being initiated from a vcpu context then the 246 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1. 247 * 248 * The caller cannot hold any locks when initiating the rendezvous. 249 * 250 * The implementation of this API may cause vcpus other than those specified 251 * by 'dest' to be stalled. The caller should not rely on any vcpus making 252 * forward progress when the rendezvous is in progress. 253 */ 254typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg); 255void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest, 256 vm_rendezvous_func_t func, void *arg); 257cpuset_t vm_active_cpus(struct vm *vm); 258cpuset_t vm_suspended_cpus(struct vm *vm); 259#endif /* _SYS__CPUSET_H_ */ 260 261static __inline int 262vcpu_rendezvous_pending(struct vm_eventinfo *info) 263{ 264 265 return (*((uintptr_t *)(info->rptr)) != 0); 266} 267 268static __inline int 269vcpu_suspended(struct vm_eventinfo *info) 270{ 271 272 return (*info->sptr); 273} 274 275static __inline int 276vcpu_reqidle(struct vm_eventinfo *info) 277{ 278 279 return (*info->iptr); 280} 281 282/* 283 * Return 1 if device indicated by bus/slot/func is supposed to be a 284 * pci passthrough device. 285 * 286 * Return 0 otherwise. 287 */ 288int vmm_is_pptdev(int bus, int slot, int func); 289 290void *vm_iommu_domain(struct vm *vm); 291 292enum vcpu_state { 293 VCPU_IDLE, 294 VCPU_FROZEN, 295 VCPU_RUNNING, 296 VCPU_SLEEPING, 297}; 298 299int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state, 300 bool from_idle); 301enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu); 302 303static int __inline 304vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu) 305{ 306 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING); 307} 308 309#ifdef _SYS_PROC_H_ 310static int __inline 311vcpu_should_yield(struct vm *vm, int vcpu) 312{ 313 314 if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED)) 315 return (1); 316 else if (curthread->td_owepreempt) 317 return (1); 318 else 319 return (0); 320} 321#endif 322 323void *vcpu_stats(struct vm *vm, int vcpu); 324void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr); 325struct vmspace *vm_get_vmspace(struct vm *vm); 326struct vatpic *vm_atpic(struct vm *vm); 327struct vatpit *vm_atpit(struct vm *vm); 328struct vpmtmr *vm_pmtmr(struct vm *vm); 329struct vrtc *vm_rtc(struct vm *vm); 330 331/* 332 * Inject exception 'vector' into the guest vcpu. This function returns 0 on 333 * success and non-zero on failure. 334 * 335 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling 336 * this function directly because they enforce the trap-like or fault-like 337 * behavior of an exception. 338 * 339 * This function should only be called in the context of the thread that is 340 * executing this vcpu. 341 */ 342int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid, 343 uint32_t errcode, int restart_instruction); 344 345/* 346 * This function is called after a VM-exit that occurred during exception or 347 * interrupt delivery through the IDT. The format of 'intinfo' is described 348 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2. 349 * 350 * If a VM-exit handler completes the event delivery successfully then it 351 * should call vm_exit_intinfo() to extinguish the pending event. For e.g., 352 * if the task switch emulation is triggered via a task gate then it should 353 * call this function with 'intinfo=0' to indicate that the external event 354 * is not pending anymore. 355 * 356 * Return value is 0 on success and non-zero on failure. 357 */ 358int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo); 359 360/* 361 * This function is called before every VM-entry to retrieve a pending 362 * event that should be injected into the guest. This function combines 363 * nested events into a double or triple fault. 364 * 365 * Returns 0 if there are no events that need to be injected into the guest 366 * and non-zero otherwise. 367 */ 368int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info); 369 370int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2); 371 372enum vm_reg_name vm_segment_name(int seg_encoding); 373 374struct vm_copyinfo { 375 uint64_t gpa; 376 size_t len; 377 void *hva; 378 void *cookie; 379}; 380 381/* 382 * Set up 'copyinfo[]' to copy to/from guest linear address space starting 383 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for 384 * a copyin or PROT_WRITE for a copyout. 385 * 386 * retval is_fault Interpretation 387 * 0 0 Success 388 * 0 1 An exception was injected into the guest 389 * EFAULT N/A Unrecoverable error 390 * 391 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if 392 * the return value is 0. The 'copyinfo[]' resources should be freed by calling 393 * 'vm_copy_teardown()' after the copy is done. 394 */ 395int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging, 396 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo, 397 int num_copyinfo, int *is_fault); 398void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 399 int num_copyinfo); 400void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 401 void *kaddr, size_t len); 402void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr, 403 struct vm_copyinfo *copyinfo, size_t len); 404 405int vcpu_trace_exceptions(struct vm *vm, int vcpuid); 406#endif /* KERNEL */ 407 408#define VM_MAXCPU 16 /* maximum virtual cpus */ 409 410/* 411 * Identifiers for optional vmm capabilities 412 */ 413enum vm_cap_type { 414 VM_CAP_HALT_EXIT, 415 VM_CAP_MTRAP_EXIT, 416 VM_CAP_PAUSE_EXIT, 417 VM_CAP_UNRESTRICTED_GUEST, 418 VM_CAP_ENABLE_INVPCID, 419 VM_CAP_MAX 420}; 421 422enum vm_intr_trigger { 423 EDGE_TRIGGER, 424 LEVEL_TRIGGER 425}; 426 427/* 428 * The 'access' field has the format specified in Table 21-2 of the Intel 429 * Architecture Manual vol 3b. 430 * 431 * XXX The contents of the 'access' field are architecturally defined except 432 * bit 16 - Segment Unusable. 433 */ 434struct seg_desc { 435 uint64_t base; 436 uint32_t limit; 437 uint32_t access; 438}; 439#define SEG_DESC_TYPE(access) ((access) & 0x001f) 440#define SEG_DESC_DPL(access) (((access) >> 5) & 0x3) 441#define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0) 442#define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0) 443#define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0) 444#define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0) 445 446enum vm_cpu_mode { 447 CPU_MODE_REAL, 448 CPU_MODE_PROTECTED, 449 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 450 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 451}; 452 453enum vm_paging_mode { 454 PAGING_MODE_FLAT, 455 PAGING_MODE_32, 456 PAGING_MODE_PAE, 457 PAGING_MODE_64, 458}; 459 460struct vm_guest_paging { 461 uint64_t cr3; 462 int cpl; 463 enum vm_cpu_mode cpu_mode; 464 enum vm_paging_mode paging_mode; 465}; 466 467/* 468 * The data structures 'vie' and 'vie_op' are meant to be opaque to the 469 * consumers of instruction decoding. The only reason why their contents 470 * need to be exposed is because they are part of the 'vm_exit' structure. 471 */ 472struct vie_op { 473 uint8_t op_byte; /* actual opcode byte */ 474 uint8_t op_type; /* type of operation (e.g. MOV) */ 475 uint16_t op_flags; 476}; 477 478#define VIE_INST_SIZE 15 479struct vie { 480 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */ 481 uint8_t num_valid; /* size of the instruction */ 482 uint8_t num_processed; 483 484 uint8_t addrsize:4, opsize:4; /* address and operand sizes */ 485 uint8_t rex_w:1, /* REX prefix */ 486 rex_r:1, 487 rex_x:1, 488 rex_b:1, 489 rex_present:1, 490 repz_present:1, /* REP/REPE/REPZ prefix */ 491 repnz_present:1, /* REPNE/REPNZ prefix */ 492 opsize_override:1, /* Operand size override */ 493 addrsize_override:1, /* Address size override */ 494 segment_override:1; /* Segment override */ 495 496 uint8_t mod:2, /* ModRM byte */ 497 reg:4, 498 rm:4; 499 500 uint8_t ss:2, /* SIB byte */ 501 index:4, 502 base:4; 503 504 uint8_t disp_bytes; 505 uint8_t imm_bytes; 506 507 uint8_t scale; 508 int base_register; /* VM_REG_GUEST_xyz */ 509 int index_register; /* VM_REG_GUEST_xyz */ 510 int segment_register; /* VM_REG_GUEST_xyz */ 511 512 int64_t displacement; /* optional addr displacement */ 513 int64_t immediate; /* optional immediate operand */ 514 515 uint8_t decoded; /* set to 1 if successfully decoded */ 516 517 struct vie_op op; /* opcode description */ 518}; 519 520enum vm_exitcode { 521 VM_EXITCODE_INOUT, 522 VM_EXITCODE_VMX, 523 VM_EXITCODE_BOGUS, 524 VM_EXITCODE_RDMSR, 525 VM_EXITCODE_WRMSR, 526 VM_EXITCODE_HLT, 527 VM_EXITCODE_MTRAP, 528 VM_EXITCODE_PAUSE, 529 VM_EXITCODE_PAGING, 530 VM_EXITCODE_INST_EMUL, 531 VM_EXITCODE_SPINUP_AP, 532 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */ 533 VM_EXITCODE_RENDEZVOUS, 534 VM_EXITCODE_IOAPIC_EOI, 535 VM_EXITCODE_SUSPENDED, 536 VM_EXITCODE_INOUT_STR, 537 VM_EXITCODE_TASK_SWITCH, 538 VM_EXITCODE_MONITOR, 539 VM_EXITCODE_MWAIT, 540 VM_EXITCODE_SVM, 541 VM_EXITCODE_REQIDLE, 542 VM_EXITCODE_MAX 543}; 544 545struct vm_inout { 546 uint16_t bytes:3; /* 1 or 2 or 4 */ 547 uint16_t in:1; 548 uint16_t string:1; 549 uint16_t rep:1; 550 uint16_t port; 551 uint32_t eax; /* valid for out */ 552}; 553 554struct vm_inout_str { 555 struct vm_inout inout; /* must be the first element */ 556 struct vm_guest_paging paging; 557 uint64_t rflags; 558 uint64_t cr0; 559 uint64_t index; 560 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */ 561 int addrsize; 562 enum vm_reg_name seg_name; 563 struct seg_desc seg_desc; 564}; 565 566enum task_switch_reason { 567 TSR_CALL, 568 TSR_IRET, 569 TSR_JMP, 570 TSR_IDT_GATE, /* task gate in IDT */ 571}; 572 573struct vm_task_switch { 574 uint16_t tsssel; /* new TSS selector */ 575 int ext; /* task switch due to external event */ 576 uint32_t errcode; 577 int errcode_valid; /* push 'errcode' on the new stack */ 578 enum task_switch_reason reason; 579 struct vm_guest_paging paging; 580}; 581 582struct vm_exit { 583 enum vm_exitcode exitcode; 584 int inst_length; /* 0 means unknown */ 585 uint64_t rip; 586 union { 587 struct vm_inout inout; 588 struct vm_inout_str inout_str; 589 struct { 590 uint64_t gpa; 591 int fault_type; 592 } paging; 593 struct { 594 uint64_t gpa; 595 uint64_t gla; 596 uint64_t cs_base; 597 int cs_d; /* CS.D */ 598 struct vm_guest_paging paging; 599 struct vie vie; 600 } inst_emul; 601 /* 602 * VMX specific payload. Used when there is no "better" 603 * exitcode to represent the VM-exit. 604 */ 605 struct { 606 int status; /* vmx inst status */ 607 /* 608 * 'exit_reason' and 'exit_qualification' are valid 609 * only if 'status' is zero. 610 */ 611 uint32_t exit_reason; 612 uint64_t exit_qualification; 613 /* 614 * 'inst_error' and 'inst_type' are valid 615 * only if 'status' is non-zero. 616 */ 617 int inst_type; 618 int inst_error; 619 } vmx; 620 /* 621 * SVM specific payload. 622 */ 623 struct { 624 uint64_t exitcode; 625 uint64_t exitinfo1; 626 uint64_t exitinfo2; 627 } svm; 628 struct { 629 uint32_t code; /* ecx value */ 630 uint64_t wval; 631 } msr; 632 struct { 633 int vcpu; 634 uint64_t rip; 635 } spinup_ap; 636 struct { 637 uint64_t rflags; 638 uint64_t intr_status; 639 } hlt; 640 struct { 641 int vector; 642 } ioapic_eoi; 643 struct { 644 enum vm_suspend_how how; 645 } suspended; 646 struct vm_task_switch task_switch; 647 } u; 648}; 649 650/* APIs to inject faults into the guest */ 651void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid, 652 int errcode); 653 654static __inline void 655vm_inject_ud(void *vm, int vcpuid) 656{ 657 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0); 658} 659 660static __inline void 661vm_inject_gp(void *vm, int vcpuid) 662{ 663 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0); 664} 665 666static __inline void 667vm_inject_ac(void *vm, int vcpuid, int errcode) 668{ 669 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode); 670} 671 672static __inline void 673vm_inject_ss(void *vm, int vcpuid, int errcode) 674{ 675 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode); 676} 677 678void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2); 679 680int vm_restart_instruction(void *vm, int vcpuid); 681 682#endif /* _VMM_H_ */ 683