vmm.h revision 347435
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 347435 2019-05-10 16:32:44Z jhb $ 27 */ 28 29#ifndef _VMM_H_ 30#define _VMM_H_ 31 32#include <sys/sdt.h> 33#include <x86/segments.h> 34 35#ifdef _KERNEL 36SDT_PROVIDER_DECLARE(vmm); 37#endif 38 39enum vm_suspend_how { 40 VM_SUSPEND_NONE, 41 VM_SUSPEND_RESET, 42 VM_SUSPEND_POWEROFF, 43 VM_SUSPEND_HALT, 44 VM_SUSPEND_TRIPLEFAULT, 45 VM_SUSPEND_LAST 46}; 47 48/* 49 * Identifiers for architecturally defined registers. 50 */ 51enum vm_reg_name { 52 VM_REG_GUEST_RAX, 53 VM_REG_GUEST_RBX, 54 VM_REG_GUEST_RCX, 55 VM_REG_GUEST_RDX, 56 VM_REG_GUEST_RSI, 57 VM_REG_GUEST_RDI, 58 VM_REG_GUEST_RBP, 59 VM_REG_GUEST_R8, 60 VM_REG_GUEST_R9, 61 VM_REG_GUEST_R10, 62 VM_REG_GUEST_R11, 63 VM_REG_GUEST_R12, 64 VM_REG_GUEST_R13, 65 VM_REG_GUEST_R14, 66 VM_REG_GUEST_R15, 67 VM_REG_GUEST_CR0, 68 VM_REG_GUEST_CR3, 69 VM_REG_GUEST_CR4, 70 VM_REG_GUEST_DR7, 71 VM_REG_GUEST_RSP, 72 VM_REG_GUEST_RIP, 73 VM_REG_GUEST_RFLAGS, 74 VM_REG_GUEST_ES, 75 VM_REG_GUEST_CS, 76 VM_REG_GUEST_SS, 77 VM_REG_GUEST_DS, 78 VM_REG_GUEST_FS, 79 VM_REG_GUEST_GS, 80 VM_REG_GUEST_LDTR, 81 VM_REG_GUEST_TR, 82 VM_REG_GUEST_IDTR, 83 VM_REG_GUEST_GDTR, 84 VM_REG_GUEST_EFER, 85 VM_REG_GUEST_CR2, 86 VM_REG_GUEST_PDPTE0, 87 VM_REG_GUEST_PDPTE1, 88 VM_REG_GUEST_PDPTE2, 89 VM_REG_GUEST_PDPTE3, 90 VM_REG_GUEST_INTR_SHADOW, 91 VM_REG_GUEST_DR0, 92 VM_REG_GUEST_DR1, 93 VM_REG_GUEST_DR2, 94 VM_REG_GUEST_DR3, 95 VM_REG_GUEST_DR6, 96 VM_REG_LAST 97}; 98 99enum x2apic_state { 100 X2APIC_DISABLED, 101 X2APIC_ENABLED, 102 X2APIC_STATE_LAST 103}; 104 105#define VM_INTINFO_VECTOR(info) ((info) & 0xff) 106#define VM_INTINFO_DEL_ERRCODE 0x800 107#define VM_INTINFO_RSVD 0x7ffff000 108#define VM_INTINFO_VALID 0x80000000 109#define VM_INTINFO_TYPE 0x700 110#define VM_INTINFO_HWINTR (0 << 8) 111#define VM_INTINFO_NMI (2 << 8) 112#define VM_INTINFO_HWEXCEPTION (3 << 8) 113#define VM_INTINFO_SWINTR (4 << 8) 114 115#ifdef _KERNEL 116 117#define VM_MAX_NAMELEN 32 118 119struct vm; 120struct vm_exception; 121struct seg_desc; 122struct vm_exit; 123struct vm_run; 124struct vhpet; 125struct vioapic; 126struct vlapic; 127struct vmspace; 128struct vm_object; 129struct vm_guest_paging; 130struct pmap; 131 132struct vm_eventinfo { 133 void *rptr; /* rendezvous cookie */ 134 int *sptr; /* suspend cookie */ 135 int *iptr; /* reqidle cookie */ 136}; 137 138typedef int (*vmm_init_func_t)(int ipinum); 139typedef int (*vmm_cleanup_func_t)(void); 140typedef void (*vmm_resume_func_t)(void); 141typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap); 142typedef int (*vmi_run_func_t)(void *vmi, int vcpu, register_t rip, 143 struct pmap *pmap, struct vm_eventinfo *info); 144typedef void (*vmi_cleanup_func_t)(void *vmi); 145typedef int (*vmi_get_register_t)(void *vmi, int vcpu, int num, 146 uint64_t *retval); 147typedef int (*vmi_set_register_t)(void *vmi, int vcpu, int num, 148 uint64_t val); 149typedef int (*vmi_get_desc_t)(void *vmi, int vcpu, int num, 150 struct seg_desc *desc); 151typedef int (*vmi_set_desc_t)(void *vmi, int vcpu, int num, 152 struct seg_desc *desc); 153typedef int (*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval); 154typedef int (*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val); 155typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max); 156typedef void (*vmi_vmspace_free)(struct vmspace *vmspace); 157typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu); 158typedef void (*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic); 159 160struct vmm_ops { 161 vmm_init_func_t init; /* module wide initialization */ 162 vmm_cleanup_func_t cleanup; 163 vmm_resume_func_t resume; 164 165 vmi_init_func_t vminit; /* vm-specific initialization */ 166 vmi_run_func_t vmrun; 167 vmi_cleanup_func_t vmcleanup; 168 vmi_get_register_t vmgetreg; 169 vmi_set_register_t vmsetreg; 170 vmi_get_desc_t vmgetdesc; 171 vmi_set_desc_t vmsetdesc; 172 vmi_get_cap_t vmgetcap; 173 vmi_set_cap_t vmsetcap; 174 vmi_vmspace_alloc vmspace_alloc; 175 vmi_vmspace_free vmspace_free; 176 vmi_vlapic_init vlapic_init; 177 vmi_vlapic_cleanup vlapic_cleanup; 178}; 179 180extern struct vmm_ops vmm_ops_intel; 181extern struct vmm_ops vmm_ops_amd; 182 183int vm_create(const char *name, struct vm **retvm); 184void vm_destroy(struct vm *vm); 185int vm_reinit(struct vm *vm); 186const char *vm_name(struct vm *vm); 187 188/* 189 * APIs that modify the guest memory map require all vcpus to be frozen. 190 */ 191int vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t off, 192 size_t len, int prot, int flags); 193int vm_alloc_memseg(struct vm *vm, int ident, size_t len, bool sysmem); 194void vm_free_memseg(struct vm *vm, int ident); 195int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa); 196int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len); 197int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func); 198int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func); 199 200/* 201 * APIs that inspect the guest memory map require only a *single* vcpu to 202 * be frozen. This acts like a read lock on the guest memory map since any 203 * modification requires *all* vcpus to be frozen. 204 */ 205int vm_mmap_getnext(struct vm *vm, vm_paddr_t *gpa, int *segid, 206 vm_ooffset_t *segoff, size_t *len, int *prot, int *flags); 207int vm_get_memseg(struct vm *vm, int ident, size_t *len, bool *sysmem, 208 struct vm_object **objptr); 209vm_paddr_t vmm_sysmem_maxaddr(struct vm *vm); 210void *vm_gpa_hold(struct vm *, int vcpuid, vm_paddr_t gpa, size_t len, 211 int prot, void **cookie); 212void vm_gpa_release(void *cookie); 213bool vm_mem_allocated(struct vm *vm, int vcpuid, vm_paddr_t gpa); 214 215int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval); 216int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val); 217int vm_get_seg_desc(struct vm *vm, int vcpu, int reg, 218 struct seg_desc *ret_desc); 219int vm_set_seg_desc(struct vm *vm, int vcpu, int reg, 220 struct seg_desc *desc); 221int vm_run(struct vm *vm, struct vm_run *vmrun); 222int vm_suspend(struct vm *vm, enum vm_suspend_how how); 223int vm_inject_nmi(struct vm *vm, int vcpu); 224int vm_nmi_pending(struct vm *vm, int vcpuid); 225void vm_nmi_clear(struct vm *vm, int vcpuid); 226int vm_inject_extint(struct vm *vm, int vcpu); 227int vm_extint_pending(struct vm *vm, int vcpuid); 228void vm_extint_clear(struct vm *vm, int vcpuid); 229struct vlapic *vm_lapic(struct vm *vm, int cpu); 230struct vioapic *vm_ioapic(struct vm *vm); 231struct vhpet *vm_hpet(struct vm *vm); 232int vm_get_capability(struct vm *vm, int vcpu, int type, int *val); 233int vm_set_capability(struct vm *vm, int vcpu, int type, int val); 234int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state); 235int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state); 236int vm_apicid2vcpuid(struct vm *vm, int apicid); 237int vm_activate_cpu(struct vm *vm, int vcpu); 238struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid); 239void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip); 240void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip); 241void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip); 242void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip); 243 244#ifdef _SYS__CPUSET_H_ 245/* 246 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'. 247 * The rendezvous 'func(arg)' is not allowed to do anything that will 248 * cause the thread to be put to sleep. 249 * 250 * If the rendezvous is being initiated from a vcpu context then the 251 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1. 252 * 253 * The caller cannot hold any locks when initiating the rendezvous. 254 * 255 * The implementation of this API may cause vcpus other than those specified 256 * by 'dest' to be stalled. The caller should not rely on any vcpus making 257 * forward progress when the rendezvous is in progress. 258 */ 259typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg); 260void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest, 261 vm_rendezvous_func_t func, void *arg); 262cpuset_t vm_active_cpus(struct vm *vm); 263cpuset_t vm_suspended_cpus(struct vm *vm); 264#endif /* _SYS__CPUSET_H_ */ 265 266static __inline int 267vcpu_rendezvous_pending(struct vm_eventinfo *info) 268{ 269 270 return (*((uintptr_t *)(info->rptr)) != 0); 271} 272 273static __inline int 274vcpu_suspended(struct vm_eventinfo *info) 275{ 276 277 return (*info->sptr); 278} 279 280static __inline int 281vcpu_reqidle(struct vm_eventinfo *info) 282{ 283 284 return (*info->iptr); 285} 286 287/* 288 * Return 1 if device indicated by bus/slot/func is supposed to be a 289 * pci passthrough device. 290 * 291 * Return 0 otherwise. 292 */ 293int vmm_is_pptdev(int bus, int slot, int func); 294 295void *vm_iommu_domain(struct vm *vm); 296 297enum vcpu_state { 298 VCPU_IDLE, 299 VCPU_FROZEN, 300 VCPU_RUNNING, 301 VCPU_SLEEPING, 302}; 303 304int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state, 305 bool from_idle); 306enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu); 307 308static int __inline 309vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu) 310{ 311 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING); 312} 313 314#ifdef _SYS_PROC_H_ 315static int __inline 316vcpu_should_yield(struct vm *vm, int vcpu) 317{ 318 319 if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED)) 320 return (1); 321 else if (curthread->td_owepreempt) 322 return (1); 323 else 324 return (0); 325} 326#endif 327 328void *vcpu_stats(struct vm *vm, int vcpu); 329void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr); 330struct vmspace *vm_get_vmspace(struct vm *vm); 331struct vatpic *vm_atpic(struct vm *vm); 332struct vatpit *vm_atpit(struct vm *vm); 333struct vpmtmr *vm_pmtmr(struct vm *vm); 334struct vrtc *vm_rtc(struct vm *vm); 335 336/* 337 * Inject exception 'vector' into the guest vcpu. This function returns 0 on 338 * success and non-zero on failure. 339 * 340 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling 341 * this function directly because they enforce the trap-like or fault-like 342 * behavior of an exception. 343 * 344 * This function should only be called in the context of the thread that is 345 * executing this vcpu. 346 */ 347int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid, 348 uint32_t errcode, int restart_instruction); 349 350/* 351 * This function is called after a VM-exit that occurred during exception or 352 * interrupt delivery through the IDT. The format of 'intinfo' is described 353 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2. 354 * 355 * If a VM-exit handler completes the event delivery successfully then it 356 * should call vm_exit_intinfo() to extinguish the pending event. For e.g., 357 * if the task switch emulation is triggered via a task gate then it should 358 * call this function with 'intinfo=0' to indicate that the external event 359 * is not pending anymore. 360 * 361 * Return value is 0 on success and non-zero on failure. 362 */ 363int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo); 364 365/* 366 * This function is called before every VM-entry to retrieve a pending 367 * event that should be injected into the guest. This function combines 368 * nested events into a double or triple fault. 369 * 370 * Returns 0 if there are no events that need to be injected into the guest 371 * and non-zero otherwise. 372 */ 373int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info); 374 375int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2); 376 377enum vm_reg_name vm_segment_name(int seg_encoding); 378 379struct vm_copyinfo { 380 uint64_t gpa; 381 size_t len; 382 void *hva; 383 void *cookie; 384}; 385 386/* 387 * Set up 'copyinfo[]' to copy to/from guest linear address space starting 388 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for 389 * a copyin or PROT_WRITE for a copyout. 390 * 391 * retval is_fault Interpretation 392 * 0 0 Success 393 * 0 1 An exception was injected into the guest 394 * EFAULT N/A Unrecoverable error 395 * 396 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if 397 * the return value is 0. The 'copyinfo[]' resources should be freed by calling 398 * 'vm_copy_teardown()' after the copy is done. 399 */ 400int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging, 401 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo, 402 int num_copyinfo, int *is_fault); 403void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 404 int num_copyinfo); 405void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 406 void *kaddr, size_t len); 407void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr, 408 struct vm_copyinfo *copyinfo, size_t len); 409 410int vcpu_trace_exceptions(struct vm *vm, int vcpuid); 411#endif /* KERNEL */ 412 413#define VM_MAXCPU 16 /* maximum virtual cpus */ 414 415/* 416 * Identifiers for optional vmm capabilities 417 */ 418enum vm_cap_type { 419 VM_CAP_HALT_EXIT, 420 VM_CAP_MTRAP_EXIT, 421 VM_CAP_PAUSE_EXIT, 422 VM_CAP_UNRESTRICTED_GUEST, 423 VM_CAP_ENABLE_INVPCID, 424 VM_CAP_MAX 425}; 426 427enum vm_intr_trigger { 428 EDGE_TRIGGER, 429 LEVEL_TRIGGER 430}; 431 432/* 433 * The 'access' field has the format specified in Table 21-2 of the Intel 434 * Architecture Manual vol 3b. 435 * 436 * XXX The contents of the 'access' field are architecturally defined except 437 * bit 16 - Segment Unusable. 438 */ 439struct seg_desc { 440 uint64_t base; 441 uint32_t limit; 442 uint32_t access; 443}; 444#define SEG_DESC_TYPE(access) ((access) & 0x001f) 445#define SEG_DESC_DPL(access) (((access) >> 5) & 0x3) 446#define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0) 447#define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0) 448#define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0) 449#define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0) 450 451enum vm_cpu_mode { 452 CPU_MODE_REAL, 453 CPU_MODE_PROTECTED, 454 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 455 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 456}; 457 458enum vm_paging_mode { 459 PAGING_MODE_FLAT, 460 PAGING_MODE_32, 461 PAGING_MODE_PAE, 462 PAGING_MODE_64, 463}; 464 465struct vm_guest_paging { 466 uint64_t cr3; 467 int cpl; 468 enum vm_cpu_mode cpu_mode; 469 enum vm_paging_mode paging_mode; 470}; 471 472/* 473 * The data structures 'vie' and 'vie_op' are meant to be opaque to the 474 * consumers of instruction decoding. The only reason why their contents 475 * need to be exposed is because they are part of the 'vm_exit' structure. 476 */ 477struct vie_op { 478 uint8_t op_byte; /* actual opcode byte */ 479 uint8_t op_type; /* type of operation (e.g. MOV) */ 480 uint16_t op_flags; 481}; 482 483#define VIE_INST_SIZE 15 484struct vie { 485 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */ 486 uint8_t num_valid; /* size of the instruction */ 487 uint8_t num_processed; 488 489 uint8_t addrsize:4, opsize:4; /* address and operand sizes */ 490 uint8_t rex_w:1, /* REX prefix */ 491 rex_r:1, 492 rex_x:1, 493 rex_b:1, 494 rex_present:1, 495 repz_present:1, /* REP/REPE/REPZ prefix */ 496 repnz_present:1, /* REPNE/REPNZ prefix */ 497 opsize_override:1, /* Operand size override */ 498 addrsize_override:1, /* Address size override */ 499 segment_override:1; /* Segment override */ 500 501 uint8_t mod:2, /* ModRM byte */ 502 reg:4, 503 rm:4; 504 505 uint8_t ss:2, /* SIB byte */ 506 index:4, 507 base:4; 508 509 uint8_t disp_bytes; 510 uint8_t imm_bytes; 511 512 uint8_t scale; 513 int base_register; /* VM_REG_GUEST_xyz */ 514 int index_register; /* VM_REG_GUEST_xyz */ 515 int segment_register; /* VM_REG_GUEST_xyz */ 516 517 int64_t displacement; /* optional addr displacement */ 518 int64_t immediate; /* optional immediate operand */ 519 520 uint8_t decoded; /* set to 1 if successfully decoded */ 521 522 struct vie_op op; /* opcode description */ 523}; 524 525enum vm_exitcode { 526 VM_EXITCODE_INOUT, 527 VM_EXITCODE_VMX, 528 VM_EXITCODE_BOGUS, 529 VM_EXITCODE_RDMSR, 530 VM_EXITCODE_WRMSR, 531 VM_EXITCODE_HLT, 532 VM_EXITCODE_MTRAP, 533 VM_EXITCODE_PAUSE, 534 VM_EXITCODE_PAGING, 535 VM_EXITCODE_INST_EMUL, 536 VM_EXITCODE_SPINUP_AP, 537 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */ 538 VM_EXITCODE_RENDEZVOUS, 539 VM_EXITCODE_IOAPIC_EOI, 540 VM_EXITCODE_SUSPENDED, 541 VM_EXITCODE_INOUT_STR, 542 VM_EXITCODE_TASK_SWITCH, 543 VM_EXITCODE_MONITOR, 544 VM_EXITCODE_MWAIT, 545 VM_EXITCODE_SVM, 546 VM_EXITCODE_REQIDLE, 547 VM_EXITCODE_VMINSN, 548 VM_EXITCODE_MAX 549}; 550 551struct vm_inout { 552 uint16_t bytes:3; /* 1 or 2 or 4 */ 553 uint16_t in:1; 554 uint16_t string:1; 555 uint16_t rep:1; 556 uint16_t port; 557 uint32_t eax; /* valid for out */ 558}; 559 560struct vm_inout_str { 561 struct vm_inout inout; /* must be the first element */ 562 struct vm_guest_paging paging; 563 uint64_t rflags; 564 uint64_t cr0; 565 uint64_t index; 566 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */ 567 int addrsize; 568 enum vm_reg_name seg_name; 569 struct seg_desc seg_desc; 570}; 571 572enum task_switch_reason { 573 TSR_CALL, 574 TSR_IRET, 575 TSR_JMP, 576 TSR_IDT_GATE, /* task gate in IDT */ 577}; 578 579struct vm_task_switch { 580 uint16_t tsssel; /* new TSS selector */ 581 int ext; /* task switch due to external event */ 582 uint32_t errcode; 583 int errcode_valid; /* push 'errcode' on the new stack */ 584 enum task_switch_reason reason; 585 struct vm_guest_paging paging; 586}; 587 588struct vm_exit { 589 enum vm_exitcode exitcode; 590 int inst_length; /* 0 means unknown */ 591 uint64_t rip; 592 union { 593 struct vm_inout inout; 594 struct vm_inout_str inout_str; 595 struct { 596 uint64_t gpa; 597 int fault_type; 598 } paging; 599 struct { 600 uint64_t gpa; 601 uint64_t gla; 602 uint64_t cs_base; 603 int cs_d; /* CS.D */ 604 struct vm_guest_paging paging; 605 struct vie vie; 606 } inst_emul; 607 /* 608 * VMX specific payload. Used when there is no "better" 609 * exitcode to represent the VM-exit. 610 */ 611 struct { 612 int status; /* vmx inst status */ 613 /* 614 * 'exit_reason' and 'exit_qualification' are valid 615 * only if 'status' is zero. 616 */ 617 uint32_t exit_reason; 618 uint64_t exit_qualification; 619 /* 620 * 'inst_error' and 'inst_type' are valid 621 * only if 'status' is non-zero. 622 */ 623 int inst_type; 624 int inst_error; 625 } vmx; 626 /* 627 * SVM specific payload. 628 */ 629 struct { 630 uint64_t exitcode; 631 uint64_t exitinfo1; 632 uint64_t exitinfo2; 633 } svm; 634 struct { 635 uint32_t code; /* ecx value */ 636 uint64_t wval; 637 } msr; 638 struct { 639 int vcpu; 640 uint64_t rip; 641 } spinup_ap; 642 struct { 643 uint64_t rflags; 644 uint64_t intr_status; 645 } hlt; 646 struct { 647 int vector; 648 } ioapic_eoi; 649 struct { 650 enum vm_suspend_how how; 651 } suspended; 652 struct vm_task_switch task_switch; 653 } u; 654}; 655 656/* APIs to inject faults into the guest */ 657void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid, 658 int errcode); 659 660static __inline void 661vm_inject_ud(void *vm, int vcpuid) 662{ 663 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0); 664} 665 666static __inline void 667vm_inject_gp(void *vm, int vcpuid) 668{ 669 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0); 670} 671 672static __inline void 673vm_inject_ac(void *vm, int vcpuid, int errcode) 674{ 675 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode); 676} 677 678static __inline void 679vm_inject_ss(void *vm, int vcpuid, int errcode) 680{ 681 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode); 682} 683 684void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2); 685 686int vm_restart_instruction(void *vm, int vcpuid); 687 688#endif /* _VMM_H_ */ 689