vmm.h revision 348201
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 348201 2019-05-23 21:23:18Z rgrimes $ 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); 187void vm_get_topology(struct vm *vm, uint16_t *sockets, uint16_t *cores, 188 uint16_t *threads, uint16_t *maxcpus); 189int vm_set_topology(struct vm *vm, uint16_t sockets, uint16_t cores, 190 uint16_t threads, uint16_t maxcpus); 191 192/* 193 * APIs that modify the guest memory map require all vcpus to be frozen. 194 */ 195int vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t off, 196 size_t len, int prot, int flags); 197int vm_alloc_memseg(struct vm *vm, int ident, size_t len, bool sysmem); 198void vm_free_memseg(struct vm *vm, int ident); 199int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa); 200int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len); 201int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func); 202int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func); 203 204/* 205 * APIs that inspect the guest memory map require only a *single* vcpu to 206 * be frozen. This acts like a read lock on the guest memory map since any 207 * modification requires *all* vcpus to be frozen. 208 */ 209int vm_mmap_getnext(struct vm *vm, vm_paddr_t *gpa, int *segid, 210 vm_ooffset_t *segoff, size_t *len, int *prot, int *flags); 211int vm_get_memseg(struct vm *vm, int ident, size_t *len, bool *sysmem, 212 struct vm_object **objptr); 213vm_paddr_t vmm_sysmem_maxaddr(struct vm *vm); 214void *vm_gpa_hold(struct vm *, int vcpuid, vm_paddr_t gpa, size_t len, 215 int prot, void **cookie); 216void vm_gpa_release(void *cookie); 217bool vm_mem_allocated(struct vm *vm, int vcpuid, vm_paddr_t gpa); 218 219int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval); 220int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val); 221int vm_get_seg_desc(struct vm *vm, int vcpu, int reg, 222 struct seg_desc *ret_desc); 223int vm_set_seg_desc(struct vm *vm, int vcpu, int reg, 224 struct seg_desc *desc); 225int vm_run(struct vm *vm, struct vm_run *vmrun); 226int vm_suspend(struct vm *vm, enum vm_suspend_how how); 227int vm_inject_nmi(struct vm *vm, int vcpu); 228int vm_nmi_pending(struct vm *vm, int vcpuid); 229void vm_nmi_clear(struct vm *vm, int vcpuid); 230int vm_inject_extint(struct vm *vm, int vcpu); 231int vm_extint_pending(struct vm *vm, int vcpuid); 232void vm_extint_clear(struct vm *vm, int vcpuid); 233struct vlapic *vm_lapic(struct vm *vm, int cpu); 234struct vioapic *vm_ioapic(struct vm *vm); 235struct vhpet *vm_hpet(struct vm *vm); 236int vm_get_capability(struct vm *vm, int vcpu, int type, int *val); 237int vm_set_capability(struct vm *vm, int vcpu, int type, int val); 238int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state); 239int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state); 240int vm_apicid2vcpuid(struct vm *vm, int apicid); 241int vm_activate_cpu(struct vm *vm, int vcpu); 242struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid); 243void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip); 244void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip); 245void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip); 246void vm_exit_reqidle(struct vm *vm, int vcpuid, uint64_t rip); 247 248#ifdef _SYS__CPUSET_H_ 249/* 250 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'. 251 * The rendezvous 'func(arg)' is not allowed to do anything that will 252 * cause the thread to be put to sleep. 253 * 254 * If the rendezvous is being initiated from a vcpu context then the 255 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1. 256 * 257 * The caller cannot hold any locks when initiating the rendezvous. 258 * 259 * The implementation of this API may cause vcpus other than those specified 260 * by 'dest' to be stalled. The caller should not rely on any vcpus making 261 * forward progress when the rendezvous is in progress. 262 */ 263typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg); 264void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest, 265 vm_rendezvous_func_t func, void *arg); 266cpuset_t vm_active_cpus(struct vm *vm); 267cpuset_t vm_suspended_cpus(struct vm *vm); 268#endif /* _SYS__CPUSET_H_ */ 269 270static __inline int 271vcpu_rendezvous_pending(struct vm_eventinfo *info) 272{ 273 274 return (*((uintptr_t *)(info->rptr)) != 0); 275} 276 277static __inline int 278vcpu_suspended(struct vm_eventinfo *info) 279{ 280 281 return (*info->sptr); 282} 283 284static __inline int 285vcpu_reqidle(struct vm_eventinfo *info) 286{ 287 288 return (*info->iptr); 289} 290 291/* 292 * Return 1 if device indicated by bus/slot/func is supposed to be a 293 * pci passthrough device. 294 * 295 * Return 0 otherwise. 296 */ 297int vmm_is_pptdev(int bus, int slot, int func); 298 299void *vm_iommu_domain(struct vm *vm); 300 301enum vcpu_state { 302 VCPU_IDLE, 303 VCPU_FROZEN, 304 VCPU_RUNNING, 305 VCPU_SLEEPING, 306}; 307 308int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state, 309 bool from_idle); 310enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu); 311 312static int __inline 313vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu) 314{ 315 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING); 316} 317 318#ifdef _SYS_PROC_H_ 319static int __inline 320vcpu_should_yield(struct vm *vm, int vcpu) 321{ 322 323 if (curthread->td_flags & (TDF_ASTPENDING | TDF_NEEDRESCHED)) 324 return (1); 325 else if (curthread->td_owepreempt) 326 return (1); 327 else 328 return (0); 329} 330#endif 331 332void *vcpu_stats(struct vm *vm, int vcpu); 333void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr); 334struct vmspace *vm_get_vmspace(struct vm *vm); 335struct vatpic *vm_atpic(struct vm *vm); 336struct vatpit *vm_atpit(struct vm *vm); 337struct vpmtmr *vm_pmtmr(struct vm *vm); 338struct vrtc *vm_rtc(struct vm *vm); 339 340/* 341 * Inject exception 'vector' into the guest vcpu. This function returns 0 on 342 * success and non-zero on failure. 343 * 344 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling 345 * this function directly because they enforce the trap-like or fault-like 346 * behavior of an exception. 347 * 348 * This function should only be called in the context of the thread that is 349 * executing this vcpu. 350 */ 351int vm_inject_exception(struct vm *vm, int vcpuid, int vector, int err_valid, 352 uint32_t errcode, int restart_instruction); 353 354/* 355 * This function is called after a VM-exit that occurred during exception or 356 * interrupt delivery through the IDT. The format of 'intinfo' is described 357 * in Figure 15-1, "EXITINTINFO for All Intercepts", APM, Vol 2. 358 * 359 * If a VM-exit handler completes the event delivery successfully then it 360 * should call vm_exit_intinfo() to extinguish the pending event. For e.g., 361 * if the task switch emulation is triggered via a task gate then it should 362 * call this function with 'intinfo=0' to indicate that the external event 363 * is not pending anymore. 364 * 365 * Return value is 0 on success and non-zero on failure. 366 */ 367int vm_exit_intinfo(struct vm *vm, int vcpuid, uint64_t intinfo); 368 369/* 370 * This function is called before every VM-entry to retrieve a pending 371 * event that should be injected into the guest. This function combines 372 * nested events into a double or triple fault. 373 * 374 * Returns 0 if there are no events that need to be injected into the guest 375 * and non-zero otherwise. 376 */ 377int vm_entry_intinfo(struct vm *vm, int vcpuid, uint64_t *info); 378 379int vm_get_intinfo(struct vm *vm, int vcpuid, uint64_t *info1, uint64_t *info2); 380 381enum vm_reg_name vm_segment_name(int seg_encoding); 382 383struct vm_copyinfo { 384 uint64_t gpa; 385 size_t len; 386 void *hva; 387 void *cookie; 388}; 389 390/* 391 * Set up 'copyinfo[]' to copy to/from guest linear address space starting 392 * at 'gla' and 'len' bytes long. The 'prot' should be set to PROT_READ for 393 * a copyin or PROT_WRITE for a copyout. 394 * 395 * retval is_fault Interpretation 396 * 0 0 Success 397 * 0 1 An exception was injected into the guest 398 * EFAULT N/A Unrecoverable error 399 * 400 * The 'copyinfo[]' can be passed to 'vm_copyin()' or 'vm_copyout()' only if 401 * the return value is 0. The 'copyinfo[]' resources should be freed by calling 402 * 'vm_copy_teardown()' after the copy is done. 403 */ 404int vm_copy_setup(struct vm *vm, int vcpuid, struct vm_guest_paging *paging, 405 uint64_t gla, size_t len, int prot, struct vm_copyinfo *copyinfo, 406 int num_copyinfo, int *is_fault); 407void vm_copy_teardown(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 408 int num_copyinfo); 409void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, 410 void *kaddr, size_t len); 411void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr, 412 struct vm_copyinfo *copyinfo, size_t len); 413 414int vcpu_trace_exceptions(struct vm *vm, int vcpuid); 415#endif /* KERNEL */ 416 417#define VM_MAXCPU 16 /* maximum virtual cpus */ 418 419/* 420 * Identifiers for optional vmm capabilities 421 */ 422enum vm_cap_type { 423 VM_CAP_HALT_EXIT, 424 VM_CAP_MTRAP_EXIT, 425 VM_CAP_PAUSE_EXIT, 426 VM_CAP_UNRESTRICTED_GUEST, 427 VM_CAP_ENABLE_INVPCID, 428 VM_CAP_MAX 429}; 430 431enum vm_intr_trigger { 432 EDGE_TRIGGER, 433 LEVEL_TRIGGER 434}; 435 436/* 437 * The 'access' field has the format specified in Table 21-2 of the Intel 438 * Architecture Manual vol 3b. 439 * 440 * XXX The contents of the 'access' field are architecturally defined except 441 * bit 16 - Segment Unusable. 442 */ 443struct seg_desc { 444 uint64_t base; 445 uint32_t limit; 446 uint32_t access; 447}; 448#define SEG_DESC_TYPE(access) ((access) & 0x001f) 449#define SEG_DESC_DPL(access) (((access) >> 5) & 0x3) 450#define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0) 451#define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0) 452#define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0) 453#define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0) 454 455enum vm_cpu_mode { 456 CPU_MODE_REAL, 457 CPU_MODE_PROTECTED, 458 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 459 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 460}; 461 462enum vm_paging_mode { 463 PAGING_MODE_FLAT, 464 PAGING_MODE_32, 465 PAGING_MODE_PAE, 466 PAGING_MODE_64, 467}; 468 469struct vm_guest_paging { 470 uint64_t cr3; 471 int cpl; 472 enum vm_cpu_mode cpu_mode; 473 enum vm_paging_mode paging_mode; 474}; 475 476/* 477 * The data structures 'vie' and 'vie_op' are meant to be opaque to the 478 * consumers of instruction decoding. The only reason why their contents 479 * need to be exposed is because they are part of the 'vm_exit' structure. 480 */ 481struct vie_op { 482 uint8_t op_byte; /* actual opcode byte */ 483 uint8_t op_type; /* type of operation (e.g. MOV) */ 484 uint16_t op_flags; 485}; 486 487#define VIE_INST_SIZE 15 488struct vie { 489 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */ 490 uint8_t num_valid; /* size of the instruction */ 491 uint8_t num_processed; 492 493 uint8_t addrsize:4, opsize:4; /* address and operand sizes */ 494 uint8_t rex_w:1, /* REX prefix */ 495 rex_r:1, 496 rex_x:1, 497 rex_b:1, 498 rex_present:1, 499 repz_present:1, /* REP/REPE/REPZ prefix */ 500 repnz_present:1, /* REPNE/REPNZ prefix */ 501 opsize_override:1, /* Operand size override */ 502 addrsize_override:1, /* Address size override */ 503 segment_override:1; /* Segment override */ 504 505 uint8_t mod:2, /* ModRM byte */ 506 reg:4, 507 rm:4; 508 509 uint8_t ss:2, /* SIB byte */ 510 index:4, 511 base:4; 512 513 uint8_t disp_bytes; 514 uint8_t imm_bytes; 515 516 uint8_t scale; 517 int base_register; /* VM_REG_GUEST_xyz */ 518 int index_register; /* VM_REG_GUEST_xyz */ 519 int segment_register; /* VM_REG_GUEST_xyz */ 520 521 int64_t displacement; /* optional addr displacement */ 522 int64_t immediate; /* optional immediate operand */ 523 524 uint8_t decoded; /* set to 1 if successfully decoded */ 525 526 struct vie_op op; /* opcode description */ 527}; 528 529enum vm_exitcode { 530 VM_EXITCODE_INOUT, 531 VM_EXITCODE_VMX, 532 VM_EXITCODE_BOGUS, 533 VM_EXITCODE_RDMSR, 534 VM_EXITCODE_WRMSR, 535 VM_EXITCODE_HLT, 536 VM_EXITCODE_MTRAP, 537 VM_EXITCODE_PAUSE, 538 VM_EXITCODE_PAGING, 539 VM_EXITCODE_INST_EMUL, 540 VM_EXITCODE_SPINUP_AP, 541 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */ 542 VM_EXITCODE_RENDEZVOUS, 543 VM_EXITCODE_IOAPIC_EOI, 544 VM_EXITCODE_SUSPENDED, 545 VM_EXITCODE_INOUT_STR, 546 VM_EXITCODE_TASK_SWITCH, 547 VM_EXITCODE_MONITOR, 548 VM_EXITCODE_MWAIT, 549 VM_EXITCODE_SVM, 550 VM_EXITCODE_REQIDLE, 551 VM_EXITCODE_VMINSN, 552 VM_EXITCODE_MAX 553}; 554 555struct vm_inout { 556 uint16_t bytes:3; /* 1 or 2 or 4 */ 557 uint16_t in:1; 558 uint16_t string:1; 559 uint16_t rep:1; 560 uint16_t port; 561 uint32_t eax; /* valid for out */ 562}; 563 564struct vm_inout_str { 565 struct vm_inout inout; /* must be the first element */ 566 struct vm_guest_paging paging; 567 uint64_t rflags; 568 uint64_t cr0; 569 uint64_t index; 570 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */ 571 int addrsize; 572 enum vm_reg_name seg_name; 573 struct seg_desc seg_desc; 574}; 575 576enum task_switch_reason { 577 TSR_CALL, 578 TSR_IRET, 579 TSR_JMP, 580 TSR_IDT_GATE, /* task gate in IDT */ 581}; 582 583struct vm_task_switch { 584 uint16_t tsssel; /* new TSS selector */ 585 int ext; /* task switch due to external event */ 586 uint32_t errcode; 587 int errcode_valid; /* push 'errcode' on the new stack */ 588 enum task_switch_reason reason; 589 struct vm_guest_paging paging; 590}; 591 592struct vm_exit { 593 enum vm_exitcode exitcode; 594 int inst_length; /* 0 means unknown */ 595 uint64_t rip; 596 union { 597 struct vm_inout inout; 598 struct vm_inout_str inout_str; 599 struct { 600 uint64_t gpa; 601 int fault_type; 602 } paging; 603 struct { 604 uint64_t gpa; 605 uint64_t gla; 606 uint64_t cs_base; 607 int cs_d; /* CS.D */ 608 struct vm_guest_paging paging; 609 struct vie vie; 610 } inst_emul; 611 /* 612 * VMX specific payload. Used when there is no "better" 613 * exitcode to represent the VM-exit. 614 */ 615 struct { 616 int status; /* vmx inst status */ 617 /* 618 * 'exit_reason' and 'exit_qualification' are valid 619 * only if 'status' is zero. 620 */ 621 uint32_t exit_reason; 622 uint64_t exit_qualification; 623 /* 624 * 'inst_error' and 'inst_type' are valid 625 * only if 'status' is non-zero. 626 */ 627 int inst_type; 628 int inst_error; 629 } vmx; 630 /* 631 * SVM specific payload. 632 */ 633 struct { 634 uint64_t exitcode; 635 uint64_t exitinfo1; 636 uint64_t exitinfo2; 637 } svm; 638 struct { 639 uint32_t code; /* ecx value */ 640 uint64_t wval; 641 } msr; 642 struct { 643 int vcpu; 644 uint64_t rip; 645 } spinup_ap; 646 struct { 647 uint64_t rflags; 648 uint64_t intr_status; 649 } hlt; 650 struct { 651 int vector; 652 } ioapic_eoi; 653 struct { 654 enum vm_suspend_how how; 655 } suspended; 656 struct vm_task_switch task_switch; 657 } u; 658}; 659 660/* APIs to inject faults into the guest */ 661void vm_inject_fault(void *vm, int vcpuid, int vector, int errcode_valid, 662 int errcode); 663 664static __inline void 665vm_inject_ud(void *vm, int vcpuid) 666{ 667 vm_inject_fault(vm, vcpuid, IDT_UD, 0, 0); 668} 669 670static __inline void 671vm_inject_gp(void *vm, int vcpuid) 672{ 673 vm_inject_fault(vm, vcpuid, IDT_GP, 1, 0); 674} 675 676static __inline void 677vm_inject_ac(void *vm, int vcpuid, int errcode) 678{ 679 vm_inject_fault(vm, vcpuid, IDT_AC, 1, errcode); 680} 681 682static __inline void 683vm_inject_ss(void *vm, int vcpuid, int errcode) 684{ 685 vm_inject_fault(vm, vcpuid, IDT_SS, 1, errcode); 686} 687 688void vm_inject_pf(void *vm, int vcpuid, int error_code, uint64_t cr2); 689 690int vm_restart_instruction(void *vm, int vcpuid); 691 692#endif /* _VMM_H_ */ 693