vmm.h revision 270074
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/10/sys/amd64/include/vmm.h 270074 2014-08-17 01:23:52Z grehan $ 27 */ 28 29#ifndef _VMM_H_ 30#define _VMM_H_ 31 32enum vm_suspend_how { 33 VM_SUSPEND_NONE, 34 VM_SUSPEND_RESET, 35 VM_SUSPEND_POWEROFF, 36 VM_SUSPEND_HALT, 37 VM_SUSPEND_LAST 38}; 39 40#ifdef _KERNEL 41 42#define VM_MAX_NAMELEN 32 43 44struct vm; 45struct vm_exception; 46struct vm_memory_segment; 47struct seg_desc; 48struct vm_exit; 49struct vm_run; 50struct vhpet; 51struct vioapic; 52struct vlapic; 53struct vmspace; 54struct vm_object; 55struct pmap; 56 57enum vm_reg_name; 58enum x2apic_state; 59 60typedef int (*vmm_init_func_t)(int ipinum); 61typedef int (*vmm_cleanup_func_t)(void); 62typedef void (*vmm_resume_func_t)(void); 63typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap); 64typedef int (*vmi_run_func_t)(void *vmi, int vcpu, register_t rip, 65 struct pmap *pmap, void *rendezvous_cookie, 66 void *suspend_cookie); 67typedef void (*vmi_cleanup_func_t)(void *vmi); 68typedef int (*vmi_get_register_t)(void *vmi, int vcpu, int num, 69 uint64_t *retval); 70typedef int (*vmi_set_register_t)(void *vmi, int vcpu, int num, 71 uint64_t val); 72typedef int (*vmi_get_desc_t)(void *vmi, int vcpu, int num, 73 struct seg_desc *desc); 74typedef int (*vmi_set_desc_t)(void *vmi, int vcpu, int num, 75 struct seg_desc *desc); 76typedef int (*vmi_get_cap_t)(void *vmi, int vcpu, int num, int *retval); 77typedef int (*vmi_set_cap_t)(void *vmi, int vcpu, int num, int val); 78typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max); 79typedef void (*vmi_vmspace_free)(struct vmspace *vmspace); 80typedef struct vlapic * (*vmi_vlapic_init)(void *vmi, int vcpu); 81typedef void (*vmi_vlapic_cleanup)(void *vmi, struct vlapic *vlapic); 82 83struct vmm_ops { 84 vmm_init_func_t init; /* module wide initialization */ 85 vmm_cleanup_func_t cleanup; 86 vmm_resume_func_t resume; 87 88 vmi_init_func_t vminit; /* vm-specific initialization */ 89 vmi_run_func_t vmrun; 90 vmi_cleanup_func_t vmcleanup; 91 vmi_get_register_t vmgetreg; 92 vmi_set_register_t vmsetreg; 93 vmi_get_desc_t vmgetdesc; 94 vmi_set_desc_t vmsetdesc; 95 vmi_get_cap_t vmgetcap; 96 vmi_set_cap_t vmsetcap; 97 vmi_vmspace_alloc vmspace_alloc; 98 vmi_vmspace_free vmspace_free; 99 vmi_vlapic_init vlapic_init; 100 vmi_vlapic_cleanup vlapic_cleanup; 101}; 102 103extern struct vmm_ops vmm_ops_intel; 104extern struct vmm_ops vmm_ops_amd; 105 106int vm_create(const char *name, struct vm **retvm); 107void vm_destroy(struct vm *vm); 108int vm_reinit(struct vm *vm); 109const char *vm_name(struct vm *vm); 110int vm_malloc(struct vm *vm, vm_paddr_t gpa, size_t len); 111int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa); 112int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len); 113void *vm_gpa_hold(struct vm *, vm_paddr_t gpa, size_t len, int prot, 114 void **cookie); 115void vm_gpa_release(void *cookie); 116int vm_gpabase2memseg(struct vm *vm, vm_paddr_t gpabase, 117 struct vm_memory_segment *seg); 118int vm_get_memobj(struct vm *vm, vm_paddr_t gpa, size_t len, 119 vm_offset_t *offset, struct vm_object **object); 120boolean_t vm_mem_allocated(struct vm *vm, vm_paddr_t gpa); 121int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval); 122int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val); 123int vm_get_seg_desc(struct vm *vm, int vcpu, int reg, 124 struct seg_desc *ret_desc); 125int vm_set_seg_desc(struct vm *vm, int vcpu, int reg, 126 struct seg_desc *desc); 127int vm_run(struct vm *vm, struct vm_run *vmrun); 128int vm_suspend(struct vm *vm, enum vm_suspend_how how); 129int vm_inject_nmi(struct vm *vm, int vcpu); 130int vm_nmi_pending(struct vm *vm, int vcpuid); 131void vm_nmi_clear(struct vm *vm, int vcpuid); 132int vm_inject_extint(struct vm *vm, int vcpu); 133int vm_extint_pending(struct vm *vm, int vcpuid); 134void vm_extint_clear(struct vm *vm, int vcpuid); 135uint64_t *vm_guest_msrs(struct vm *vm, int cpu); 136struct vlapic *vm_lapic(struct vm *vm, int cpu); 137struct vioapic *vm_ioapic(struct vm *vm); 138struct vhpet *vm_hpet(struct vm *vm); 139int vm_get_capability(struct vm *vm, int vcpu, int type, int *val); 140int vm_set_capability(struct vm *vm, int vcpu, int type, int val); 141int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state); 142int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state); 143int vm_apicid2vcpuid(struct vm *vm, int apicid); 144int vm_activate_cpu(struct vm *vm, int vcpu); 145cpuset_t vm_active_cpus(struct vm *vm); 146cpuset_t vm_suspended_cpus(struct vm *vm); 147struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid); 148void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip); 149void vm_exit_rendezvous(struct vm *vm, int vcpuid, uint64_t rip); 150void vm_exit_astpending(struct vm *vm, int vcpuid, uint64_t rip); 151 152/* 153 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'. 154 * The rendezvous 'func(arg)' is not allowed to do anything that will 155 * cause the thread to be put to sleep. 156 * 157 * If the rendezvous is being initiated from a vcpu context then the 158 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1. 159 * 160 * The caller cannot hold any locks when initiating the rendezvous. 161 * 162 * The implementation of this API may cause vcpus other than those specified 163 * by 'dest' to be stalled. The caller should not rely on any vcpus making 164 * forward progress when the rendezvous is in progress. 165 */ 166typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg); 167void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest, 168 vm_rendezvous_func_t func, void *arg); 169 170static __inline int 171vcpu_rendezvous_pending(void *rendezvous_cookie) 172{ 173 174 return (*(uintptr_t *)rendezvous_cookie != 0); 175} 176 177static __inline int 178vcpu_suspended(void *suspend_cookie) 179{ 180 181 return (*(int *)suspend_cookie); 182} 183 184/* 185 * Return 1 if device indicated by bus/slot/func is supposed to be a 186 * pci passthrough device. 187 * 188 * Return 0 otherwise. 189 */ 190int vmm_is_pptdev(int bus, int slot, int func); 191 192void *vm_iommu_domain(struct vm *vm); 193 194enum vcpu_state { 195 VCPU_IDLE, 196 VCPU_FROZEN, 197 VCPU_RUNNING, 198 VCPU_SLEEPING, 199}; 200 201int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state, 202 bool from_idle); 203enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu); 204 205static int __inline 206vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu) 207{ 208 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING); 209} 210 211void *vcpu_stats(struct vm *vm, int vcpu); 212void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr); 213struct vmspace *vm_get_vmspace(struct vm *vm); 214int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func); 215int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func); 216struct vatpic *vm_atpic(struct vm *vm); 217struct vatpit *vm_atpit(struct vm *vm); 218 219/* 220 * Inject exception 'vme' into the guest vcpu. This function returns 0 on 221 * success and non-zero on failure. 222 * 223 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling 224 * this function directly because they enforce the trap-like or fault-like 225 * behavior of an exception. 226 * 227 * This function should only be called in the context of the thread that is 228 * executing this vcpu. 229 */ 230int vm_inject_exception(struct vm *vm, int vcpuid, struct vm_exception *vme); 231 232/* 233 * Returns 0 if there is no exception pending for this vcpu. Returns 1 if an 234 * exception is pending and also updates 'vme'. The pending exception is 235 * cleared when this function returns. 236 * 237 * This function should only be called in the context of the thread that is 238 * executing this vcpu. 239 */ 240int vm_exception_pending(struct vm *vm, int vcpuid, struct vm_exception *vme); 241 242void vm_inject_gp(struct vm *vm, int vcpuid); /* general protection fault */ 243void vm_inject_ud(struct vm *vm, int vcpuid); /* undefined instruction fault */ 244void vm_inject_pf(struct vm *vm, int vcpuid, int error_code, uint64_t cr2); 245 246enum vm_reg_name vm_segment_name(int seg_encoding); 247 248#endif /* KERNEL */ 249 250#define VM_MAXCPU 16 /* maximum virtual cpus */ 251 252/* 253 * Identifiers for architecturally defined registers. 254 */ 255enum vm_reg_name { 256 VM_REG_GUEST_RAX, 257 VM_REG_GUEST_RBX, 258 VM_REG_GUEST_RCX, 259 VM_REG_GUEST_RDX, 260 VM_REG_GUEST_RSI, 261 VM_REG_GUEST_RDI, 262 VM_REG_GUEST_RBP, 263 VM_REG_GUEST_R8, 264 VM_REG_GUEST_R9, 265 VM_REG_GUEST_R10, 266 VM_REG_GUEST_R11, 267 VM_REG_GUEST_R12, 268 VM_REG_GUEST_R13, 269 VM_REG_GUEST_R14, 270 VM_REG_GUEST_R15, 271 VM_REG_GUEST_CR0, 272 VM_REG_GUEST_CR3, 273 VM_REG_GUEST_CR4, 274 VM_REG_GUEST_DR7, 275 VM_REG_GUEST_RSP, 276 VM_REG_GUEST_RIP, 277 VM_REG_GUEST_RFLAGS, 278 VM_REG_GUEST_ES, 279 VM_REG_GUEST_CS, 280 VM_REG_GUEST_SS, 281 VM_REG_GUEST_DS, 282 VM_REG_GUEST_FS, 283 VM_REG_GUEST_GS, 284 VM_REG_GUEST_LDTR, 285 VM_REG_GUEST_TR, 286 VM_REG_GUEST_IDTR, 287 VM_REG_GUEST_GDTR, 288 VM_REG_GUEST_EFER, 289 VM_REG_GUEST_CR2, 290 VM_REG_LAST 291}; 292 293/* 294 * Identifiers for optional vmm capabilities 295 */ 296enum vm_cap_type { 297 VM_CAP_HALT_EXIT, 298 VM_CAP_MTRAP_EXIT, 299 VM_CAP_PAUSE_EXIT, 300 VM_CAP_UNRESTRICTED_GUEST, 301 VM_CAP_ENABLE_INVPCID, 302 VM_CAP_MAX 303}; 304 305enum x2apic_state { 306 X2APIC_DISABLED, 307 X2APIC_ENABLED, 308 X2APIC_STATE_LAST 309}; 310 311enum vm_intr_trigger { 312 EDGE_TRIGGER, 313 LEVEL_TRIGGER 314}; 315 316/* 317 * The 'access' field has the format specified in Table 21-2 of the Intel 318 * Architecture Manual vol 3b. 319 * 320 * XXX The contents of the 'access' field are architecturally defined except 321 * bit 16 - Segment Unusable. 322 */ 323struct seg_desc { 324 uint64_t base; 325 uint32_t limit; 326 uint32_t access; 327}; 328#define SEG_DESC_TYPE(desc) ((desc)->access & 0x001f) 329#define SEG_DESC_PRESENT(desc) ((desc)->access & 0x0080) 330#define SEG_DESC_DEF32(desc) ((desc)->access & 0x4000) 331#define SEG_DESC_GRANULARITY(desc) ((desc)->access & 0x8000) 332#define SEG_DESC_UNUSABLE(desc) ((desc)->access & 0x10000) 333 334enum vm_cpu_mode { 335 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 336 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 337}; 338 339enum vm_paging_mode { 340 PAGING_MODE_FLAT, 341 PAGING_MODE_32, 342 PAGING_MODE_PAE, 343 PAGING_MODE_64, 344}; 345 346struct vm_guest_paging { 347 uint64_t cr3; 348 int cpl; 349 enum vm_cpu_mode cpu_mode; 350 enum vm_paging_mode paging_mode; 351}; 352 353/* 354 * The data structures 'vie' and 'vie_op' are meant to be opaque to the 355 * consumers of instruction decoding. The only reason why their contents 356 * need to be exposed is because they are part of the 'vm_exit' structure. 357 */ 358struct vie_op { 359 uint8_t op_byte; /* actual opcode byte */ 360 uint8_t op_type; /* type of operation (e.g. MOV) */ 361 uint16_t op_flags; 362}; 363 364#define VIE_INST_SIZE 15 365struct vie { 366 uint8_t inst[VIE_INST_SIZE]; /* instruction bytes */ 367 uint8_t num_valid; /* size of the instruction */ 368 uint8_t num_processed; 369 370 uint8_t rex_w:1, /* REX prefix */ 371 rex_r:1, 372 rex_x:1, 373 rex_b:1, 374 rex_present:1; 375 376 uint8_t mod:2, /* ModRM byte */ 377 reg:4, 378 rm:4; 379 380 uint8_t ss:2, /* SIB byte */ 381 index:4, 382 base:4; 383 384 uint8_t disp_bytes; 385 uint8_t imm_bytes; 386 387 uint8_t scale; 388 int base_register; /* VM_REG_GUEST_xyz */ 389 int index_register; /* VM_REG_GUEST_xyz */ 390 391 int64_t displacement; /* optional addr displacement */ 392 int64_t immediate; /* optional immediate operand */ 393 394 uint8_t decoded; /* set to 1 if successfully decoded */ 395 396 struct vie_op op; /* opcode description */ 397}; 398 399enum vm_exitcode { 400 VM_EXITCODE_INOUT, 401 VM_EXITCODE_VMX, 402 VM_EXITCODE_BOGUS, 403 VM_EXITCODE_RDMSR, 404 VM_EXITCODE_WRMSR, 405 VM_EXITCODE_HLT, 406 VM_EXITCODE_MTRAP, 407 VM_EXITCODE_PAUSE, 408 VM_EXITCODE_PAGING, 409 VM_EXITCODE_INST_EMUL, 410 VM_EXITCODE_SPINUP_AP, 411 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */ 412 VM_EXITCODE_RENDEZVOUS, 413 VM_EXITCODE_IOAPIC_EOI, 414 VM_EXITCODE_SUSPENDED, 415 VM_EXITCODE_INOUT_STR, 416 VM_EXITCODE_MAX 417}; 418 419struct vm_inout { 420 uint16_t bytes:3; /* 1 or 2 or 4 */ 421 uint16_t in:1; 422 uint16_t string:1; 423 uint16_t rep:1; 424 uint16_t port; 425 uint32_t eax; /* valid for out */ 426}; 427 428struct vm_inout_str { 429 struct vm_inout inout; /* must be the first element */ 430 struct vm_guest_paging paging; 431 uint64_t rflags; 432 uint64_t cr0; 433 uint64_t index; 434 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */ 435 int addrsize; 436 enum vm_reg_name seg_name; 437 struct seg_desc seg_desc; 438}; 439 440struct vm_exit { 441 enum vm_exitcode exitcode; 442 int inst_length; /* 0 means unknown */ 443 uint64_t rip; 444 union { 445 struct vm_inout inout; 446 struct vm_inout_str inout_str; 447 struct { 448 uint64_t gpa; 449 int fault_type; 450 } paging; 451 struct { 452 uint64_t gpa; 453 uint64_t gla; 454 struct vm_guest_paging paging; 455 struct vie vie; 456 } inst_emul; 457 /* 458 * VMX specific payload. Used when there is no "better" 459 * exitcode to represent the VM-exit. 460 */ 461 struct { 462 int status; /* vmx inst status */ 463 /* 464 * 'exit_reason' and 'exit_qualification' are valid 465 * only if 'status' is zero. 466 */ 467 uint32_t exit_reason; 468 uint64_t exit_qualification; 469 /* 470 * 'inst_error' and 'inst_type' are valid 471 * only if 'status' is non-zero. 472 */ 473 int inst_type; 474 int inst_error; 475 } vmx; 476 struct { 477 uint32_t code; /* ecx value */ 478 uint64_t wval; 479 } msr; 480 struct { 481 int vcpu; 482 uint64_t rip; 483 } spinup_ap; 484 struct { 485 uint64_t rflags; 486 } hlt; 487 struct { 488 int vector; 489 } ioapic_eoi; 490 struct { 491 enum vm_suspend_how how; 492 } suspended; 493 } u; 494}; 495 496#endif /* _VMM_H_ */ 497