vmm.h revision 266573
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: head/sys/amd64/include/vmm.h 266573 2014-05-23 05:15:17Z neel $ 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); 108const char *vm_name(struct vm *vm); 109int vm_malloc(struct vm *vm, vm_paddr_t gpa, size_t len); 110int vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa); 111int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len); 112void *vm_gpa_hold(struct vm *, vm_paddr_t gpa, size_t len, int prot, 113 void **cookie); 114void vm_gpa_release(void *cookie); 115int vm_gpabase2memseg(struct vm *vm, vm_paddr_t gpabase, 116 struct vm_memory_segment *seg); 117int vm_get_memobj(struct vm *vm, vm_paddr_t gpa, size_t len, 118 vm_offset_t *offset, struct vm_object **object); 119boolean_t vm_mem_allocated(struct vm *vm, vm_paddr_t gpa); 120int vm_get_register(struct vm *vm, int vcpu, int reg, uint64_t *retval); 121int vm_set_register(struct vm *vm, int vcpu, int reg, uint64_t val); 122int vm_get_seg_desc(struct vm *vm, int vcpu, int reg, 123 struct seg_desc *ret_desc); 124int vm_set_seg_desc(struct vm *vm, int vcpu, int reg, 125 struct seg_desc *desc); 126int vm_run(struct vm *vm, struct vm_run *vmrun); 127int vm_suspend(struct vm *vm, enum vm_suspend_how how); 128int vm_inject_nmi(struct vm *vm, int vcpu); 129int vm_nmi_pending(struct vm *vm, int vcpuid); 130void vm_nmi_clear(struct vm *vm, int vcpuid); 131int vm_inject_extint(struct vm *vm, int vcpu); 132int vm_extint_pending(struct vm *vm, int vcpuid); 133void vm_extint_clear(struct vm *vm, int vcpuid); 134uint64_t *vm_guest_msrs(struct vm *vm, int cpu); 135struct vlapic *vm_lapic(struct vm *vm, int cpu); 136struct vioapic *vm_ioapic(struct vm *vm); 137struct vhpet *vm_hpet(struct vm *vm); 138int vm_get_capability(struct vm *vm, int vcpu, int type, int *val); 139int vm_set_capability(struct vm *vm, int vcpu, int type, int val); 140int vm_get_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state *state); 141int vm_set_x2apic_state(struct vm *vm, int vcpu, enum x2apic_state state); 142int vm_apicid2vcpuid(struct vm *vm, int apicid); 143void vm_activate_cpu(struct vm *vm, int vcpu); 144cpuset_t vm_active_cpus(struct vm *vm); 145struct vm_exit *vm_exitinfo(struct vm *vm, int vcpuid); 146void vm_exit_suspended(struct vm *vm, int vcpuid, uint64_t rip); 147 148/* 149 * Rendezvous all vcpus specified in 'dest' and execute 'func(arg)'. 150 * The rendezvous 'func(arg)' is not allowed to do anything that will 151 * cause the thread to be put to sleep. 152 * 153 * If the rendezvous is being initiated from a vcpu context then the 154 * 'vcpuid' must refer to that vcpu, otherwise it should be set to -1. 155 * 156 * The caller cannot hold any locks when initiating the rendezvous. 157 * 158 * The implementation of this API may cause vcpus other than those specified 159 * by 'dest' to be stalled. The caller should not rely on any vcpus making 160 * forward progress when the rendezvous is in progress. 161 */ 162typedef void (*vm_rendezvous_func_t)(struct vm *vm, int vcpuid, void *arg); 163void vm_smp_rendezvous(struct vm *vm, int vcpuid, cpuset_t dest, 164 vm_rendezvous_func_t func, void *arg); 165 166static __inline int 167vcpu_rendezvous_pending(void *rendezvous_cookie) 168{ 169 170 return (*(uintptr_t *)rendezvous_cookie != 0); 171} 172 173static __inline int 174vcpu_suspended(void *suspend_cookie) 175{ 176 177 return (*(int *)suspend_cookie); 178} 179 180/* 181 * Return 1 if device indicated by bus/slot/func is supposed to be a 182 * pci passthrough device. 183 * 184 * Return 0 otherwise. 185 */ 186int vmm_is_pptdev(int bus, int slot, int func); 187 188void *vm_iommu_domain(struct vm *vm); 189 190enum vcpu_state { 191 VCPU_IDLE, 192 VCPU_FROZEN, 193 VCPU_RUNNING, 194 VCPU_SLEEPING, 195}; 196 197int vcpu_set_state(struct vm *vm, int vcpu, enum vcpu_state state, 198 bool from_idle); 199enum vcpu_state vcpu_get_state(struct vm *vm, int vcpu, int *hostcpu); 200 201static int __inline 202vcpu_is_running(struct vm *vm, int vcpu, int *hostcpu) 203{ 204 return (vcpu_get_state(vm, vcpu, hostcpu) == VCPU_RUNNING); 205} 206 207void *vcpu_stats(struct vm *vm, int vcpu); 208void vcpu_notify_event(struct vm *vm, int vcpuid, bool lapic_intr); 209struct vmspace *vm_get_vmspace(struct vm *vm); 210int vm_assign_pptdev(struct vm *vm, int bus, int slot, int func); 211int vm_unassign_pptdev(struct vm *vm, int bus, int slot, int func); 212struct vatpic *vm_atpic(struct vm *vm); 213struct vatpit *vm_atpit(struct vm *vm); 214 215/* 216 * Inject exception 'vme' into the guest vcpu. This function returns 0 on 217 * success and non-zero on failure. 218 * 219 * Wrapper functions like 'vm_inject_gp()' should be preferred to calling 220 * this function directly because they enforce the trap-like or fault-like 221 * behavior of an exception. 222 * 223 * This function should only be called in the context of the thread that is 224 * executing this vcpu. 225 */ 226int vm_inject_exception(struct vm *vm, int vcpuid, struct vm_exception *vme); 227 228/* 229 * Returns 0 if there is no exception pending for this vcpu. Returns 1 if an 230 * exception is pending and also updates 'vme'. The pending exception is 231 * cleared when this function returns. 232 * 233 * This function should only be called in the context of the thread that is 234 * executing this vcpu. 235 */ 236int vm_exception_pending(struct vm *vm, int vcpuid, struct vm_exception *vme); 237 238void vm_inject_gp(struct vm *vm, int vcpuid); /* general protection fault */ 239void vm_inject_ud(struct vm *vm, int vcpuid); /* undefined instruction fault */ 240void vm_inject_pf(struct vm *vm, int vcpuid, int error_code); /* page fault */ 241 242enum vm_reg_name vm_segment_name(int seg_encoding); 243 244#endif /* KERNEL */ 245 246#include <machine/vmm_instruction_emul.h> 247 248#define VM_MAXCPU 16 /* maximum virtual cpus */ 249 250/* 251 * Identifiers for architecturally defined registers. 252 */ 253enum vm_reg_name { 254 VM_REG_GUEST_RAX, 255 VM_REG_GUEST_RBX, 256 VM_REG_GUEST_RCX, 257 VM_REG_GUEST_RDX, 258 VM_REG_GUEST_RSI, 259 VM_REG_GUEST_RDI, 260 VM_REG_GUEST_RBP, 261 VM_REG_GUEST_R8, 262 VM_REG_GUEST_R9, 263 VM_REG_GUEST_R10, 264 VM_REG_GUEST_R11, 265 VM_REG_GUEST_R12, 266 VM_REG_GUEST_R13, 267 VM_REG_GUEST_R14, 268 VM_REG_GUEST_R15, 269 VM_REG_GUEST_CR0, 270 VM_REG_GUEST_CR3, 271 VM_REG_GUEST_CR4, 272 VM_REG_GUEST_DR7, 273 VM_REG_GUEST_RSP, 274 VM_REG_GUEST_RIP, 275 VM_REG_GUEST_RFLAGS, 276 VM_REG_GUEST_ES, 277 VM_REG_GUEST_CS, 278 VM_REG_GUEST_SS, 279 VM_REG_GUEST_DS, 280 VM_REG_GUEST_FS, 281 VM_REG_GUEST_GS, 282 VM_REG_GUEST_LDTR, 283 VM_REG_GUEST_TR, 284 VM_REG_GUEST_IDTR, 285 VM_REG_GUEST_GDTR, 286 VM_REG_GUEST_EFER, 287 VM_REG_LAST 288}; 289 290/* 291 * Identifiers for optional vmm capabilities 292 */ 293enum vm_cap_type { 294 VM_CAP_HALT_EXIT, 295 VM_CAP_MTRAP_EXIT, 296 VM_CAP_PAUSE_EXIT, 297 VM_CAP_UNRESTRICTED_GUEST, 298 VM_CAP_ENABLE_INVPCID, 299 VM_CAP_MAX 300}; 301 302enum x2apic_state { 303 X2APIC_DISABLED, 304 X2APIC_ENABLED, 305 X2APIC_STATE_LAST 306}; 307 308enum vm_intr_trigger { 309 EDGE_TRIGGER, 310 LEVEL_TRIGGER 311}; 312 313/* 314 * The 'access' field has the format specified in Table 21-2 of the Intel 315 * Architecture Manual vol 3b. 316 * 317 * XXX The contents of the 'access' field are architecturally defined except 318 * bit 16 - Segment Unusable. 319 */ 320struct seg_desc { 321 uint64_t base; 322 uint32_t limit; 323 uint32_t access; 324}; 325 326enum vm_exitcode { 327 VM_EXITCODE_INOUT, 328 VM_EXITCODE_VMX, 329 VM_EXITCODE_BOGUS, 330 VM_EXITCODE_RDMSR, 331 VM_EXITCODE_WRMSR, 332 VM_EXITCODE_HLT, 333 VM_EXITCODE_MTRAP, 334 VM_EXITCODE_PAUSE, 335 VM_EXITCODE_PAGING, 336 VM_EXITCODE_INST_EMUL, 337 VM_EXITCODE_SPINUP_AP, 338 VM_EXITCODE_DEPRECATED1, /* used to be SPINDOWN_CPU */ 339 VM_EXITCODE_RENDEZVOUS, 340 VM_EXITCODE_IOAPIC_EOI, 341 VM_EXITCODE_SUSPENDED, 342 VM_EXITCODE_INOUT_STR, 343 VM_EXITCODE_MAX 344}; 345 346struct vm_inout { 347 uint16_t bytes:3; /* 1 or 2 or 4 */ 348 uint16_t in:1; 349 uint16_t string:1; 350 uint16_t rep:1; 351 uint16_t port; 352 uint32_t eax; /* valid for out */ 353}; 354 355struct vm_inout_str { 356 struct vm_inout inout; /* must be the first element */ 357 enum vie_cpu_mode cpu_mode; 358 enum vie_paging_mode paging_mode; 359 uint64_t rflags; 360 uint64_t cr0; 361 uint64_t cr3; 362 uint64_t index; 363 uint64_t count; /* rep=1 (%rcx), rep=0 (1) */ 364 int cpl; 365 int addrsize; 366 enum vm_reg_name seg_name; 367 struct seg_desc seg_desc; 368 uint64_t gla; /* may be set to VIE_INVALID_GLA */ 369 uint64_t gpa; 370}; 371 372struct vm_exit { 373 enum vm_exitcode exitcode; 374 int inst_length; /* 0 means unknown */ 375 uint64_t rip; 376 union { 377 struct vm_inout inout; 378 struct vm_inout_str inout_str; 379 struct { 380 uint64_t gpa; 381 int fault_type; 382 } paging; 383 struct { 384 uint64_t gpa; 385 uint64_t gla; 386 uint64_t cr3; 387 enum vie_cpu_mode cpu_mode; 388 enum vie_paging_mode paging_mode; 389 int cpl; 390 struct vie vie; 391 } inst_emul; 392 /* 393 * VMX specific payload. Used when there is no "better" 394 * exitcode to represent the VM-exit. 395 */ 396 struct { 397 int status; /* vmx inst status */ 398 /* 399 * 'exit_reason' and 'exit_qualification' are valid 400 * only if 'status' is zero. 401 */ 402 uint32_t exit_reason; 403 uint64_t exit_qualification; 404 /* 405 * 'inst_error' and 'inst_type' are valid 406 * only if 'status' is non-zero. 407 */ 408 int inst_type; 409 int inst_error; 410 } vmx; 411 struct { 412 uint32_t code; /* ecx value */ 413 uint64_t wval; 414 } msr; 415 struct { 416 int vcpu; 417 uint64_t rip; 418 } spinup_ap; 419 struct { 420 uint64_t rflags; 421 } hlt; 422 struct { 423 int vector; 424 } ioapic_eoi; 425 struct { 426 enum vm_suspend_how how; 427 } suspended; 428 } u; 429}; 430 431#endif /* _VMM_H_ */ 432