1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2011 NetApp, Inc. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND 17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 19 * ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE 20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 * 28 * $FreeBSD$ 29 */ 30 /* 31 * This file and its contents are supplied under the terms of the 32 * Common Development and Distribution License ("CDDL"), version 1.0. 33 * You may only use this file in accordance with the terms of version 34 * 1.0 of the CDDL. 35 * 36 * A full copy of the text of the CDDL should have accompanied this 37 * source. A copy of the CDDL is also available via the Internet at 38 * http://www.illumos.org/license/CDDL. 39 * 40 * Copyright 2015 Pluribus Networks Inc. 41 * Copyright 2019 Joyent, Inc. 42 * Copyright 2022 Oxide Computer Company 43 */ 44 45 #ifndef _VMM_H_ 46 #define _VMM_H_ 47 48 enum vm_suspend_how { 49 VM_SUSPEND_NONE, 50 VM_SUSPEND_RESET, 51 VM_SUSPEND_POWEROFF, 52 VM_SUSPEND_HALT, 53 VM_SUSPEND_TRIPLEFAULT, 54 VM_SUSPEND_LAST 55 }; 56 57 /* 58 * Identifiers for architecturally defined registers. 59 */ 60 enum vm_reg_name { 61 VM_REG_GUEST_RAX, 62 VM_REG_GUEST_RBX, 63 VM_REG_GUEST_RCX, 64 VM_REG_GUEST_RDX, 65 VM_REG_GUEST_RSI, 66 VM_REG_GUEST_RDI, 67 VM_REG_GUEST_RBP, 68 VM_REG_GUEST_R8, 69 VM_REG_GUEST_R9, 70 VM_REG_GUEST_R10, 71 VM_REG_GUEST_R11, 72 VM_REG_GUEST_R12, 73 VM_REG_GUEST_R13, 74 VM_REG_GUEST_R14, 75 VM_REG_GUEST_R15, 76 VM_REG_GUEST_CR0, 77 VM_REG_GUEST_CR3, 78 VM_REG_GUEST_CR4, 79 VM_REG_GUEST_DR7, 80 VM_REG_GUEST_RSP, 81 VM_REG_GUEST_RIP, 82 VM_REG_GUEST_RFLAGS, 83 VM_REG_GUEST_ES, 84 VM_REG_GUEST_CS, 85 VM_REG_GUEST_SS, 86 VM_REG_GUEST_DS, 87 VM_REG_GUEST_FS, 88 VM_REG_GUEST_GS, 89 VM_REG_GUEST_LDTR, 90 VM_REG_GUEST_TR, 91 VM_REG_GUEST_IDTR, 92 VM_REG_GUEST_GDTR, 93 VM_REG_GUEST_EFER, 94 VM_REG_GUEST_CR2, 95 VM_REG_GUEST_PDPTE0, 96 VM_REG_GUEST_PDPTE1, 97 VM_REG_GUEST_PDPTE2, 98 VM_REG_GUEST_PDPTE3, 99 VM_REG_GUEST_INTR_SHADOW, 100 VM_REG_GUEST_DR0, 101 VM_REG_GUEST_DR1, 102 VM_REG_GUEST_DR2, 103 VM_REG_GUEST_DR3, 104 VM_REG_GUEST_DR6, 105 VM_REG_GUEST_ENTRY_INST_LENGTH, 106 VM_REG_GUEST_XCR0, 107 VM_REG_LAST 108 }; 109 110 enum x2apic_state { 111 X2APIC_DISABLED, 112 X2APIC_ENABLED, 113 X2APIC_STATE_LAST 114 }; 115 116 #define VM_INTINFO_MASK_VECTOR 0xffUL 117 #define VM_INTINFO_MASK_TYPE 0x700UL 118 #define VM_INTINFO_MASK_RSVD 0x7ffff000UL 119 #define VM_INTINFO_SHIFT_ERRCODE 32 120 121 #define VM_INTINFO_VECTOR(val) ((val) & VM_INTINFO_MASK_VECTOR) 122 #define VM_INTINFO_TYPE(val) ((val) & VM_INTINFO_MASK_TYPE) 123 #define VM_INTINFO_ERRCODE(val) ((val) >> VM_INTINFO_SHIFT_ERRCODE) 124 #define VM_INTINFO_PENDING(val) (((val) & VM_INTINFO_VALID) != 0) 125 #define VM_INTINFO_HAS_ERRCODE(val) (((val) & VM_INTINFO_DEL_ERRCODE) != 0) 126 127 #define VM_INTINFO_VALID (1UL << 31) 128 #define VM_INTINFO_DEL_ERRCODE (1UL << 11) 129 130 #define VM_INTINFO_HWINTR (0 << 8) 131 #define VM_INTINFO_NMI (2 << 8) 132 #define VM_INTINFO_HWEXCP (3 << 8) 133 #define VM_INTINFO_SWINTR (4 << 8) 134 /* Reserved for CPU (read: Intel) specific types */ 135 #define VM_INTINFO_RESV1 (1 << 8) 136 #define VM_INTINFO_RESV5 (5 << 8) 137 #define VM_INTINFO_RESV6 (6 << 8) 138 #define VM_INTINFO_RESV7 (7 << 8) 139 140 /* 141 * illumos doesn't have a limitation based on SPECNAMELEN like FreeBSD does. 142 * To simplify structure definitions, an arbitrary limit has been chosen. 143 * This same limit is used for memory segment names 144 */ 145 146 #define VM_MAX_NAMELEN 128 147 #define VM_MAX_SEG_NAMELEN 128 148 149 #ifdef _KERNEL 150 #define VM_MAXCPU 32 /* maximum virtual cpus */ 151 #endif 152 153 /* 154 * Identifiers for optional vmm capabilities 155 */ 156 enum vm_cap_type { 157 VM_CAP_HALT_EXIT, 158 VM_CAP_MTRAP_EXIT, 159 VM_CAP_PAUSE_EXIT, 160 VM_CAP_ENABLE_INVPCID, 161 VM_CAP_BPT_EXIT, 162 VM_CAP_MAX 163 }; 164 165 enum vmx_caps { 166 VMX_CAP_NONE = 0, 167 VMX_CAP_TPR_SHADOW = (1UL << 0), 168 VMX_CAP_APICV = (1UL << 1), 169 VMX_CAP_APICV_X2APIC = (1UL << 2), 170 VMX_CAP_APICV_PIR = (1UL << 3), 171 }; 172 173 enum vm_intr_trigger { 174 EDGE_TRIGGER, 175 LEVEL_TRIGGER 176 }; 177 178 /* 179 * The 'access' field has the format specified in Table 21-2 of the Intel 180 * Architecture Manual vol 3b. 181 * 182 * XXX The contents of the 'access' field are architecturally defined except 183 * bit 16 - Segment Unusable. 184 */ 185 struct seg_desc { 186 uint64_t base; 187 uint32_t limit; 188 uint32_t access; 189 }; 190 #define SEG_DESC_TYPE(access) ((access) & 0x001f) 191 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3) 192 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0) 193 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0) 194 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0) 195 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0) 196 197 enum vm_cpu_mode { 198 CPU_MODE_REAL, 199 CPU_MODE_PROTECTED, 200 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 201 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 202 }; 203 204 enum vm_paging_mode { 205 PAGING_MODE_FLAT, 206 PAGING_MODE_32, 207 PAGING_MODE_PAE, 208 PAGING_MODE_64, 209 }; 210 211 struct vm_guest_paging { 212 uint64_t cr3; 213 int cpl; 214 enum vm_cpu_mode cpu_mode; 215 enum vm_paging_mode paging_mode; 216 }; 217 218 enum vm_exitcode { 219 VM_EXITCODE_INOUT, 220 VM_EXITCODE_VMX, 221 VM_EXITCODE_BOGUS, 222 VM_EXITCODE_RDMSR, 223 VM_EXITCODE_WRMSR, 224 VM_EXITCODE_HLT, 225 VM_EXITCODE_MTRAP, 226 VM_EXITCODE_PAUSE, 227 VM_EXITCODE_PAGING, 228 VM_EXITCODE_INST_EMUL, 229 VM_EXITCODE_RUN_STATE, 230 VM_EXITCODE_MMIO_EMUL, 231 VM_EXITCODE_DEPRECATED, /* formerly RUNBLOCK */ 232 VM_EXITCODE_IOAPIC_EOI, 233 VM_EXITCODE_SUSPENDED, 234 VM_EXITCODE_MMIO, 235 VM_EXITCODE_TASK_SWITCH, 236 VM_EXITCODE_MONITOR, 237 VM_EXITCODE_MWAIT, 238 VM_EXITCODE_SVM, 239 VM_EXITCODE_REQIDLE, 240 VM_EXITCODE_DEBUG, 241 VM_EXITCODE_VMINSN, 242 VM_EXITCODE_BPT, 243 VM_EXITCODE_HT, 244 VM_EXITCODE_MAX 245 }; 246 247 enum inout_flags { 248 INOUT_IN = (1U << 0), /* direction: 'in' when set, else 'out' */ 249 250 /* 251 * The following flags are used only for in-kernel emulation logic and 252 * are not exposed to userspace. 253 */ 254 INOUT_STR = (1U << 1), /* ins/outs operation */ 255 INOUT_REP = (1U << 2), /* 'rep' prefix present on instruction */ 256 }; 257 258 struct vm_inout { 259 uint32_t eax; 260 uint16_t port; 261 uint8_t bytes; /* 1 or 2 or 4 */ 262 uint8_t flags; /* see: inout_flags */ 263 264 /* 265 * The address size and segment are relevant to INS/OUTS operations. 266 * Userspace is not concerned with them since the in-kernel emulation 267 * handles those specific aspects. 268 */ 269 uint8_t addrsize; 270 uint8_t segment; 271 }; 272 273 struct vm_mmio { 274 uint8_t bytes; /* 1/2/4/8 bytes */ 275 uint8_t read; /* read: 1, write: 0 */ 276 uint16_t _pad[3]; 277 uint64_t gpa; 278 uint64_t data; 279 }; 280 281 enum task_switch_reason { 282 TSR_CALL, 283 TSR_IRET, 284 TSR_JMP, 285 TSR_IDT_GATE, /* task gate in IDT */ 286 }; 287 288 struct vm_task_switch { 289 uint16_t tsssel; /* new TSS selector */ 290 int ext; /* task switch due to external event */ 291 uint32_t errcode; 292 int errcode_valid; /* push 'errcode' on the new stack */ 293 enum task_switch_reason reason; 294 struct vm_guest_paging paging; 295 }; 296 297 enum vcpu_run_state { 298 VRS_HALT = 0, 299 VRS_INIT = (1 << 0), 300 VRS_RUN = (1 << 1), 301 302 VRS_PEND_INIT = (1 << 14), 303 VRS_PEND_SIPI = (1 << 15), 304 }; 305 #define VRS_MASK_VALID(v) \ 306 ((v) & (VRS_INIT | VRS_RUN | VRS_PEND_SIPI | VRS_PEND_SIPI)) 307 #define VRS_IS_VALID(v) ((v) == VRS_MASK_VALID(v)) 308 309 struct vm_exit { 310 enum vm_exitcode exitcode; 311 int inst_length; /* 0 means unknown */ 312 uint64_t rip; 313 union { 314 struct vm_inout inout; 315 struct vm_mmio mmio; 316 struct { 317 uint64_t gpa; 318 int fault_type; 319 } paging; 320 /* 321 * Kernel-internal MMIO decoding and emulation. 322 * Userspace should not expect to see this, but rather a 323 * VM_EXITCODE_MMIO with the above 'mmio' context. 324 */ 325 struct { 326 uint64_t gpa; 327 uint64_t gla; 328 uint64_t cs_base; 329 int cs_d; /* CS.D */ 330 } mmio_emul; 331 struct { 332 uint8_t inst[15]; 333 uint8_t num_valid; 334 } inst_emul; 335 /* 336 * VMX specific payload. Used when there is no "better" 337 * exitcode to represent the VM-exit. 338 */ 339 struct { 340 int status; /* vmx inst status */ 341 /* 342 * 'exit_reason' and 'exit_qualification' are valid 343 * only if 'status' is zero. 344 */ 345 uint32_t exit_reason; 346 uint64_t exit_qualification; 347 /* 348 * 'inst_error' and 'inst_type' are valid 349 * only if 'status' is non-zero. 350 */ 351 int inst_type; 352 int inst_error; 353 } vmx; 354 /* 355 * SVM specific payload. 356 */ 357 struct { 358 uint64_t exitcode; 359 uint64_t exitinfo1; 360 uint64_t exitinfo2; 361 } svm; 362 struct { 363 int inst_length; 364 } bpt; 365 struct { 366 uint32_t code; /* ecx value */ 367 uint64_t wval; 368 } msr; 369 struct { 370 uint64_t rflags; 371 } hlt; 372 struct { 373 int vector; 374 } ioapic_eoi; 375 struct { 376 enum vm_suspend_how how; 377 } suspended; 378 struct vm_task_switch task_switch; 379 } u; 380 }; 381 382 enum vm_entry_cmds { 383 VEC_DEFAULT = 0, 384 VEC_DISCARD_INSTR, /* discard inst emul state */ 385 VEC_FULFILL_MMIO, /* entry includes result for mmio emul */ 386 VEC_FULFILL_INOUT, /* entry includes result for inout emul */ 387 }; 388 389 struct vm_entry { 390 int cpuid; 391 uint_t cmd; /* see: vm_entry_cmds */ 392 void *exit_data; 393 union { 394 struct vm_inout inout; 395 struct vm_mmio mmio; 396 } u; 397 }; 398 399 int vm_restart_instruction(void *vm, int vcpuid); 400 401 enum vm_create_flags { 402 /* 403 * Allocate guest memory segments from existing reservoir capacity, 404 * rather than attempting to create transient allocations. 405 */ 406 VCF_RESERVOIR_MEM = (1 << 0), 407 408 /* 409 * Enable dirty page tracking for the guest. 410 */ 411 VCF_TRACK_DIRTY = (1 << 1), 412 }; 413 414 /* 415 * Describes an entry for `cpuid` emulation. 416 * Used internally by bhyve (kernel) in addition to exposed ioctl(2) interface. 417 */ 418 struct vcpu_cpuid_entry { 419 uint32_t vce_function; 420 uint32_t vce_index; 421 uint32_t vce_flags; 422 uint32_t vce_eax; 423 uint32_t vce_ebx; 424 uint32_t vce_ecx; 425 uint32_t vce_edx; 426 uint32_t _pad; 427 }; 428 429 /* 430 * Defined flags for vcpu_cpuid_entry`vce_flags are below. 431 */ 432 433 /* Use index (ecx) input value when matching entry */ 434 #define VCE_FLAG_MATCH_INDEX (1 << 0) 435 436 /* All valid flacts for vcpu_cpuid_entry`vce_flags */ 437 #define VCE_FLAGS_VALID VCE_FLAG_MATCH_INDEX 438 439 /* 440 * Defined flags for vcpu_cpuid configuration are below. 441 * These are used by both the ioctl(2) interface via vm_vcpu_cpuid_config and 442 * internally in the kernel vmm. 443 */ 444 445 /* Use legacy hard-coded cpuid masking tables applied to the host CPU */ 446 #define VCC_FLAG_LEGACY_HANDLING (1 << 0) 447 /* 448 * Emulate Intel-style fallback behavior (emit highest "standard" entry) if the 449 * queried function/index do not match. If not set, emulate AMD-style, where 450 * all zeroes are returned in such cases. 451 */ 452 #define VCC_FLAG_INTEL_FALLBACK (1 << 1) 453 454 /* All valid flacts for vm_vcpu_cpuid_config`vvcc_flags */ 455 #define VCC_FLAGS_VALID \ 456 (VCC_FLAG_LEGACY_HANDLING | VCC_FLAG_INTEL_FALLBACK) 457 458 /* Maximum vcpu_cpuid_entry records per vCPU */ 459 #define VMM_MAX_CPUID_ENTRIES 256 460 461 #endif /* _VMM_H_ */ 462