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 2021 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_LAST 107 }; 108 109 enum x2apic_state { 110 X2APIC_DISABLED, 111 X2APIC_ENABLED, 112 X2APIC_STATE_LAST 113 }; 114 115 #define VM_INTINFO_VECTOR(info) ((info) & 0xff) 116 #define VM_INTINFO_DEL_ERRCODE 0x800 117 #define VM_INTINFO_RSVD 0x7ffff000 118 #define VM_INTINFO_VALID 0x80000000 119 #define VM_INTINFO_TYPE 0x700 120 #define VM_INTINFO_HWINTR (0 << 8) 121 #define VM_INTINFO_NMI (2 << 8) 122 #define VM_INTINFO_HWEXCEPTION (3 << 8) 123 #define VM_INTINFO_SWINTR (4 << 8) 124 125 /* 126 * illumos doesn't have a limitation based on SPECNAMELEN like FreeBSD does. 127 * To simplify structure definitions, an arbitrary limit has been chosen. 128 * This same limit is used for memory segment names 129 */ 130 131 #define VM_MAX_NAMELEN 128 132 #define VM_MAX_SEG_NAMELEN 128 133 134 #ifdef _KERNEL 135 #define VM_MAXCPU 32 /* maximum virtual cpus */ 136 #endif 137 138 /* 139 * Identifiers for optional vmm capabilities 140 */ 141 enum vm_cap_type { 142 VM_CAP_HALT_EXIT, 143 VM_CAP_MTRAP_EXIT, 144 VM_CAP_PAUSE_EXIT, 145 VM_CAP_ENABLE_INVPCID, 146 VM_CAP_BPT_EXIT, 147 VM_CAP_MAX 148 }; 149 150 enum vmx_caps { 151 VMX_CAP_NONE = 0, 152 VMX_CAP_TPR_SHADOW = (1UL << 0), 153 VMX_CAP_APICV = (1UL << 1), 154 VMX_CAP_APICV_X2APIC = (1UL << 2), 155 VMX_CAP_APICV_PIR = (1UL << 3), 156 }; 157 158 enum vm_intr_trigger { 159 EDGE_TRIGGER, 160 LEVEL_TRIGGER 161 }; 162 163 /* 164 * The 'access' field has the format specified in Table 21-2 of the Intel 165 * Architecture Manual vol 3b. 166 * 167 * XXX The contents of the 'access' field are architecturally defined except 168 * bit 16 - Segment Unusable. 169 */ 170 struct seg_desc { 171 uint64_t base; 172 uint32_t limit; 173 uint32_t access; 174 }; 175 #define SEG_DESC_TYPE(access) ((access) & 0x001f) 176 #define SEG_DESC_DPL(access) (((access) >> 5) & 0x3) 177 #define SEG_DESC_PRESENT(access) (((access) & 0x0080) ? 1 : 0) 178 #define SEG_DESC_DEF32(access) (((access) & 0x4000) ? 1 : 0) 179 #define SEG_DESC_GRANULARITY(access) (((access) & 0x8000) ? 1 : 0) 180 #define SEG_DESC_UNUSABLE(access) (((access) & 0x10000) ? 1 : 0) 181 182 enum vm_cpu_mode { 183 CPU_MODE_REAL, 184 CPU_MODE_PROTECTED, 185 CPU_MODE_COMPATIBILITY, /* IA-32E mode (CS.L = 0) */ 186 CPU_MODE_64BIT, /* IA-32E mode (CS.L = 1) */ 187 }; 188 189 enum vm_paging_mode { 190 PAGING_MODE_FLAT, 191 PAGING_MODE_32, 192 PAGING_MODE_PAE, 193 PAGING_MODE_64, 194 }; 195 196 struct vm_guest_paging { 197 uint64_t cr3; 198 int cpl; 199 enum vm_cpu_mode cpu_mode; 200 enum vm_paging_mode paging_mode; 201 }; 202 203 enum vm_exitcode { 204 VM_EXITCODE_INOUT, 205 VM_EXITCODE_VMX, 206 VM_EXITCODE_BOGUS, 207 VM_EXITCODE_RDMSR, 208 VM_EXITCODE_WRMSR, 209 VM_EXITCODE_HLT, 210 VM_EXITCODE_MTRAP, 211 VM_EXITCODE_PAUSE, 212 VM_EXITCODE_PAGING, 213 VM_EXITCODE_INST_EMUL, 214 VM_EXITCODE_RUN_STATE, 215 VM_EXITCODE_MMIO_EMUL, 216 VM_EXITCODE_DEPRECATED, /* formerly RUNBLOCK */ 217 VM_EXITCODE_IOAPIC_EOI, 218 VM_EXITCODE_SUSPENDED, 219 VM_EXITCODE_MMIO, 220 VM_EXITCODE_TASK_SWITCH, 221 VM_EXITCODE_MONITOR, 222 VM_EXITCODE_MWAIT, 223 VM_EXITCODE_SVM, 224 VM_EXITCODE_REQIDLE, 225 VM_EXITCODE_DEBUG, 226 VM_EXITCODE_VMINSN, 227 VM_EXITCODE_BPT, 228 VM_EXITCODE_HT, 229 VM_EXITCODE_MAX 230 }; 231 232 enum inout_flags { 233 INOUT_IN = (1U << 0), /* direction: 'in' when set, else 'out' */ 234 235 /* 236 * The following flags are used only for in-kernel emulation logic and 237 * are not exposed to userspace. 238 */ 239 INOUT_STR = (1U << 1), /* ins/outs operation */ 240 INOUT_REP = (1U << 2), /* 'rep' prefix present on instruction */ 241 }; 242 243 struct vm_inout { 244 uint32_t eax; 245 uint16_t port; 246 uint8_t bytes; /* 1 or 2 or 4 */ 247 uint8_t flags; /* see: inout_flags */ 248 249 /* 250 * The address size and segment are relevant to INS/OUTS operations. 251 * Userspace is not concerned with them since the in-kernel emulation 252 * handles those specific aspects. 253 */ 254 uint8_t addrsize; 255 uint8_t segment; 256 }; 257 258 struct vm_mmio { 259 uint8_t bytes; /* 1/2/4/8 bytes */ 260 uint8_t read; /* read: 1, write: 0 */ 261 uint16_t _pad[3]; 262 uint64_t gpa; 263 uint64_t data; 264 }; 265 266 enum task_switch_reason { 267 TSR_CALL, 268 TSR_IRET, 269 TSR_JMP, 270 TSR_IDT_GATE, /* task gate in IDT */ 271 }; 272 273 struct vm_task_switch { 274 uint16_t tsssel; /* new TSS selector */ 275 int ext; /* task switch due to external event */ 276 uint32_t errcode; 277 int errcode_valid; /* push 'errcode' on the new stack */ 278 enum task_switch_reason reason; 279 struct vm_guest_paging paging; 280 }; 281 282 enum vcpu_run_state { 283 VRS_HALT = 0, 284 VRS_INIT = (1 << 0), 285 VRS_RUN = (1 << 1), 286 287 VRS_PEND_INIT = (1 << 14), 288 VRS_PEND_SIPI = (1 << 15), 289 }; 290 #define VRS_MASK_VALID(v) \ 291 ((v) & (VRS_INIT | VRS_RUN | VRS_PEND_SIPI | VRS_PEND_SIPI)) 292 #define VRS_IS_VALID(v) ((v) == VRS_MASK_VALID(v)) 293 294 struct vm_exit { 295 enum vm_exitcode exitcode; 296 int inst_length; /* 0 means unknown */ 297 uint64_t rip; 298 union { 299 struct vm_inout inout; 300 struct vm_mmio mmio; 301 struct { 302 uint64_t gpa; 303 int fault_type; 304 } paging; 305 /* 306 * Kernel-internal MMIO decoding and emulation. 307 * Userspace should not expect to see this, but rather a 308 * VM_EXITCODE_MMIO with the above 'mmio' context. 309 */ 310 struct { 311 uint64_t gpa; 312 uint64_t gla; 313 uint64_t cs_base; 314 int cs_d; /* CS.D */ 315 } mmio_emul; 316 struct { 317 uint8_t inst[15]; 318 uint8_t num_valid; 319 } inst_emul; 320 /* 321 * VMX specific payload. Used when there is no "better" 322 * exitcode to represent the VM-exit. 323 */ 324 struct { 325 int status; /* vmx inst status */ 326 /* 327 * 'exit_reason' and 'exit_qualification' are valid 328 * only if 'status' is zero. 329 */ 330 uint32_t exit_reason; 331 uint64_t exit_qualification; 332 /* 333 * 'inst_error' and 'inst_type' are valid 334 * only if 'status' is non-zero. 335 */ 336 int inst_type; 337 int inst_error; 338 } vmx; 339 /* 340 * SVM specific payload. 341 */ 342 struct { 343 uint64_t exitcode; 344 uint64_t exitinfo1; 345 uint64_t exitinfo2; 346 } svm; 347 struct { 348 int inst_length; 349 } bpt; 350 struct { 351 uint32_t code; /* ecx value */ 352 uint64_t wval; 353 } msr; 354 struct { 355 uint64_t rflags; 356 } hlt; 357 struct { 358 int vector; 359 } ioapic_eoi; 360 struct { 361 enum vm_suspend_how how; 362 } suspended; 363 struct vm_task_switch task_switch; 364 } u; 365 }; 366 367 enum vm_entry_cmds { 368 VEC_DEFAULT = 0, 369 VEC_DISCARD_INSTR, /* discard inst emul state */ 370 VEC_FULFILL_MMIO, /* entry includes result for mmio emul */ 371 VEC_FULFILL_INOUT, /* entry includes result for inout emul */ 372 }; 373 374 struct vm_entry { 375 int cpuid; 376 uint_t cmd; /* see: vm_entry_cmds */ 377 void *exit_data; 378 union { 379 struct vm_inout inout; 380 struct vm_mmio mmio; 381 } u; 382 }; 383 384 int vm_restart_instruction(void *vm, int vcpuid); 385 386 enum vm_create_flags { 387 /* 388 * Allocate guest memory segments from existing reservoir capacity, 389 * rather than attempting to create transient allocations. 390 */ 391 VCF_RESERVOIR_MEM = (1 << 0), 392 }; 393 394 #endif /* _VMM_H_ */ 395