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 2018 Joyent, Inc. 42 */ 43 44 #include <sys/cdefs.h> 45 __FBSDID("$FreeBSD$"); 46 47 #include <sys/types.h> 48 #ifndef WITHOUT_CAPSICUM 49 #include <sys/capsicum.h> 50 #endif 51 #include <sys/mman.h> 52 #include <sys/time.h> 53 #include <sys/cpuset.h> 54 55 #ifdef __FreeBSD__ 56 #include <amd64/vmm/intel/vmcs.h> 57 #else 58 #include <intel/vmcs.h> 59 #endif 60 61 #include <machine/atomic.h> 62 #include <machine/segments.h> 63 64 #ifndef WITHOUT_CAPSICUM 65 #include <capsicum_helpers.h> 66 #endif 67 #include <stdio.h> 68 #include <stdlib.h> 69 #include <string.h> 70 #include <err.h> 71 #include <errno.h> 72 #include <libgen.h> 73 #include <unistd.h> 74 #include <assert.h> 75 #include <pthread.h> 76 #include <pthread_np.h> 77 #include <sysexits.h> 78 #include <stdbool.h> 79 #include <stdint.h> 80 81 #include <machine/vmm.h> 82 #ifndef WITHOUT_CAPSICUM 83 #include <machine/vmm_dev.h> 84 #endif 85 #include <vmmapi.h> 86 87 #include "bhyverun.h" 88 #include "acpi.h" 89 #include "atkbdc.h" 90 #include "console.h" 91 #include "bootrom.h" 92 #include "inout.h" 93 #include "dbgport.h" 94 #include "debug.h" 95 #include "fwctl.h" 96 #include "gdb.h" 97 #include "ioapic.h" 98 #include "kernemu_dev.h" 99 #include "mem.h" 100 #include "mevent.h" 101 #include "mptbl.h" 102 #include "pci_emul.h" 103 #include "pci_irq.h" 104 #include "pci_lpc.h" 105 #include "smbiostbl.h" 106 #include "xmsr.h" 107 #include "spinup_ap.h" 108 #include "rfb.h" 109 #include "rtc.h" 110 #include "vga.h" 111 #include "vmgenc.h" 112 113 #define GUEST_NIO_PORT 0x488 /* guest upcalls via i/o port */ 114 115 #define MB (1024UL * 1024) 116 #define GB (1024UL * MB) 117 118 static const char * const vmx_exit_reason_desc[] = { 119 [EXIT_REASON_EXCEPTION] = "Exception or non-maskable interrupt (NMI)", 120 [EXIT_REASON_EXT_INTR] = "External interrupt", 121 [EXIT_REASON_TRIPLE_FAULT] = "Triple fault", 122 [EXIT_REASON_INIT] = "INIT signal", 123 [EXIT_REASON_SIPI] = "Start-up IPI (SIPI)", 124 [EXIT_REASON_IO_SMI] = "I/O system-management interrupt (SMI)", 125 [EXIT_REASON_SMI] = "Other SMI", 126 [EXIT_REASON_INTR_WINDOW] = "Interrupt window", 127 [EXIT_REASON_NMI_WINDOW] = "NMI window", 128 [EXIT_REASON_TASK_SWITCH] = "Task switch", 129 [EXIT_REASON_CPUID] = "CPUID", 130 [EXIT_REASON_GETSEC] = "GETSEC", 131 [EXIT_REASON_HLT] = "HLT", 132 [EXIT_REASON_INVD] = "INVD", 133 [EXIT_REASON_INVLPG] = "INVLPG", 134 [EXIT_REASON_RDPMC] = "RDPMC", 135 [EXIT_REASON_RDTSC] = "RDTSC", 136 [EXIT_REASON_RSM] = "RSM", 137 [EXIT_REASON_VMCALL] = "VMCALL", 138 [EXIT_REASON_VMCLEAR] = "VMCLEAR", 139 [EXIT_REASON_VMLAUNCH] = "VMLAUNCH", 140 [EXIT_REASON_VMPTRLD] = "VMPTRLD", 141 [EXIT_REASON_VMPTRST] = "VMPTRST", 142 [EXIT_REASON_VMREAD] = "VMREAD", 143 [EXIT_REASON_VMRESUME] = "VMRESUME", 144 [EXIT_REASON_VMWRITE] = "VMWRITE", 145 [EXIT_REASON_VMXOFF] = "VMXOFF", 146 [EXIT_REASON_VMXON] = "VMXON", 147 [EXIT_REASON_CR_ACCESS] = "Control-register accesses", 148 [EXIT_REASON_DR_ACCESS] = "MOV DR", 149 [EXIT_REASON_INOUT] = "I/O instruction", 150 [EXIT_REASON_RDMSR] = "RDMSR", 151 [EXIT_REASON_WRMSR] = "WRMSR", 152 [EXIT_REASON_INVAL_VMCS] = 153 "VM-entry failure due to invalid guest state", 154 [EXIT_REASON_INVAL_MSR] = "VM-entry failure due to MSR loading", 155 [EXIT_REASON_MWAIT] = "MWAIT", 156 [EXIT_REASON_MTF] = "Monitor trap flag", 157 [EXIT_REASON_MONITOR] = "MONITOR", 158 [EXIT_REASON_PAUSE] = "PAUSE", 159 [EXIT_REASON_MCE_DURING_ENTRY] = 160 "VM-entry failure due to machine-check event", 161 [EXIT_REASON_TPR] = "TPR below threshold", 162 [EXIT_REASON_APIC_ACCESS] = "APIC access", 163 [EXIT_REASON_VIRTUALIZED_EOI] = "Virtualized EOI", 164 [EXIT_REASON_GDTR_IDTR] = "Access to GDTR or IDTR", 165 [EXIT_REASON_LDTR_TR] = "Access to LDTR or TR", 166 [EXIT_REASON_EPT_FAULT] = "EPT violation", 167 [EXIT_REASON_EPT_MISCONFIG] = "EPT misconfiguration", 168 [EXIT_REASON_INVEPT] = "INVEPT", 169 [EXIT_REASON_RDTSCP] = "RDTSCP", 170 [EXIT_REASON_VMX_PREEMPT] = "VMX-preemption timer expired", 171 [EXIT_REASON_INVVPID] = "INVVPID", 172 [EXIT_REASON_WBINVD] = "WBINVD", 173 [EXIT_REASON_XSETBV] = "XSETBV", 174 [EXIT_REASON_APIC_WRITE] = "APIC write", 175 [EXIT_REASON_RDRAND] = "RDRAND", 176 [EXIT_REASON_INVPCID] = "INVPCID", 177 [EXIT_REASON_VMFUNC] = "VMFUNC", 178 [EXIT_REASON_ENCLS] = "ENCLS", 179 [EXIT_REASON_RDSEED] = "RDSEED", 180 [EXIT_REASON_PM_LOG_FULL] = "Page-modification log full", 181 [EXIT_REASON_XSAVES] = "XSAVES", 182 [EXIT_REASON_XRSTORS] = "XRSTORS" 183 }; 184 185 typedef int (*vmexit_handler_t)(struct vmctx *, struct vm_exit *, int *vcpu); 186 extern int vmexit_task_switch(struct vmctx *, struct vm_exit *, int *vcpu); 187 188 char *vmname; 189 190 int guest_ncpus; 191 uint16_t cores, maxcpus, sockets, threads; 192 193 char *guest_uuid_str; 194 195 int raw_stdio = 0; 196 197 static int gdb_port = 0; 198 static int guest_vmexit_on_hlt, guest_vmexit_on_pause; 199 static int virtio_msix = 1; 200 static int x2apic_mode = 0; /* default is xAPIC */ 201 202 static int strictio; 203 static int strictmsr = 1; 204 205 static int acpi; 206 207 static char *progname; 208 static const int BSP = 0; 209 210 static cpuset_t cpumask; 211 212 static void vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip); 213 214 static struct vm_exit vmexit[VM_MAXCPU]; 215 216 struct bhyvestats { 217 uint64_t vmexit_bogus; 218 uint64_t vmexit_reqidle; 219 uint64_t vmexit_hlt; 220 uint64_t vmexit_pause; 221 uint64_t vmexit_mtrap; 222 uint64_t vmexit_inst_emul; 223 uint64_t cpu_switch_rotate; 224 uint64_t cpu_switch_direct; 225 } stats; 226 227 struct mt_vmm_info { 228 pthread_t mt_thr; 229 struct vmctx *mt_ctx; 230 int mt_vcpu; 231 } mt_vmm_info[VM_MAXCPU]; 232 233 #ifdef __FreeBSD__ 234 static cpuset_t *vcpumap[VM_MAXCPU] = { NULL }; 235 #endif 236 237 static void 238 usage(int code) 239 { 240 241 fprintf(stderr, 242 "Usage: %s [-abehuwxACHPSWY]\n" 243 " %*s [-c [[cpus=]numcpus][,sockets=n][,cores=n][,threads=n]]\n" 244 " %*s [-g <gdb port>] [-l <lpc>]\n" 245 #ifdef __FreeBSD__ 246 " %*s [-m mem] [-p vcpu:hostcpu] [-s <pci>] [-U uuid] <vm>\n" 247 #else 248 " %*s [-m mem] [-s <pci>] [-U uuid] <vm>\n" 249 #endif 250 " -a: local apic is in xAPIC mode (deprecated)\n" 251 " -A: create ACPI tables\n" 252 " -c: number of cpus and/or topology specification\n" 253 " -C: include guest memory in core file\n" 254 #ifndef __FreeBSD__ 255 " -d: suspend cpu at boot\n" 256 #endif 257 " -e: exit on unhandled I/O access\n" 258 " -g: gdb port\n" 259 " -h: help\n" 260 " -H: vmexit from the guest on hlt\n" 261 " -l: LPC device configuration\n" 262 " -m: memory size\n" 263 #ifdef __FreeBSD__ 264 " -p: pin 'vcpu' to 'hostcpu'\n" 265 #endif 266 " -P: vmexit from the guest on pause\n" 267 " -s: <slot,driver,configinfo> PCI slot config\n" 268 " -S: guest memory cannot be swapped\n" 269 " -u: RTC keeps UTC time\n" 270 " -U: uuid\n" 271 " -w: ignore unimplemented MSRs\n" 272 " -W: force virtio to use single-vector MSI\n" 273 " -x: local apic is in x2APIC mode\n" 274 " -Y: disable MPtable generation\n", 275 progname, (int)strlen(progname), "", (int)strlen(progname), "", 276 (int)strlen(progname), ""); 277 278 exit(code); 279 } 280 281 /* 282 * XXX This parser is known to have the following issues: 283 * 1. It accepts null key=value tokens ",,". 284 * 2. It accepts whitespace after = and before value. 285 * 3. Values out of range of INT are silently wrapped. 286 * 4. It doesn't check non-final values. 287 * 5. The apparently bogus limits of UINT16_MAX are for future expansion. 288 * 289 * The acceptance of a null specification ('-c ""') is by design to match the 290 * manual page syntax specification, this results in a topology of 1 vCPU. 291 */ 292 static int 293 topology_parse(const char *opt) 294 { 295 uint64_t ncpus; 296 int c, chk, n, s, t, tmp; 297 char *cp, *str; 298 bool ns, scts; 299 300 c = 1, n = 1, s = 1, t = 1; 301 ns = false, scts = false; 302 str = strdup(opt); 303 if (str == NULL) 304 goto out; 305 306 while ((cp = strsep(&str, ",")) != NULL) { 307 if (sscanf(cp, "%i%n", &tmp, &chk) == 1) { 308 n = tmp; 309 ns = true; 310 } else if (sscanf(cp, "cpus=%i%n", &tmp, &chk) == 1) { 311 n = tmp; 312 ns = true; 313 } else if (sscanf(cp, "sockets=%i%n", &tmp, &chk) == 1) { 314 s = tmp; 315 scts = true; 316 } else if (sscanf(cp, "cores=%i%n", &tmp, &chk) == 1) { 317 c = tmp; 318 scts = true; 319 } else if (sscanf(cp, "threads=%i%n", &tmp, &chk) == 1) { 320 t = tmp; 321 scts = true; 322 #ifdef notyet /* Do not expose this until vmm.ko implements it */ 323 } else if (sscanf(cp, "maxcpus=%i%n", &tmp, &chk) == 1) { 324 m = tmp; 325 #endif 326 /* Skip the empty argument case from -c "" */ 327 } else if (cp[0] == '\0') 328 continue; 329 else 330 goto out; 331 /* Any trailing garbage causes an error */ 332 if (cp[chk] != '\0') 333 goto out; 334 } 335 free(str); 336 str = NULL; 337 338 /* 339 * Range check 1 <= n <= UINT16_MAX all values 340 */ 341 if (n < 1 || s < 1 || c < 1 || t < 1 || 342 n > UINT16_MAX || s > UINT16_MAX || c > UINT16_MAX || 343 t > UINT16_MAX) 344 return (-1); 345 346 /* If only the cpus was specified, use that as sockets */ 347 if (!scts) 348 s = n; 349 /* 350 * Compute sockets * cores * threads avoiding overflow 351 * The range check above insures these are 16 bit values 352 * If n was specified check it against computed ncpus 353 */ 354 ncpus = (uint64_t)s * c * t; 355 if (ncpus > UINT16_MAX || (ns && n != ncpus)) 356 return (-1); 357 358 guest_ncpus = ncpus; 359 sockets = s; 360 cores = c; 361 threads = t; 362 return(0); 363 364 out: 365 free(str); 366 return (-1); 367 } 368 369 #ifndef WITHOUT_CAPSICUM 370 /* 371 * 11-stable capsicum helpers 372 */ 373 static void 374 bhyve_caph_cache_catpages(void) 375 { 376 377 (void)catopen("libc", NL_CAT_LOCALE); 378 } 379 380 static int 381 bhyve_caph_limit_stdoe(void) 382 { 383 cap_rights_t rights; 384 unsigned long cmds[] = { TIOCGETA, TIOCGWINSZ }; 385 int i, fds[] = { STDOUT_FILENO, STDERR_FILENO }; 386 387 cap_rights_init(&rights, CAP_FCNTL, CAP_FSTAT, CAP_IOCTL); 388 cap_rights_set(&rights, CAP_WRITE); 389 390 for (i = 0; i < nitems(fds); i++) { 391 if (cap_rights_limit(fds[i], &rights) < 0 && errno != ENOSYS) 392 return (-1); 393 394 if (cap_ioctls_limit(fds[i], cmds, nitems(cmds)) < 0 && errno != ENOSYS) 395 return (-1); 396 397 if (cap_fcntls_limit(fds[i], CAP_FCNTL_GETFL) < 0 && errno != ENOSYS) 398 return (-1); 399 } 400 401 return (0); 402 } 403 404 #endif 405 406 #ifdef __FreeBSD__ 407 static int 408 pincpu_parse(const char *opt) 409 { 410 int vcpu, pcpu; 411 412 if (sscanf(opt, "%d:%d", &vcpu, &pcpu) != 2) { 413 fprintf(stderr, "invalid format: %s\n", opt); 414 return (-1); 415 } 416 417 if (vcpu < 0 || vcpu >= VM_MAXCPU) { 418 fprintf(stderr, "vcpu '%d' outside valid range from 0 to %d\n", 419 vcpu, VM_MAXCPU - 1); 420 return (-1); 421 } 422 423 if (pcpu < 0 || pcpu >= CPU_SETSIZE) { 424 fprintf(stderr, "hostcpu '%d' outside valid range from " 425 "0 to %d\n", pcpu, CPU_SETSIZE - 1); 426 return (-1); 427 } 428 429 if (vcpumap[vcpu] == NULL) { 430 if ((vcpumap[vcpu] = malloc(sizeof(cpuset_t))) == NULL) { 431 perror("malloc"); 432 return (-1); 433 } 434 CPU_ZERO(vcpumap[vcpu]); 435 } 436 CPU_SET(pcpu, vcpumap[vcpu]); 437 return (0); 438 } 439 440 void 441 vm_inject_fault(void *arg, int vcpu, int vector, int errcode_valid, 442 int errcode) 443 { 444 struct vmctx *ctx; 445 int error, restart_instruction; 446 447 ctx = arg; 448 restart_instruction = 1; 449 450 error = vm_inject_exception(ctx, vcpu, vector, errcode_valid, errcode, 451 restart_instruction); 452 assert(error == 0); 453 } 454 #endif /* __FreeBSD__ */ 455 456 void * 457 paddr_guest2host(struct vmctx *ctx, uintptr_t gaddr, size_t len) 458 { 459 460 return (vm_map_gpa(ctx, gaddr, len)); 461 } 462 463 int 464 fbsdrun_vmexit_on_pause(void) 465 { 466 467 return (guest_vmexit_on_pause); 468 } 469 470 int 471 fbsdrun_vmexit_on_hlt(void) 472 { 473 474 return (guest_vmexit_on_hlt); 475 } 476 477 int 478 fbsdrun_virtio_msix(void) 479 { 480 481 return (virtio_msix); 482 } 483 484 static void * 485 fbsdrun_start_thread(void *param) 486 { 487 char tname[MAXCOMLEN + 1]; 488 struct mt_vmm_info *mtp; 489 int vcpu; 490 491 mtp = param; 492 vcpu = mtp->mt_vcpu; 493 494 snprintf(tname, sizeof(tname), "vcpu %d", vcpu); 495 pthread_set_name_np(mtp->mt_thr, tname); 496 497 if (gdb_port != 0) 498 gdb_cpu_add(vcpu); 499 500 vm_loop(mtp->mt_ctx, vcpu, vmexit[vcpu].rip); 501 502 /* not reached */ 503 exit(1); 504 return (NULL); 505 } 506 507 #ifdef __FreeBSD__ 508 void 509 fbsdrun_addcpu(struct vmctx *ctx, int fromcpu, int newcpu, uint64_t rip) 510 #else 511 void 512 fbsdrun_addcpu(struct vmctx *ctx, int fromcpu, int newcpu, uint64_t rip, 513 bool suspend) 514 #endif 515 { 516 int error; 517 518 assert(fromcpu == BSP); 519 520 /* 521 * The 'newcpu' must be activated in the context of 'fromcpu'. If 522 * vm_activate_cpu() is delayed until newcpu's pthread starts running 523 * then vmm.ko is out-of-sync with bhyve and this can create a race 524 * with vm_suspend(). 525 */ 526 error = vm_activate_cpu(ctx, newcpu); 527 if (error != 0) 528 err(EX_OSERR, "could not activate CPU %d", newcpu); 529 530 CPU_SET_ATOMIC(newcpu, &cpumask); 531 532 #ifndef __FreeBSD__ 533 if (suspend) 534 (void) vm_suspend_cpu(ctx, newcpu); 535 #endif 536 537 /* 538 * Set up the vmexit struct to allow execution to start 539 * at the given RIP 540 */ 541 vmexit[newcpu].rip = rip; 542 vmexit[newcpu].inst_length = 0; 543 544 mt_vmm_info[newcpu].mt_ctx = ctx; 545 mt_vmm_info[newcpu].mt_vcpu = newcpu; 546 547 error = pthread_create(&mt_vmm_info[newcpu].mt_thr, NULL, 548 fbsdrun_start_thread, &mt_vmm_info[newcpu]); 549 assert(error == 0); 550 } 551 552 static int 553 fbsdrun_deletecpu(struct vmctx *ctx, int vcpu) 554 { 555 556 if (!CPU_ISSET(vcpu, &cpumask)) { 557 fprintf(stderr, "Attempting to delete unknown cpu %d\n", vcpu); 558 exit(4); 559 } 560 561 CPU_CLR_ATOMIC(vcpu, &cpumask); 562 return (CPU_EMPTY(&cpumask)); 563 } 564 565 static int 566 vmexit_handle_notify(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu, 567 uint32_t eax) 568 { 569 #if BHYVE_DEBUG 570 /* 571 * put guest-driven debug here 572 */ 573 #endif 574 return (VMEXIT_CONTINUE); 575 } 576 577 static int 578 vmexit_inout(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 579 { 580 int error; 581 int bytes, port, in, out; 582 int vcpu; 583 584 vcpu = *pvcpu; 585 586 port = vme->u.inout.port; 587 bytes = vme->u.inout.bytes; 588 in = vme->u.inout.in; 589 out = !in; 590 591 /* Extra-special case of host notifications */ 592 if (out && port == GUEST_NIO_PORT) { 593 error = vmexit_handle_notify(ctx, vme, pvcpu, vme->u.inout.eax); 594 return (error); 595 } 596 597 error = emulate_inout(ctx, vcpu, vme, strictio); 598 if (error) { 599 fprintf(stderr, "Unhandled %s%c 0x%04x at 0x%lx\n", 600 in ? "in" : "out", 601 bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), 602 port, vmexit->rip); 603 return (VMEXIT_ABORT); 604 } else { 605 return (VMEXIT_CONTINUE); 606 } 607 } 608 609 static int 610 vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 611 { 612 uint64_t val; 613 uint32_t eax, edx; 614 int error; 615 616 val = 0; 617 error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val); 618 if (error != 0) { 619 fprintf(stderr, "rdmsr to register %#x on vcpu %d\n", 620 vme->u.msr.code, *pvcpu); 621 if (strictmsr) { 622 vm_inject_gp(ctx, *pvcpu); 623 return (VMEXIT_CONTINUE); 624 } 625 } 626 627 eax = val; 628 error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax); 629 assert(error == 0); 630 631 edx = val >> 32; 632 error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx); 633 assert(error == 0); 634 635 return (VMEXIT_CONTINUE); 636 } 637 638 static int 639 vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 640 { 641 int error; 642 643 error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval); 644 if (error != 0) { 645 fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n", 646 vme->u.msr.code, vme->u.msr.wval, *pvcpu); 647 if (strictmsr) { 648 vm_inject_gp(ctx, *pvcpu); 649 return (VMEXIT_CONTINUE); 650 } 651 } 652 return (VMEXIT_CONTINUE); 653 } 654 655 static int 656 vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 657 { 658 659 (void)spinup_ap(ctx, *pvcpu, 660 vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip); 661 662 return (VMEXIT_CONTINUE); 663 } 664 665 #define DEBUG_EPT_MISCONFIG 666 #ifdef DEBUG_EPT_MISCONFIG 667 #define VMCS_GUEST_PHYSICAL_ADDRESS 0x00002400 668 669 static uint64_t ept_misconfig_gpa, ept_misconfig_pte[4]; 670 static int ept_misconfig_ptenum; 671 #endif 672 673 static const char * 674 vmexit_vmx_desc(uint32_t exit_reason) 675 { 676 677 if (exit_reason >= nitems(vmx_exit_reason_desc) || 678 vmx_exit_reason_desc[exit_reason] == NULL) 679 return ("Unknown"); 680 return (vmx_exit_reason_desc[exit_reason]); 681 } 682 683 static int 684 vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 685 { 686 687 fprintf(stderr, "vm exit[%d]\n", *pvcpu); 688 fprintf(stderr, "\treason\t\tVMX\n"); 689 fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip); 690 fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length); 691 fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status); 692 fprintf(stderr, "\texit_reason\t%u (%s)\n", vmexit->u.vmx.exit_reason, 693 vmexit_vmx_desc(vmexit->u.vmx.exit_reason)); 694 fprintf(stderr, "\tqualification\t0x%016lx\n", 695 vmexit->u.vmx.exit_qualification); 696 fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type); 697 fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error); 698 #ifdef DEBUG_EPT_MISCONFIG 699 if (vmexit->u.vmx.exit_reason == EXIT_REASON_EPT_MISCONFIG) { 700 vm_get_register(ctx, *pvcpu, 701 VMCS_IDENT(VMCS_GUEST_PHYSICAL_ADDRESS), 702 &ept_misconfig_gpa); 703 vm_get_gpa_pmap(ctx, ept_misconfig_gpa, ept_misconfig_pte, 704 &ept_misconfig_ptenum); 705 fprintf(stderr, "\tEPT misconfiguration:\n"); 706 fprintf(stderr, "\t\tGPA: %#lx\n", ept_misconfig_gpa); 707 fprintf(stderr, "\t\tPTE(%d): %#lx %#lx %#lx %#lx\n", 708 ept_misconfig_ptenum, ept_misconfig_pte[0], 709 ept_misconfig_pte[1], ept_misconfig_pte[2], 710 ept_misconfig_pte[3]); 711 } 712 #endif /* DEBUG_EPT_MISCONFIG */ 713 return (VMEXIT_ABORT); 714 } 715 716 static int 717 vmexit_svm(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 718 { 719 720 fprintf(stderr, "vm exit[%d]\n", *pvcpu); 721 fprintf(stderr, "\treason\t\tSVM\n"); 722 fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip); 723 fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length); 724 fprintf(stderr, "\texitcode\t%#lx\n", vmexit->u.svm.exitcode); 725 fprintf(stderr, "\texitinfo1\t%#lx\n", vmexit->u.svm.exitinfo1); 726 fprintf(stderr, "\texitinfo2\t%#lx\n", vmexit->u.svm.exitinfo2); 727 return (VMEXIT_ABORT); 728 } 729 730 static int 731 vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 732 { 733 734 assert(vmexit->inst_length == 0); 735 736 stats.vmexit_bogus++; 737 738 return (VMEXIT_CONTINUE); 739 } 740 741 static int 742 vmexit_reqidle(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 743 { 744 745 assert(vmexit->inst_length == 0); 746 747 stats.vmexit_reqidle++; 748 749 return (VMEXIT_CONTINUE); 750 } 751 752 static int 753 vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 754 { 755 756 stats.vmexit_hlt++; 757 758 /* 759 * Just continue execution with the next instruction. We use 760 * the HLT VM exit as a way to be friendly with the host 761 * scheduler. 762 */ 763 return (VMEXIT_CONTINUE); 764 } 765 766 static int 767 vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 768 { 769 770 stats.vmexit_pause++; 771 772 return (VMEXIT_CONTINUE); 773 } 774 775 static int 776 vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 777 { 778 779 assert(vmexit->inst_length == 0); 780 781 stats.vmexit_mtrap++; 782 783 if (gdb_port == 0) { 784 fprintf(stderr, "vm_loop: unexpected VMEXIT_MTRAP\n"); 785 exit(4); 786 } 787 gdb_cpu_mtrap(*pvcpu); 788 return (VMEXIT_CONTINUE); 789 } 790 791 static int 792 vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 793 { 794 int err, i; 795 struct vie *vie; 796 797 stats.vmexit_inst_emul++; 798 799 vie = &vmexit->u.inst_emul.vie; 800 err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa, 801 vie, &vmexit->u.inst_emul.paging); 802 803 if (err) { 804 if (err == ESRCH) { 805 EPRINTLN("Unhandled memory access to 0x%lx\n", 806 vmexit->u.inst_emul.gpa); 807 } 808 809 fprintf(stderr, "Failed to emulate instruction sequence [ "); 810 for (i = 0; i < vie->num_valid; i++) 811 fprintf(stderr, "%02x", vie->inst[i]); 812 FPRINTLN(stderr, " ] at 0x%lx", vmexit->rip); 813 return (VMEXIT_ABORT); 814 } 815 816 return (VMEXIT_CONTINUE); 817 } 818 819 static pthread_mutex_t resetcpu_mtx = PTHREAD_MUTEX_INITIALIZER; 820 static pthread_cond_t resetcpu_cond = PTHREAD_COND_INITIALIZER; 821 822 static int 823 vmexit_suspend(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 824 { 825 enum vm_suspend_how how; 826 827 how = vmexit->u.suspended.how; 828 829 fbsdrun_deletecpu(ctx, *pvcpu); 830 831 if (*pvcpu != BSP) { 832 pthread_mutex_lock(&resetcpu_mtx); 833 pthread_cond_signal(&resetcpu_cond); 834 pthread_mutex_unlock(&resetcpu_mtx); 835 pthread_exit(NULL); 836 } 837 838 pthread_mutex_lock(&resetcpu_mtx); 839 while (!CPU_EMPTY(&cpumask)) { 840 pthread_cond_wait(&resetcpu_cond, &resetcpu_mtx); 841 } 842 pthread_mutex_unlock(&resetcpu_mtx); 843 844 switch (how) { 845 case VM_SUSPEND_RESET: 846 exit(0); 847 case VM_SUSPEND_POWEROFF: 848 exit(1); 849 case VM_SUSPEND_HALT: 850 exit(2); 851 case VM_SUSPEND_TRIPLEFAULT: 852 exit(3); 853 default: 854 fprintf(stderr, "vmexit_suspend: invalid reason %d\n", how); 855 exit(100); 856 } 857 return (0); /* NOTREACHED */ 858 } 859 860 static int 861 vmexit_debug(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 862 { 863 864 if (gdb_port == 0) { 865 fprintf(stderr, "vm_loop: unexpected VMEXIT_DEBUG\n"); 866 exit(4); 867 } 868 gdb_cpu_suspend(*pvcpu); 869 return (VMEXIT_CONTINUE); 870 } 871 872 static int 873 vmexit_breakpoint(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 874 { 875 876 if (gdb_port == 0) { 877 fprintf(stderr, "vm_loop: unexpected VMEXIT_DEBUG\n"); 878 exit(4); 879 } 880 gdb_cpu_breakpoint(*pvcpu, vmexit); 881 return (VMEXIT_CONTINUE); 882 } 883 884 static vmexit_handler_t handler[VM_EXITCODE_MAX] = { 885 [VM_EXITCODE_INOUT] = vmexit_inout, 886 [VM_EXITCODE_INOUT_STR] = vmexit_inout, 887 [VM_EXITCODE_VMX] = vmexit_vmx, 888 [VM_EXITCODE_SVM] = vmexit_svm, 889 [VM_EXITCODE_BOGUS] = vmexit_bogus, 890 [VM_EXITCODE_REQIDLE] = vmexit_reqidle, 891 [VM_EXITCODE_RDMSR] = vmexit_rdmsr, 892 [VM_EXITCODE_WRMSR] = vmexit_wrmsr, 893 [VM_EXITCODE_MTRAP] = vmexit_mtrap, 894 [VM_EXITCODE_INST_EMUL] = vmexit_inst_emul, 895 [VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap, 896 [VM_EXITCODE_SUSPENDED] = vmexit_suspend, 897 [VM_EXITCODE_TASK_SWITCH] = vmexit_task_switch, 898 [VM_EXITCODE_DEBUG] = vmexit_debug, 899 [VM_EXITCODE_BPT] = vmexit_breakpoint, 900 }; 901 902 static void 903 vm_loop(struct vmctx *ctx, int vcpu, uint64_t startrip) 904 { 905 int error, rc; 906 enum vm_exitcode exitcode; 907 cpuset_t active_cpus; 908 909 #ifdef __FreeBSD__ 910 if (vcpumap[vcpu] != NULL) { 911 error = pthread_setaffinity_np(pthread_self(), 912 sizeof(cpuset_t), vcpumap[vcpu]); 913 assert(error == 0); 914 } 915 #endif 916 error = vm_active_cpus(ctx, &active_cpus); 917 assert(CPU_ISSET(vcpu, &active_cpus)); 918 919 error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RIP, startrip); 920 assert(error == 0); 921 922 while (1) { 923 error = vm_run(ctx, vcpu, &vmexit[vcpu]); 924 if (error != 0) 925 break; 926 927 exitcode = vmexit[vcpu].exitcode; 928 if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) { 929 fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n", 930 exitcode); 931 exit(4); 932 } 933 934 rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu); 935 936 switch (rc) { 937 case VMEXIT_CONTINUE: 938 break; 939 case VMEXIT_ABORT: 940 abort(); 941 default: 942 exit(4); 943 } 944 } 945 fprintf(stderr, "vm_run error %d, errno %d\n", error, errno); 946 } 947 948 static int 949 num_vcpus_allowed(struct vmctx *ctx) 950 { 951 int tmp, error; 952 953 error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp); 954 955 /* 956 * The guest is allowed to spinup more than one processor only if the 957 * UNRESTRICTED_GUEST capability is available. 958 */ 959 if (error == 0) 960 return (VM_MAXCPU); 961 else 962 return (1); 963 } 964 965 void 966 fbsdrun_set_capabilities(struct vmctx *ctx, int cpu) 967 { 968 int err, tmp; 969 970 if (fbsdrun_vmexit_on_hlt()) { 971 err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp); 972 if (err < 0) { 973 fprintf(stderr, "VM exit on HLT not supported\n"); 974 exit(4); 975 } 976 vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1); 977 if (cpu == BSP) 978 handler[VM_EXITCODE_HLT] = vmexit_hlt; 979 } 980 981 if (fbsdrun_vmexit_on_pause()) { 982 /* 983 * pause exit support required for this mode 984 */ 985 err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp); 986 if (err < 0) { 987 fprintf(stderr, 988 "SMP mux requested, no pause support\n"); 989 exit(4); 990 } 991 vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1); 992 if (cpu == BSP) 993 handler[VM_EXITCODE_PAUSE] = vmexit_pause; 994 } 995 996 if (x2apic_mode) 997 err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED); 998 else 999 err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED); 1000 1001 if (err) { 1002 fprintf(stderr, "Unable to set x2apic state (%d)\n", err); 1003 exit(4); 1004 } 1005 1006 #ifdef __FreeBSD__ 1007 vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1); 1008 #endif 1009 } 1010 1011 static struct vmctx * 1012 do_open(const char *vmname) 1013 { 1014 struct vmctx *ctx; 1015 int error; 1016 bool reinit, romboot; 1017 #ifndef WITHOUT_CAPSICUM 1018 cap_rights_t rights; 1019 const cap_ioctl_t *cmds; 1020 size_t ncmds; 1021 #endif 1022 1023 reinit = romboot = false; 1024 1025 if (lpc_bootrom()) 1026 romboot = true; 1027 1028 error = vm_create(vmname); 1029 if (error) { 1030 if (errno == EEXIST) { 1031 if (romboot) { 1032 reinit = true; 1033 } else { 1034 /* 1035 * The virtual machine has been setup by the 1036 * userspace bootloader. 1037 */ 1038 } 1039 } else { 1040 perror("vm_create"); 1041 exit(4); 1042 } 1043 } else { 1044 if (!romboot) { 1045 /* 1046 * If the virtual machine was just created then a 1047 * bootrom must be configured to boot it. 1048 */ 1049 fprintf(stderr, "virtual machine cannot be booted\n"); 1050 exit(4); 1051 } 1052 } 1053 1054 ctx = vm_open(vmname); 1055 if (ctx == NULL) { 1056 perror("vm_open"); 1057 exit(4); 1058 } 1059 1060 #ifndef WITHOUT_CAPSICUM 1061 cap_rights_init(&rights, CAP_IOCTL, CAP_MMAP_RW); 1062 if (caph_rights_limit(vm_get_device_fd(ctx), &rights) == -1) 1063 errx(EX_OSERR, "Unable to apply rights for sandbox"); 1064 vm_get_ioctls(&ncmds); 1065 cmds = vm_get_ioctls(NULL); 1066 if (cmds == NULL) 1067 errx(EX_OSERR, "out of memory"); 1068 if (caph_ioctls_limit(vm_get_device_fd(ctx), cmds, ncmds) == -1) 1069 errx(EX_OSERR, "Unable to apply rights for sandbox"); 1070 free((cap_ioctl_t *)cmds); 1071 #endif 1072 1073 if (reinit) { 1074 error = vm_reinit(ctx); 1075 if (error) { 1076 perror("vm_reinit"); 1077 exit(4); 1078 } 1079 } 1080 error = vm_set_topology(ctx, sockets, cores, threads, maxcpus); 1081 if (error) 1082 errx(EX_OSERR, "vm_set_topology"); 1083 return (ctx); 1084 } 1085 1086 int 1087 main(int argc, char *argv[]) 1088 { 1089 int c, error, dbg_port, err, bvmcons; 1090 int max_vcpus, mptgen, memflags; 1091 int rtc_localtime; 1092 bool gdb_stop; 1093 #ifndef __FreeBSD__ 1094 bool suspend = false; 1095 #endif 1096 struct vmctx *ctx; 1097 uint64_t rip; 1098 size_t memsize; 1099 char *optstr; 1100 1101 bvmcons = 0; 1102 progname = basename(argv[0]); 1103 dbg_port = 0; 1104 gdb_stop = false; 1105 guest_ncpus = 1; 1106 sockets = cores = threads = 1; 1107 maxcpus = 0; 1108 memsize = 256 * MB; 1109 mptgen = 1; 1110 rtc_localtime = 1; 1111 memflags = 0; 1112 1113 #ifdef __FreeBSD__ 1114 optstr = "abehuwxACHIPSWYp:g:G:c:s:m:l:B:U:"; 1115 #else 1116 optstr = "abdehuwxACHIPSWYg:G:c:s:m:l:B:U:"; 1117 #endif 1118 while ((c = getopt(argc, argv, optstr)) != -1) { 1119 switch (c) { 1120 case 'a': 1121 x2apic_mode = 0; 1122 break; 1123 case 'A': 1124 acpi = 1; 1125 break; 1126 case 'b': 1127 bvmcons = 1; 1128 break; 1129 case 'B': 1130 if (smbios_parse(optarg) != 0) { 1131 errx(EX_USAGE, "invalid SMBIOS " 1132 "configuration '%s'", optarg); 1133 } 1134 break; 1135 #ifndef __FreeBSD__ 1136 case 'd': 1137 suspend = true; 1138 break; 1139 #else 1140 case 'p': 1141 if (pincpu_parse(optarg) != 0) { 1142 errx(EX_USAGE, "invalid vcpu pinning " 1143 "configuration '%s'", optarg); 1144 } 1145 break; 1146 #endif 1147 case 'c': 1148 if (topology_parse(optarg) != 0) { 1149 errx(EX_USAGE, "invalid cpu topology " 1150 "'%s'", optarg); 1151 } 1152 break; 1153 case 'C': 1154 memflags |= VM_MEM_F_INCORE; 1155 break; 1156 case 'g': 1157 dbg_port = atoi(optarg); 1158 break; 1159 case 'G': 1160 if (optarg[0] == 'w') { 1161 gdb_stop = true; 1162 optarg++; 1163 } 1164 gdb_port = atoi(optarg); 1165 break; 1166 case 'l': 1167 if (strncmp(optarg, "help", strlen(optarg)) == 0) { 1168 lpc_print_supported_devices(); 1169 exit(0); 1170 } else if (lpc_device_parse(optarg) != 0) { 1171 errx(EX_USAGE, "invalid lpc device " 1172 "configuration '%s'", optarg); 1173 } 1174 break; 1175 case 's': 1176 if (strncmp(optarg, "help", strlen(optarg)) == 0) { 1177 pci_print_supported_devices(); 1178 exit(0); 1179 } else if (pci_parse_slot(optarg) != 0) 1180 exit(4); 1181 else 1182 break; 1183 case 'S': 1184 memflags |= VM_MEM_F_WIRED; 1185 break; 1186 case 'm': 1187 error = vm_parse_memsize(optarg, &memsize); 1188 if (error) 1189 errx(EX_USAGE, "invalid memsize '%s'", optarg); 1190 break; 1191 case 'H': 1192 guest_vmexit_on_hlt = 1; 1193 break; 1194 case 'I': 1195 /* 1196 * The "-I" option was used to add an ioapic to the 1197 * virtual machine. 1198 * 1199 * An ioapic is now provided unconditionally for each 1200 * virtual machine and this option is now deprecated. 1201 */ 1202 break; 1203 case 'P': 1204 guest_vmexit_on_pause = 1; 1205 break; 1206 case 'e': 1207 strictio = 1; 1208 break; 1209 case 'u': 1210 rtc_localtime = 0; 1211 break; 1212 case 'U': 1213 guest_uuid_str = optarg; 1214 break; 1215 case 'w': 1216 strictmsr = 0; 1217 break; 1218 case 'W': 1219 virtio_msix = 0; 1220 break; 1221 case 'x': 1222 x2apic_mode = 1; 1223 break; 1224 case 'Y': 1225 mptgen = 0; 1226 break; 1227 case 'h': 1228 usage(0); 1229 default: 1230 usage(1); 1231 } 1232 } 1233 argc -= optind; 1234 argv += optind; 1235 1236 if (argc != 1) 1237 usage(1); 1238 1239 vmname = argv[0]; 1240 ctx = do_open(vmname); 1241 1242 max_vcpus = num_vcpus_allowed(ctx); 1243 if (guest_ncpus > max_vcpus) { 1244 fprintf(stderr, "%d vCPUs requested but only %d available\n", 1245 guest_ncpus, max_vcpus); 1246 exit(4); 1247 } 1248 1249 fbsdrun_set_capabilities(ctx, BSP); 1250 1251 vm_set_memflags(ctx, memflags); 1252 #ifdef __FreeBSD__ 1253 err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL); 1254 #else 1255 do { 1256 errno = 0; 1257 err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL); 1258 error = errno; 1259 if (err != 0 && error == ENOMEM) { 1260 (void) fprintf(stderr, "Unable to allocate memory " 1261 "(%llu), retrying in 1 second\n", memsize); 1262 sleep(1); 1263 } 1264 } while (error == ENOMEM); 1265 #endif 1266 if (err) { 1267 fprintf(stderr, "Unable to setup memory (%d)\n", errno); 1268 exit(4); 1269 } 1270 1271 error = init_msr(); 1272 if (error) { 1273 fprintf(stderr, "init_msr error %d", error); 1274 exit(4); 1275 } 1276 1277 init_mem(); 1278 init_inout(); 1279 #ifdef __FreeBSD__ 1280 kernemu_dev_init(); 1281 #endif 1282 init_bootrom(ctx); 1283 atkbdc_init(ctx); 1284 pci_irq_init(ctx); 1285 ioapic_init(ctx); 1286 1287 rtc_init(ctx, rtc_localtime); 1288 sci_init(ctx); 1289 1290 /* 1291 * Exit if a device emulation finds an error in its initilization 1292 */ 1293 if (init_pci(ctx) != 0) { 1294 perror("device emulation initialization error"); 1295 exit(4); 1296 } 1297 1298 /* 1299 * Initialize after PCI, to allow a bootrom file to reserve the high 1300 * region. 1301 */ 1302 if (acpi) 1303 vmgenc_init(ctx); 1304 1305 if (dbg_port != 0) 1306 init_dbgport(dbg_port); 1307 1308 if (gdb_port != 0) 1309 init_gdb(ctx, gdb_port, gdb_stop); 1310 1311 if (bvmcons) 1312 init_bvmcons(); 1313 1314 vga_init(1); 1315 1316 if (lpc_bootrom()) { 1317 if (vm_set_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, 1)) { 1318 fprintf(stderr, "ROM boot failed: unrestricted guest " 1319 "capability not available\n"); 1320 exit(4); 1321 } 1322 error = vcpu_reset(ctx, BSP); 1323 assert(error == 0); 1324 } 1325 1326 error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip); 1327 assert(error == 0); 1328 1329 /* 1330 * build the guest tables, MP etc. 1331 */ 1332 if (mptgen) { 1333 error = mptable_build(ctx, guest_ncpus); 1334 if (error) { 1335 perror("error to build the guest tables"); 1336 exit(4); 1337 } 1338 } 1339 1340 error = smbios_build(ctx); 1341 assert(error == 0); 1342 1343 if (acpi) { 1344 error = acpi_build(ctx, guest_ncpus); 1345 assert(error == 0); 1346 } 1347 1348 if (lpc_bootrom()) 1349 fwctl_init(); 1350 1351 /* 1352 * Change the proc title to include the VM name. 1353 */ 1354 setproctitle("%s", vmname); 1355 1356 #ifndef WITHOUT_CAPSICUM 1357 caph_cache_catpages(); 1358 1359 if (caph_limit_stdout() == -1 || caph_limit_stderr() == -1) 1360 errx(EX_OSERR, "Unable to apply rights for sandbox"); 1361 1362 if (caph_enter() == -1) 1363 errx(EX_OSERR, "cap_enter() failed"); 1364 #endif 1365 1366 /* 1367 * Add CPU 0 1368 */ 1369 #ifdef __FreeBSD__ 1370 fbsdrun_addcpu(ctx, BSP, BSP, rip); 1371 #else 1372 fbsdrun_addcpu(ctx, BSP, BSP, rip, suspend); 1373 #endif 1374 /* 1375 * Head off to the main event dispatch loop 1376 */ 1377 mevent_dispatch(); 1378 1379 exit(4); 1380 } 1381