bhyverun.c revision 268953
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/usr.sbin/bhyve/bhyverun.c 268953 2014-07-21 19:08:02Z jhb $ 27 */ 28 29#include <sys/cdefs.h> 30__FBSDID("$FreeBSD: stable/10/usr.sbin/bhyve/bhyverun.c 268953 2014-07-21 19:08:02Z jhb $"); 31 32#include <sys/types.h> 33#include <sys/mman.h> 34#include <sys/time.h> 35 36#include <machine/atomic.h> 37#include <machine/segments.h> 38 39#include <stdio.h> 40#include <stdlib.h> 41#include <string.h> 42#include <err.h> 43#include <libgen.h> 44#include <unistd.h> 45#include <assert.h> 46#include <errno.h> 47#include <pthread.h> 48#include <pthread_np.h> 49#include <sysexits.h> 50 51#include <machine/vmm.h> 52#include <vmmapi.h> 53 54#include "bhyverun.h" 55#include "acpi.h" 56#include "inout.h" 57#include "dbgport.h" 58#include "ioapic.h" 59#include "mem.h" 60#include "mevent.h" 61#include "mptbl.h" 62#include "pci_emul.h" 63#include "pci_lpc.h" 64#include "smbiostbl.h" 65#include "xmsr.h" 66#include "spinup_ap.h" 67#include "rtc.h" 68 69#define GUEST_NIO_PORT 0x488 /* guest upcalls via i/o port */ 70 71#define VMEXIT_CONTINUE 1 /* continue from next instruction */ 72#define VMEXIT_RESTART 2 /* restart current instruction */ 73#define VMEXIT_ABORT 3 /* abort the vm run loop */ 74#define VMEXIT_RESET 4 /* guest machine has reset */ 75#define VMEXIT_POWEROFF 5 /* guest machine has powered off */ 76 77#define MB (1024UL * 1024) 78#define GB (1024UL * MB) 79 80typedef int (*vmexit_handler_t)(struct vmctx *, struct vm_exit *, int *vcpu); 81 82char *vmname; 83 84int guest_ncpus; 85char *guest_uuid_str; 86 87static int guest_vmexit_on_hlt, guest_vmexit_on_pause; 88static int virtio_msix = 1; 89static int x2apic_mode = 0; /* default is xAPIC */ 90 91static int strictio; 92static int strictmsr = 1; 93 94static int acpi; 95 96static char *progname; 97static const int BSP = 0; 98 99static cpuset_t cpumask; 100 101static void vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip); 102 103struct vm_exit vmexit[VM_MAXCPU]; 104 105struct bhyvestats { 106 uint64_t vmexit_bogus; 107 uint64_t vmexit_bogus_switch; 108 uint64_t vmexit_hlt; 109 uint64_t vmexit_pause; 110 uint64_t vmexit_mtrap; 111 uint64_t vmexit_inst_emul; 112 uint64_t cpu_switch_rotate; 113 uint64_t cpu_switch_direct; 114 int io_reset; 115 int io_poweroff; 116} stats; 117 118struct mt_vmm_info { 119 pthread_t mt_thr; 120 struct vmctx *mt_ctx; 121 int mt_vcpu; 122} mt_vmm_info[VM_MAXCPU]; 123 124static cpuset_t *vcpumap[VM_MAXCPU] = { NULL }; 125 126static void 127usage(int code) 128{ 129 130 fprintf(stderr, 131 "Usage: %s [-aehwAHIPW] [-g <gdb port>] [-s <pci>] [-c vcpus]\n" 132 " %*s [-p vcpu:hostcpu] [-m mem] [-l <lpc>] <vm>\n" 133 " -a: local apic is in xAPIC mode (deprecated)\n" 134 " -A: create an ACPI table\n" 135 " -g: gdb port\n" 136 " -c: # cpus (default 1)\n" 137 " -C: include guest memory in core file\n" 138 " -p: pin 'vcpu' to 'hostcpu'\n" 139 " -H: vmexit from the guest on hlt\n" 140 " -P: vmexit from the guest on pause\n" 141 " -W: force virtio to use single-vector MSI\n" 142 " -e: exit on unhandled I/O access\n" 143 " -h: help\n" 144 " -s: <slot,driver,configinfo> PCI slot config\n" 145 " -l: LPC device configuration\n" 146 " -m: memory size in MB\n" 147 " -w: ignore unimplemented MSRs\n" 148 " -x: local apic is in x2APIC mode\n" 149 " -Y: disable MPtable generation\n" 150 " -U: uuid\n", 151 progname, (int)strlen(progname), ""); 152 153 exit(code); 154} 155 156static int 157pincpu_parse(const char *opt) 158{ 159 int vcpu, pcpu; 160 161 if (sscanf(opt, "%d:%d", &vcpu, &pcpu) != 2) { 162 fprintf(stderr, "invalid format: %s\n", opt); 163 return (-1); 164 } 165 166 if (vcpu < 0 || vcpu >= VM_MAXCPU) { 167 fprintf(stderr, "vcpu '%d' outside valid range from 0 to %d\n", 168 vcpu, VM_MAXCPU - 1); 169 return (-1); 170 } 171 172 if (pcpu < 0 || pcpu >= CPU_SETSIZE) { 173 fprintf(stderr, "hostcpu '%d' outside valid range from " 174 "0 to %d\n", pcpu, CPU_SETSIZE - 1); 175 return (-1); 176 } 177 178 if (vcpumap[vcpu] == NULL) { 179 if ((vcpumap[vcpu] = malloc(sizeof(cpuset_t))) == NULL) { 180 perror("malloc"); 181 return (-1); 182 } 183 CPU_ZERO(vcpumap[vcpu]); 184 } 185 CPU_SET(pcpu, vcpumap[vcpu]); 186 return (0); 187} 188 189void * 190paddr_guest2host(struct vmctx *ctx, uintptr_t gaddr, size_t len) 191{ 192 193 return (vm_map_gpa(ctx, gaddr, len)); 194} 195 196int 197fbsdrun_vmexit_on_pause(void) 198{ 199 200 return (guest_vmexit_on_pause); 201} 202 203int 204fbsdrun_vmexit_on_hlt(void) 205{ 206 207 return (guest_vmexit_on_hlt); 208} 209 210int 211fbsdrun_virtio_msix(void) 212{ 213 214 return (virtio_msix); 215} 216 217static void * 218fbsdrun_start_thread(void *param) 219{ 220 char tname[MAXCOMLEN + 1]; 221 struct mt_vmm_info *mtp; 222 int vcpu; 223 224 mtp = param; 225 vcpu = mtp->mt_vcpu; 226 227 snprintf(tname, sizeof(tname), "vcpu %d", vcpu); 228 pthread_set_name_np(mtp->mt_thr, tname); 229 230 vm_loop(mtp->mt_ctx, vcpu, vmexit[vcpu].rip); 231 232 /* not reached */ 233 exit(1); 234 return (NULL); 235} 236 237void 238fbsdrun_addcpu(struct vmctx *ctx, int fromcpu, int newcpu, uint64_t rip) 239{ 240 int error; 241 242 assert(fromcpu == BSP); 243 244 CPU_SET_ATOMIC(newcpu, &cpumask); 245 246 /* 247 * Set up the vmexit struct to allow execution to start 248 * at the given RIP 249 */ 250 vmexit[newcpu].rip = rip; 251 vmexit[newcpu].inst_length = 0; 252 253 mt_vmm_info[newcpu].mt_ctx = ctx; 254 mt_vmm_info[newcpu].mt_vcpu = newcpu; 255 256 error = pthread_create(&mt_vmm_info[newcpu].mt_thr, NULL, 257 fbsdrun_start_thread, &mt_vmm_info[newcpu]); 258 assert(error == 0); 259} 260 261static int 262fbsdrun_deletecpu(struct vmctx *ctx, int vcpu) 263{ 264 265 if (!CPU_ISSET(vcpu, &cpumask)) { 266 fprintf(stderr, "Attempting to delete unknown cpu %d\n", vcpu); 267 exit(1); 268 } 269 270 CPU_CLR_ATOMIC(vcpu, &cpumask); 271 return (CPU_EMPTY(&cpumask)); 272} 273 274static int 275vmexit_handle_notify(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu, 276 uint32_t eax) 277{ 278#if BHYVE_DEBUG 279 /* 280 * put guest-driven debug here 281 */ 282#endif 283 return (VMEXIT_CONTINUE); 284} 285 286static int 287vmexit_inout(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 288{ 289 int error; 290 int bytes, port, in, out; 291 uint32_t eax; 292 int vcpu; 293 294 vcpu = *pvcpu; 295 296 port = vme->u.inout.port; 297 bytes = vme->u.inout.bytes; 298 eax = vme->u.inout.eax; 299 in = vme->u.inout.in; 300 out = !in; 301 302 /* We don't deal with these */ 303 if (vme->u.inout.string || vme->u.inout.rep) 304 return (VMEXIT_ABORT); 305 306 /* Extra-special case of host notifications */ 307 if (out && port == GUEST_NIO_PORT) 308 return (vmexit_handle_notify(ctx, vme, pvcpu, eax)); 309 310 error = emulate_inout(ctx, vcpu, in, port, bytes, &eax, strictio); 311 if (error == INOUT_OK && in) 312 error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RAX, eax); 313 314 switch (error) { 315 case INOUT_OK: 316 return (VMEXIT_CONTINUE); 317 case INOUT_RESET: 318 stats.io_reset++; 319 return (VMEXIT_RESET); 320 case INOUT_POWEROFF: 321 stats.io_poweroff++; 322 return (VMEXIT_POWEROFF); 323 default: 324 fprintf(stderr, "Unhandled %s%c 0x%04x\n", 325 in ? "in" : "out", 326 bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), port); 327 return (VMEXIT_ABORT); 328 } 329} 330 331static int 332vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 333{ 334 uint64_t val; 335 uint32_t eax, edx; 336 int error; 337 338 val = 0; 339 error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val); 340 if (error != 0) { 341 fprintf(stderr, "rdmsr to register %#x on vcpu %d\n", 342 vme->u.msr.code, *pvcpu); 343 if (strictmsr) { 344 error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0); 345 assert(error == 0); 346 return (VMEXIT_RESTART); 347 } 348 } 349 350 eax = val; 351 error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax); 352 assert(error == 0); 353 354 edx = val >> 32; 355 error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx); 356 assert(error == 0); 357 358 return (VMEXIT_CONTINUE); 359} 360 361static int 362vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 363{ 364 int error; 365 366 error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval); 367 if (error != 0) { 368 fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n", 369 vme->u.msr.code, vme->u.msr.wval, *pvcpu); 370 if (strictmsr) { 371 error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0); 372 assert(error == 0); 373 return (VMEXIT_RESTART); 374 } 375 } 376 return (VMEXIT_CONTINUE); 377} 378 379static int 380vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu) 381{ 382 int newcpu; 383 int retval = VMEXIT_CONTINUE; 384 385 newcpu = spinup_ap(ctx, *pvcpu, 386 vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip); 387 388 return (retval); 389} 390 391static int 392vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 393{ 394 395 fprintf(stderr, "vm exit[%d]\n", *pvcpu); 396 fprintf(stderr, "\treason\t\tVMX\n"); 397 fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip); 398 fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length); 399 fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status); 400 fprintf(stderr, "\texit_reason\t%u\n", vmexit->u.vmx.exit_reason); 401 fprintf(stderr, "\tqualification\t0x%016lx\n", 402 vmexit->u.vmx.exit_qualification); 403 fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type); 404 fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error); 405 406 return (VMEXIT_ABORT); 407} 408 409static int 410vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 411{ 412 413 stats.vmexit_bogus++; 414 415 return (VMEXIT_RESTART); 416} 417 418static int 419vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 420{ 421 422 stats.vmexit_hlt++; 423 424 /* 425 * Just continue execution with the next instruction. We use 426 * the HLT VM exit as a way to be friendly with the host 427 * scheduler. 428 */ 429 return (VMEXIT_CONTINUE); 430} 431 432static int 433vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 434{ 435 436 stats.vmexit_pause++; 437 438 return (VMEXIT_CONTINUE); 439} 440 441static int 442vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 443{ 444 445 stats.vmexit_mtrap++; 446 447 return (VMEXIT_RESTART); 448} 449 450static int 451vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 452{ 453 int err; 454 stats.vmexit_inst_emul++; 455 456 err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa, 457 &vmexit->u.inst_emul.vie); 458 459 if (err) { 460 if (err == EINVAL) { 461 fprintf(stderr, 462 "Failed to emulate instruction at 0x%lx\n", 463 vmexit->rip); 464 } else if (err == ESRCH) { 465 fprintf(stderr, "Unhandled memory access to 0x%lx\n", 466 vmexit->u.inst_emul.gpa); 467 } 468 469 return (VMEXIT_ABORT); 470 } 471 472 return (VMEXIT_CONTINUE); 473} 474 475static pthread_mutex_t resetcpu_mtx = PTHREAD_MUTEX_INITIALIZER; 476static pthread_cond_t resetcpu_cond = PTHREAD_COND_INITIALIZER; 477 478static int 479vmexit_suspend(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu) 480{ 481 enum vm_suspend_how how; 482 483 how = vmexit->u.suspended.how; 484 485 fbsdrun_deletecpu(ctx, *pvcpu); 486 487 if (*pvcpu != BSP) { 488 pthread_mutex_lock(&resetcpu_mtx); 489 pthread_cond_signal(&resetcpu_cond); 490 pthread_mutex_unlock(&resetcpu_mtx); 491 pthread_exit(NULL); 492 } 493 494 pthread_mutex_lock(&resetcpu_mtx); 495 while (!CPU_EMPTY(&cpumask)) { 496 pthread_cond_wait(&resetcpu_cond, &resetcpu_mtx); 497 } 498 pthread_mutex_unlock(&resetcpu_mtx); 499 500 switch (how) { 501 case VM_SUSPEND_RESET: 502 exit(0); 503 case VM_SUSPEND_POWEROFF: 504 exit(1); 505 case VM_SUSPEND_HALT: 506 exit(2); 507 default: 508 fprintf(stderr, "vmexit_suspend: invalid reason %d\n", how); 509 exit(100); 510 } 511 return (0); /* NOTREACHED */ 512} 513 514static vmexit_handler_t handler[VM_EXITCODE_MAX] = { 515 [VM_EXITCODE_INOUT] = vmexit_inout, 516 [VM_EXITCODE_VMX] = vmexit_vmx, 517 [VM_EXITCODE_BOGUS] = vmexit_bogus, 518 [VM_EXITCODE_RDMSR] = vmexit_rdmsr, 519 [VM_EXITCODE_WRMSR] = vmexit_wrmsr, 520 [VM_EXITCODE_MTRAP] = vmexit_mtrap, 521 [VM_EXITCODE_INST_EMUL] = vmexit_inst_emul, 522 [VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap, 523 [VM_EXITCODE_SUSPENDED] = vmexit_suspend 524}; 525 526static void 527vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip) 528{ 529 int error, rc, prevcpu; 530 enum vm_exitcode exitcode; 531 enum vm_suspend_how how; 532 533 if (vcpumap[vcpu] != NULL) { 534 error = pthread_setaffinity_np(pthread_self(), 535 sizeof(cpuset_t), vcpumap[vcpu]); 536 assert(error == 0); 537 } 538 539 while (1) { 540 error = vm_run(ctx, vcpu, rip, &vmexit[vcpu]); 541 if (error != 0) 542 break; 543 544 prevcpu = vcpu; 545 546 exitcode = vmexit[vcpu].exitcode; 547 if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) { 548 fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n", 549 exitcode); 550 exit(1); 551 } 552 553 rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu); 554 555 switch (rc) { 556 case VMEXIT_CONTINUE: 557 rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length; 558 break; 559 case VMEXIT_RESTART: 560 rip = vmexit[vcpu].rip; 561 break; 562 case VMEXIT_RESET: 563 case VMEXIT_POWEROFF: 564 if (rc == VMEXIT_RESET) 565 how = VM_SUSPEND_RESET; 566 else 567 how = VM_SUSPEND_POWEROFF; 568 error = vm_suspend(ctx, how); 569 assert(error == 0 || errno == EALREADY); 570 rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length; 571 break; 572 case VMEXIT_ABORT: 573 abort(); 574 default: 575 exit(1); 576 } 577 } 578 fprintf(stderr, "vm_run error %d, errno %d\n", error, errno); 579} 580 581static int 582num_vcpus_allowed(struct vmctx *ctx) 583{ 584 int tmp, error; 585 586 error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp); 587 588 /* 589 * The guest is allowed to spinup more than one processor only if the 590 * UNRESTRICTED_GUEST capability is available. 591 */ 592 if (error == 0) 593 return (VM_MAXCPU); 594 else 595 return (1); 596} 597 598void 599fbsdrun_set_capabilities(struct vmctx *ctx, int cpu) 600{ 601 int err, tmp; 602 603 if (fbsdrun_vmexit_on_hlt()) { 604 err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp); 605 if (err < 0) { 606 fprintf(stderr, "VM exit on HLT not supported\n"); 607 exit(1); 608 } 609 vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1); 610 if (cpu == BSP) 611 handler[VM_EXITCODE_HLT] = vmexit_hlt; 612 } 613 614 if (fbsdrun_vmexit_on_pause()) { 615 /* 616 * pause exit support required for this mode 617 */ 618 err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp); 619 if (err < 0) { 620 fprintf(stderr, 621 "SMP mux requested, no pause support\n"); 622 exit(1); 623 } 624 vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1); 625 if (cpu == BSP) 626 handler[VM_EXITCODE_PAUSE] = vmexit_pause; 627 } 628 629 if (x2apic_mode) 630 err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED); 631 else 632 err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED); 633 634 if (err) { 635 fprintf(stderr, "Unable to set x2apic state (%d)\n", err); 636 exit(1); 637 } 638 639 vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1); 640} 641 642int 643main(int argc, char *argv[]) 644{ 645 int c, error, gdb_port, err, bvmcons; 646 int dump_guest_memory, max_vcpus, mptgen; 647 struct vmctx *ctx; 648 uint64_t rip; 649 size_t memsize; 650 651 bvmcons = 0; 652 dump_guest_memory = 0; 653 progname = basename(argv[0]); 654 gdb_port = 0; 655 guest_ncpus = 1; 656 memsize = 256 * MB; 657 mptgen = 1; 658 659 while ((c = getopt(argc, argv, "abehwxACHIPWYp:g:c:s:m:l:U:")) != -1) { 660 switch (c) { 661 case 'a': 662 x2apic_mode = 0; 663 break; 664 case 'A': 665 acpi = 1; 666 break; 667 case 'b': 668 bvmcons = 1; 669 break; 670 case 'p': 671 if (pincpu_parse(optarg) != 0) { 672 errx(EX_USAGE, "invalid vcpu pinning " 673 "configuration '%s'", optarg); 674 } 675 break; 676 case 'c': 677 guest_ncpus = atoi(optarg); 678 break; 679 case 'C': 680 dump_guest_memory = 1; 681 break; 682 case 'g': 683 gdb_port = atoi(optarg); 684 break; 685 case 'l': 686 if (lpc_device_parse(optarg) != 0) { 687 errx(EX_USAGE, "invalid lpc device " 688 "configuration '%s'", optarg); 689 } 690 break; 691 case 's': 692 if (pci_parse_slot(optarg) != 0) 693 exit(1); 694 else 695 break; 696 case 'm': 697 error = vm_parse_memsize(optarg, &memsize); 698 if (error) 699 errx(EX_USAGE, "invalid memsize '%s'", optarg); 700 break; 701 case 'H': 702 guest_vmexit_on_hlt = 1; 703 break; 704 case 'I': 705 /* 706 * The "-I" option was used to add an ioapic to the 707 * virtual machine. 708 * 709 * An ioapic is now provided unconditionally for each 710 * virtual machine and this option is now deprecated. 711 */ 712 break; 713 case 'P': 714 guest_vmexit_on_pause = 1; 715 break; 716 case 'e': 717 strictio = 1; 718 break; 719 case 'U': 720 guest_uuid_str = optarg; 721 break; 722 case 'w': 723 strictmsr = 0; 724 break; 725 case 'W': 726 virtio_msix = 0; 727 break; 728 case 'x': 729 x2apic_mode = 1; 730 break; 731 case 'Y': 732 mptgen = 0; 733 break; 734 case 'h': 735 usage(0); 736 default: 737 usage(1); 738 } 739 } 740 argc -= optind; 741 argv += optind; 742 743 if (argc != 1) 744 usage(1); 745 746 vmname = argv[0]; 747 748 ctx = vm_open(vmname); 749 if (ctx == NULL) { 750 perror("vm_open"); 751 exit(1); 752 } 753 754 max_vcpus = num_vcpus_allowed(ctx); 755 if (guest_ncpus > max_vcpus) { 756 fprintf(stderr, "%d vCPUs requested but only %d available\n", 757 guest_ncpus, max_vcpus); 758 exit(1); 759 } 760 761 fbsdrun_set_capabilities(ctx, BSP); 762 763 if (dump_guest_memory) 764 vm_set_memflags(ctx, VM_MEM_F_INCORE); 765 err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL); 766 if (err) { 767 fprintf(stderr, "Unable to setup memory (%d)\n", err); 768 exit(1); 769 } 770 771 init_mem(); 772 init_inout(); 773 ioapic_init(ctx); 774 775 rtc_init(ctx); 776 777 /* 778 * Exit if a device emulation finds an error in it's initilization 779 */ 780 if (init_pci(ctx) != 0) 781 exit(1); 782 783 if (gdb_port != 0) 784 init_dbgport(gdb_port); 785 786 if (bvmcons) 787 init_bvmcons(); 788 789 error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip); 790 assert(error == 0); 791 792 /* 793 * build the guest tables, MP etc. 794 */ 795 if (mptgen) { 796 error = mptable_build(ctx, guest_ncpus); 797 if (error) 798 exit(1); 799 } 800 801 error = smbios_build(ctx); 802 assert(error == 0); 803 804 if (acpi) { 805 error = acpi_build(ctx, guest_ncpus); 806 assert(error == 0); 807 } 808 809 /* 810 * Change the proc title to include the VM name. 811 */ 812 setproctitle("%s", vmname); 813 814 /* 815 * Add CPU 0 816 */ 817 fbsdrun_addcpu(ctx, BSP, BSP, rip); 818 819 /* 820 * Head off to the main event dispatch loop 821 */ 822 mevent_dispatch(); 823 824 exit(1); 825} 826