bhyverun.c revision 268972 
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 268972 2014-07-22 03:14:37Z jhb $
27 */
28
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD: stable/10/usr.sbin/bhyve/bhyverun.c 268972 2014-07-22 03:14:37Z 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_irq.h"
64#include "pci_lpc.h"
65#include "smbiostbl.h"
66#include "xmsr.h"
67#include "spinup_ap.h"
68#include "rtc.h"
69
70#define GUEST_NIO_PORT		0x488	/* guest upcalls via i/o port */
71
72#define	VMEXIT_CONTINUE		1	/* continue from next instruction */
73#define	VMEXIT_RESTART		2	/* restart current instruction */
74#define	VMEXIT_ABORT		3	/* abort the vm run loop */
75#define	VMEXIT_RESET		4	/* guest machine has reset */
76#define	VMEXIT_POWEROFF		5	/* guest machine has powered off */
77
78#define MB		(1024UL * 1024)
79#define GB		(1024UL * MB)
80
81typedef int (*vmexit_handler_t)(struct vmctx *, struct vm_exit *, int *vcpu);
82
83char *vmname;
84
85int guest_ncpus;
86char *guest_uuid_str;
87
88static int guest_vmexit_on_hlt, guest_vmexit_on_pause;
89static int virtio_msix = 1;
90static int x2apic_mode = 0;	/* default is xAPIC */
91
92static int strictio;
93static int strictmsr = 1;
94
95static int acpi;
96
97static char *progname;
98static const int BSP = 0;
99
100static cpuset_t cpumask;
101
102static void vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip);
103
104struct vm_exit vmexit[VM_MAXCPU];
105
106struct bhyvestats {
107        uint64_t        vmexit_bogus;
108        uint64_t        vmexit_bogus_switch;
109        uint64_t        vmexit_hlt;
110        uint64_t        vmexit_pause;
111        uint64_t        vmexit_mtrap;
112        uint64_t        vmexit_inst_emul;
113        uint64_t        cpu_switch_rotate;
114        uint64_t        cpu_switch_direct;
115        int             io_reset;
116	int		io_poweroff;
117} stats;
118
119struct mt_vmm_info {
120	pthread_t	mt_thr;
121	struct vmctx	*mt_ctx;
122	int		mt_vcpu;
123} mt_vmm_info[VM_MAXCPU];
124
125static cpuset_t *vcpumap[VM_MAXCPU] = { NULL };
126
127static void
128usage(int code)
129{
130
131        fprintf(stderr,
132                "Usage: %s [-aehwAHIPW] [-g <gdb port>] [-s <pci>] [-c vcpus]\n"
133		"       %*s [-p vcpu:hostcpu] [-m mem] [-l <lpc>] <vm>\n"
134		"       -a: local apic is in xAPIC mode (deprecated)\n"
135		"       -A: create an ACPI table\n"
136		"       -g: gdb port\n"
137		"       -c: # cpus (default 1)\n"
138		"       -C: include guest memory in core file\n"
139		"       -p: pin 'vcpu' to 'hostcpu'\n"
140		"       -H: vmexit from the guest on hlt\n"
141		"       -P: vmexit from the guest on pause\n"
142		"       -W: force virtio to use single-vector MSI\n"
143		"       -e: exit on unhandled I/O access\n"
144		"       -h: help\n"
145		"       -s: <slot,driver,configinfo> PCI slot config\n"
146		"       -l: LPC device configuration\n"
147		"       -m: memory size in MB\n"
148		"       -w: ignore unimplemented MSRs\n"
149		"       -x: local apic is in x2APIC mode\n"
150		"       -Y: disable MPtable generation\n"
151		"       -U: uuid\n",
152		progname, (int)strlen(progname), "");
153
154	exit(code);
155}
156
157static int
158pincpu_parse(const char *opt)
159{
160	int vcpu, pcpu;
161
162	if (sscanf(opt, "%d:%d", &vcpu, &pcpu) != 2) {
163		fprintf(stderr, "invalid format: %s\n", opt);
164		return (-1);
165	}
166
167	if (vcpu < 0 || vcpu >= VM_MAXCPU) {
168		fprintf(stderr, "vcpu '%d' outside valid range from 0 to %d\n",
169		    vcpu, VM_MAXCPU - 1);
170		return (-1);
171	}
172
173	if (pcpu < 0 || pcpu >= CPU_SETSIZE) {
174		fprintf(stderr, "hostcpu '%d' outside valid range from "
175		    "0 to %d\n", pcpu, CPU_SETSIZE - 1);
176		return (-1);
177	}
178
179	if (vcpumap[vcpu] == NULL) {
180		if ((vcpumap[vcpu] = malloc(sizeof(cpuset_t))) == NULL) {
181			perror("malloc");
182			return (-1);
183		}
184		CPU_ZERO(vcpumap[vcpu]);
185	}
186	CPU_SET(pcpu, vcpumap[vcpu]);
187	return (0);
188}
189
190void *
191paddr_guest2host(struct vmctx *ctx, uintptr_t gaddr, size_t len)
192{
193
194	return (vm_map_gpa(ctx, gaddr, len));
195}
196
197int
198fbsdrun_vmexit_on_pause(void)
199{
200
201	return (guest_vmexit_on_pause);
202}
203
204int
205fbsdrun_vmexit_on_hlt(void)
206{
207
208	return (guest_vmexit_on_hlt);
209}
210
211int
212fbsdrun_virtio_msix(void)
213{
214
215	return (virtio_msix);
216}
217
218static void *
219fbsdrun_start_thread(void *param)
220{
221	char tname[MAXCOMLEN + 1];
222	struct mt_vmm_info *mtp;
223	int vcpu;
224
225	mtp = param;
226	vcpu = mtp->mt_vcpu;
227
228	snprintf(tname, sizeof(tname), "vcpu %d", vcpu);
229	pthread_set_name_np(mtp->mt_thr, tname);
230
231	vm_loop(mtp->mt_ctx, vcpu, vmexit[vcpu].rip);
232
233	/* not reached */
234	exit(1);
235	return (NULL);
236}
237
238void
239fbsdrun_addcpu(struct vmctx *ctx, int fromcpu, int newcpu, uint64_t rip)
240{
241	int error;
242
243	assert(fromcpu == BSP);
244
245	CPU_SET_ATOMIC(newcpu, &cpumask);
246
247	/*
248	 * Set up the vmexit struct to allow execution to start
249	 * at the given RIP
250	 */
251	vmexit[newcpu].rip = rip;
252	vmexit[newcpu].inst_length = 0;
253
254	mt_vmm_info[newcpu].mt_ctx = ctx;
255	mt_vmm_info[newcpu].mt_vcpu = newcpu;
256
257	error = pthread_create(&mt_vmm_info[newcpu].mt_thr, NULL,
258	    fbsdrun_start_thread, &mt_vmm_info[newcpu]);
259	assert(error == 0);
260}
261
262static int
263fbsdrun_deletecpu(struct vmctx *ctx, int vcpu)
264{
265
266	if (!CPU_ISSET(vcpu, &cpumask)) {
267		fprintf(stderr, "Attempting to delete unknown cpu %d\n", vcpu);
268		exit(1);
269	}
270
271	CPU_CLR_ATOMIC(vcpu, &cpumask);
272	return (CPU_EMPTY(&cpumask));
273}
274
275static int
276vmexit_handle_notify(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu,
277		     uint32_t eax)
278{
279#if BHYVE_DEBUG
280	/*
281	 * put guest-driven debug here
282	 */
283#endif
284        return (VMEXIT_CONTINUE);
285}
286
287static int
288vmexit_inout(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
289{
290	int error;
291	int bytes, port, in, out;
292	uint32_t eax;
293	int vcpu;
294
295	vcpu = *pvcpu;
296
297	port = vme->u.inout.port;
298	bytes = vme->u.inout.bytes;
299	eax = vme->u.inout.eax;
300	in = vme->u.inout.in;
301	out = !in;
302
303	/* We don't deal with these */
304	if (vme->u.inout.string || vme->u.inout.rep)
305		return (VMEXIT_ABORT);
306
307        /* Extra-special case of host notifications */
308        if (out && port == GUEST_NIO_PORT)
309                return (vmexit_handle_notify(ctx, vme, pvcpu, eax));
310
311	error = emulate_inout(ctx, vcpu, in, port, bytes, &eax, strictio);
312	if (error == INOUT_OK && in)
313		error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RAX, eax);
314
315	switch (error) {
316	case INOUT_OK:
317		return (VMEXIT_CONTINUE);
318	case INOUT_RESET:
319		stats.io_reset++;
320		return (VMEXIT_RESET);
321	case INOUT_POWEROFF:
322		stats.io_poweroff++;
323		return (VMEXIT_POWEROFF);
324	default:
325		fprintf(stderr, "Unhandled %s%c 0x%04x\n",
326			in ? "in" : "out",
327			bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), port);
328		return (VMEXIT_ABORT);
329	}
330}
331
332static int
333vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
334{
335	uint64_t val;
336	uint32_t eax, edx;
337	int error;
338
339	val = 0;
340	error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val);
341	if (error != 0) {
342		fprintf(stderr, "rdmsr to register %#x on vcpu %d\n",
343		    vme->u.msr.code, *pvcpu);
344		if (strictmsr) {
345			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
346			assert(error == 0);
347			return (VMEXIT_RESTART);
348		}
349	}
350
351	eax = val;
352	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax);
353	assert(error == 0);
354
355	edx = val >> 32;
356	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx);
357	assert(error == 0);
358
359	return (VMEXIT_CONTINUE);
360}
361
362static int
363vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
364{
365	int error;
366
367	error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval);
368	if (error != 0) {
369		fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n",
370		    vme->u.msr.code, vme->u.msr.wval, *pvcpu);
371		if (strictmsr) {
372			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
373			assert(error == 0);
374			return (VMEXIT_RESTART);
375		}
376	}
377	return (VMEXIT_CONTINUE);
378}
379
380static int
381vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
382{
383	int newcpu;
384	int retval = VMEXIT_CONTINUE;
385
386	newcpu = spinup_ap(ctx, *pvcpu,
387			   vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip);
388
389	return (retval);
390}
391
392static int
393vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
394{
395
396	fprintf(stderr, "vm exit[%d]\n", *pvcpu);
397	fprintf(stderr, "\treason\t\tVMX\n");
398	fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip);
399	fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length);
400	fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status);
401	fprintf(stderr, "\texit_reason\t%u\n", vmexit->u.vmx.exit_reason);
402	fprintf(stderr, "\tqualification\t0x%016lx\n",
403	    vmexit->u.vmx.exit_qualification);
404	fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type);
405	fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error);
406
407	return (VMEXIT_ABORT);
408}
409
410static int
411vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
412{
413
414	stats.vmexit_bogus++;
415
416	return (VMEXIT_RESTART);
417}
418
419static int
420vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
421{
422
423	stats.vmexit_hlt++;
424
425	/*
426	 * Just continue execution with the next instruction. We use
427	 * the HLT VM exit as a way to be friendly with the host
428	 * scheduler.
429	 */
430	return (VMEXIT_CONTINUE);
431}
432
433static int
434vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
435{
436
437	stats.vmexit_pause++;
438
439	return (VMEXIT_CONTINUE);
440}
441
442static int
443vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
444{
445
446	stats.vmexit_mtrap++;
447
448	return (VMEXIT_RESTART);
449}
450
451static int
452vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
453{
454	int err;
455	stats.vmexit_inst_emul++;
456
457	err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa,
458			  &vmexit->u.inst_emul.vie);
459
460	if (err) {
461		if (err == EINVAL) {
462			fprintf(stderr,
463			    "Failed to emulate instruction at 0x%lx\n",
464			    vmexit->rip);
465		} else if (err == ESRCH) {
466			fprintf(stderr, "Unhandled memory access to 0x%lx\n",
467			    vmexit->u.inst_emul.gpa);
468		}
469
470		return (VMEXIT_ABORT);
471	}
472
473	return (VMEXIT_CONTINUE);
474}
475
476static pthread_mutex_t resetcpu_mtx = PTHREAD_MUTEX_INITIALIZER;
477static pthread_cond_t resetcpu_cond = PTHREAD_COND_INITIALIZER;
478
479static int
480vmexit_suspend(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
481{
482	enum vm_suspend_how how;
483
484	how = vmexit->u.suspended.how;
485
486	fbsdrun_deletecpu(ctx, *pvcpu);
487
488	if (*pvcpu != BSP) {
489		pthread_mutex_lock(&resetcpu_mtx);
490		pthread_cond_signal(&resetcpu_cond);
491		pthread_mutex_unlock(&resetcpu_mtx);
492		pthread_exit(NULL);
493	}
494
495	pthread_mutex_lock(&resetcpu_mtx);
496	while (!CPU_EMPTY(&cpumask)) {
497		pthread_cond_wait(&resetcpu_cond, &resetcpu_mtx);
498	}
499	pthread_mutex_unlock(&resetcpu_mtx);
500
501	switch (how) {
502	case VM_SUSPEND_RESET:
503		exit(0);
504	case VM_SUSPEND_POWEROFF:
505		exit(1);
506	case VM_SUSPEND_HALT:
507		exit(2);
508	default:
509		fprintf(stderr, "vmexit_suspend: invalid reason %d\n", how);
510		exit(100);
511	}
512	return (0);	/* NOTREACHED */
513}
514
515static vmexit_handler_t handler[VM_EXITCODE_MAX] = {
516	[VM_EXITCODE_INOUT]  = vmexit_inout,
517	[VM_EXITCODE_VMX]    = vmexit_vmx,
518	[VM_EXITCODE_BOGUS]  = vmexit_bogus,
519	[VM_EXITCODE_RDMSR]  = vmexit_rdmsr,
520	[VM_EXITCODE_WRMSR]  = vmexit_wrmsr,
521	[VM_EXITCODE_MTRAP]  = vmexit_mtrap,
522	[VM_EXITCODE_INST_EMUL] = vmexit_inst_emul,
523	[VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap,
524	[VM_EXITCODE_SUSPENDED] = vmexit_suspend
525};
526
527static void
528vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip)
529{
530	int error, rc, prevcpu;
531	enum vm_exitcode exitcode;
532	enum vm_suspend_how how;
533
534	if (vcpumap[vcpu] != NULL) {
535		error = pthread_setaffinity_np(pthread_self(),
536		    sizeof(cpuset_t), vcpumap[vcpu]);
537		assert(error == 0);
538	}
539
540	while (1) {
541		error = vm_run(ctx, vcpu, rip, &vmexit[vcpu]);
542		if (error != 0)
543			break;
544
545		prevcpu = vcpu;
546
547		exitcode = vmexit[vcpu].exitcode;
548		if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) {
549			fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n",
550			    exitcode);
551			exit(1);
552		}
553
554                rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu);
555
556		switch (rc) {
557		case VMEXIT_CONTINUE:
558                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
559			break;
560		case VMEXIT_RESTART:
561                        rip = vmexit[vcpu].rip;
562			break;
563		case VMEXIT_RESET:
564		case VMEXIT_POWEROFF:
565			if (rc == VMEXIT_RESET)
566				how = VM_SUSPEND_RESET;
567			else
568				how = VM_SUSPEND_POWEROFF;
569			error = vm_suspend(ctx, how);
570			assert(error == 0 || errno == EALREADY);
571                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
572			break;
573		case VMEXIT_ABORT:
574			abort();
575		default:
576			exit(1);
577		}
578	}
579	fprintf(stderr, "vm_run error %d, errno %d\n", error, errno);
580}
581
582static int
583num_vcpus_allowed(struct vmctx *ctx)
584{
585	int tmp, error;
586
587	error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp);
588
589	/*
590	 * The guest is allowed to spinup more than one processor only if the
591	 * UNRESTRICTED_GUEST capability is available.
592	 */
593	if (error == 0)
594		return (VM_MAXCPU);
595	else
596		return (1);
597}
598
599void
600fbsdrun_set_capabilities(struct vmctx *ctx, int cpu)
601{
602	int err, tmp;
603
604	if (fbsdrun_vmexit_on_hlt()) {
605		err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp);
606		if (err < 0) {
607			fprintf(stderr, "VM exit on HLT not supported\n");
608			exit(1);
609		}
610		vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1);
611		if (cpu == BSP)
612			handler[VM_EXITCODE_HLT] = vmexit_hlt;
613	}
614
615        if (fbsdrun_vmexit_on_pause()) {
616		/*
617		 * pause exit support required for this mode
618		 */
619		err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp);
620		if (err < 0) {
621			fprintf(stderr,
622			    "SMP mux requested, no pause support\n");
623			exit(1);
624		}
625		vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1);
626		if (cpu == BSP)
627			handler[VM_EXITCODE_PAUSE] = vmexit_pause;
628        }
629
630	if (x2apic_mode)
631		err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED);
632	else
633		err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED);
634
635	if (err) {
636		fprintf(stderr, "Unable to set x2apic state (%d)\n", err);
637		exit(1);
638	}
639
640	vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1);
641}
642
643int
644main(int argc, char *argv[])
645{
646	int c, error, gdb_port, err, bvmcons;
647	int dump_guest_memory, max_vcpus, mptgen;
648	struct vmctx *ctx;
649	uint64_t rip;
650	size_t memsize;
651
652	bvmcons = 0;
653	dump_guest_memory = 0;
654	progname = basename(argv[0]);
655	gdb_port = 0;
656	guest_ncpus = 1;
657	memsize = 256 * MB;
658	mptgen = 1;
659
660	while ((c = getopt(argc, argv, "abehwxACHIPWYp:g:c:s:m:l:U:")) != -1) {
661		switch (c) {
662		case 'a':
663			x2apic_mode = 0;
664			break;
665		case 'A':
666			acpi = 1;
667			break;
668		case 'b':
669			bvmcons = 1;
670			break;
671		case 'p':
672                        if (pincpu_parse(optarg) != 0) {
673                            errx(EX_USAGE, "invalid vcpu pinning "
674                                 "configuration '%s'", optarg);
675                        }
676			break;
677                case 'c':
678			guest_ncpus = atoi(optarg);
679			break;
680		case 'C':
681			dump_guest_memory = 1;
682			break;
683		case 'g':
684			gdb_port = atoi(optarg);
685			break;
686		case 'l':
687			if (lpc_device_parse(optarg) != 0) {
688				errx(EX_USAGE, "invalid lpc device "
689				    "configuration '%s'", optarg);
690			}
691			break;
692		case 's':
693			if (pci_parse_slot(optarg) != 0)
694				exit(1);
695			else
696				break;
697                case 'm':
698			error = vm_parse_memsize(optarg, &memsize);
699			if (error)
700				errx(EX_USAGE, "invalid memsize '%s'", optarg);
701			break;
702		case 'H':
703			guest_vmexit_on_hlt = 1;
704			break;
705		case 'I':
706			/*
707			 * The "-I" option was used to add an ioapic to the
708			 * virtual machine.
709			 *
710			 * An ioapic is now provided unconditionally for each
711			 * virtual machine and this option is now deprecated.
712			 */
713			break;
714		case 'P':
715			guest_vmexit_on_pause = 1;
716			break;
717		case 'e':
718			strictio = 1;
719			break;
720		case 'U':
721			guest_uuid_str = optarg;
722			break;
723		case 'w':
724			strictmsr = 0;
725			break;
726		case 'W':
727			virtio_msix = 0;
728			break;
729		case 'x':
730			x2apic_mode = 1;
731			break;
732		case 'Y':
733			mptgen = 0;
734			break;
735		case 'h':
736			usage(0);
737		default:
738			usage(1);
739		}
740	}
741	argc -= optind;
742	argv += optind;
743
744	if (argc != 1)
745		usage(1);
746
747	vmname = argv[0];
748
749	ctx = vm_open(vmname);
750	if (ctx == NULL) {
751		perror("vm_open");
752		exit(1);
753	}
754
755	max_vcpus = num_vcpus_allowed(ctx);
756	if (guest_ncpus > max_vcpus) {
757		fprintf(stderr, "%d vCPUs requested but only %d available\n",
758			guest_ncpus, max_vcpus);
759		exit(1);
760	}
761
762	fbsdrun_set_capabilities(ctx, BSP);
763
764	if (dump_guest_memory)
765		vm_set_memflags(ctx, VM_MEM_F_INCORE);
766	err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL);
767	if (err) {
768		fprintf(stderr, "Unable to setup memory (%d)\n", err);
769		exit(1);
770	}
771
772	init_mem();
773	init_inout();
774	pci_irq_init(ctx);
775	ioapic_init(ctx);
776
777	rtc_init(ctx);
778	sci_init(ctx);
779
780	/*
781	 * Exit if a device emulation finds an error in it's initilization
782	 */
783	if (init_pci(ctx) != 0)
784		exit(1);
785
786	if (gdb_port != 0)
787		init_dbgport(gdb_port);
788
789	if (bvmcons)
790		init_bvmcons();
791
792	error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip);
793	assert(error == 0);
794
795	/*
796	 * build the guest tables, MP etc.
797	 */
798	if (mptgen) {
799		error = mptable_build(ctx, guest_ncpus);
800		if (error)
801			exit(1);
802	}
803
804	error = smbios_build(ctx);
805	assert(error == 0);
806
807	if (acpi) {
808		error = acpi_build(ctx, guest_ncpus);
809		assert(error == 0);
810	}
811
812	/*
813	 * Change the proc title to include the VM name.
814	 */
815	setproctitle("%s", vmname);
816
817	/*
818	 * Add CPU 0
819	 */
820	fbsdrun_addcpu(ctx, BSP, BSP, rip);
821
822	/*
823	 * Head off to the main event dispatch loop
824	 */
825	mevent_dispatch();
826
827	exit(1);
828}
829