bhyverun.c revision 268976 
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 268976 2014-07-22 04:39:16Z jhb $
27 */
28
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD: stable/10/usr.sbin/bhyve/bhyverun.c 268976 2014-07-22 04:39:16Z 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, string;
292	int vcpu;
293
294	vcpu = *pvcpu;
295
296	port = vme->u.inout.port;
297	bytes = vme->u.inout.bytes;
298	string = vme->u.inout.string;
299	in = vme->u.inout.in;
300	out = !in;
301
302        /* Extra-special case of host notifications */
303        if (out && port == GUEST_NIO_PORT) {
304                error = vmexit_handle_notify(ctx, vme, pvcpu, vme->u.inout.eax);
305		return (error);
306	}
307
308	error = emulate_inout(ctx, vcpu, vme, strictio);
309	if (error == INOUT_OK && in && !string) {
310		error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RAX,
311		    vme->u.inout.eax);
312	}
313
314	switch (error) {
315	case INOUT_OK:
316		return (VMEXIT_CONTINUE);
317	case INOUT_RESTART:
318		return (VMEXIT_RESTART);
319	case INOUT_RESET:
320		stats.io_reset++;
321		return (VMEXIT_RESET);
322	case INOUT_POWEROFF:
323		stats.io_poweroff++;
324		return (VMEXIT_POWEROFF);
325	default:
326		fprintf(stderr, "Unhandled %s%c 0x%04x\n",
327			in ? "in" : "out",
328			bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), port);
329		return (VMEXIT_ABORT);
330	}
331}
332
333static int
334vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
335{
336	uint64_t val;
337	uint32_t eax, edx;
338	int error;
339
340	val = 0;
341	error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val);
342	if (error != 0) {
343		fprintf(stderr, "rdmsr to register %#x on vcpu %d\n",
344		    vme->u.msr.code, *pvcpu);
345		if (strictmsr) {
346			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
347			assert(error == 0);
348			return (VMEXIT_RESTART);
349		}
350	}
351
352	eax = val;
353	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax);
354	assert(error == 0);
355
356	edx = val >> 32;
357	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx);
358	assert(error == 0);
359
360	return (VMEXIT_CONTINUE);
361}
362
363static int
364vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
365{
366	int error;
367
368	error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval);
369	if (error != 0) {
370		fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n",
371		    vme->u.msr.code, vme->u.msr.wval, *pvcpu);
372		if (strictmsr) {
373			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
374			assert(error == 0);
375			return (VMEXIT_RESTART);
376		}
377	}
378	return (VMEXIT_CONTINUE);
379}
380
381static int
382vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
383{
384	int newcpu;
385	int retval = VMEXIT_CONTINUE;
386
387	newcpu = spinup_ap(ctx, *pvcpu,
388			   vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip);
389
390	return (retval);
391}
392
393static int
394vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
395{
396
397	fprintf(stderr, "vm exit[%d]\n", *pvcpu);
398	fprintf(stderr, "\treason\t\tVMX\n");
399	fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip);
400	fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length);
401	fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status);
402	fprintf(stderr, "\texit_reason\t%u\n", vmexit->u.vmx.exit_reason);
403	fprintf(stderr, "\tqualification\t0x%016lx\n",
404	    vmexit->u.vmx.exit_qualification);
405	fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type);
406	fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error);
407
408	return (VMEXIT_ABORT);
409}
410
411static int
412vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
413{
414
415	stats.vmexit_bogus++;
416
417	return (VMEXIT_RESTART);
418}
419
420static int
421vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
422{
423
424	stats.vmexit_hlt++;
425
426	/*
427	 * Just continue execution with the next instruction. We use
428	 * the HLT VM exit as a way to be friendly with the host
429	 * scheduler.
430	 */
431	return (VMEXIT_CONTINUE);
432}
433
434static int
435vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
436{
437
438	stats.vmexit_pause++;
439
440	return (VMEXIT_CONTINUE);
441}
442
443static int
444vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
445{
446
447	stats.vmexit_mtrap++;
448
449	return (VMEXIT_RESTART);
450}
451
452static int
453vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
454{
455	int err;
456	stats.vmexit_inst_emul++;
457
458	err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa,
459			  &vmexit->u.inst_emul.vie);
460
461	if (err) {
462		if (err == EINVAL) {
463			fprintf(stderr,
464			    "Failed to emulate instruction at 0x%lx\n",
465			    vmexit->rip);
466		} else if (err == ESRCH) {
467			fprintf(stderr, "Unhandled memory access to 0x%lx\n",
468			    vmexit->u.inst_emul.gpa);
469		}
470
471		return (VMEXIT_ABORT);
472	}
473
474	return (VMEXIT_CONTINUE);
475}
476
477static pthread_mutex_t resetcpu_mtx = PTHREAD_MUTEX_INITIALIZER;
478static pthread_cond_t resetcpu_cond = PTHREAD_COND_INITIALIZER;
479
480static int
481vmexit_suspend(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
482{
483	enum vm_suspend_how how;
484
485	how = vmexit->u.suspended.how;
486
487	fbsdrun_deletecpu(ctx, *pvcpu);
488
489	if (*pvcpu != BSP) {
490		pthread_mutex_lock(&resetcpu_mtx);
491		pthread_cond_signal(&resetcpu_cond);
492		pthread_mutex_unlock(&resetcpu_mtx);
493		pthread_exit(NULL);
494	}
495
496	pthread_mutex_lock(&resetcpu_mtx);
497	while (!CPU_EMPTY(&cpumask)) {
498		pthread_cond_wait(&resetcpu_cond, &resetcpu_mtx);
499	}
500	pthread_mutex_unlock(&resetcpu_mtx);
501
502	switch (how) {
503	case VM_SUSPEND_RESET:
504		exit(0);
505	case VM_SUSPEND_POWEROFF:
506		exit(1);
507	case VM_SUSPEND_HALT:
508		exit(2);
509	default:
510		fprintf(stderr, "vmexit_suspend: invalid reason %d\n", how);
511		exit(100);
512	}
513	return (0);	/* NOTREACHED */
514}
515
516static vmexit_handler_t handler[VM_EXITCODE_MAX] = {
517	[VM_EXITCODE_INOUT]  = vmexit_inout,
518	[VM_EXITCODE_INOUT_STR]  = vmexit_inout,
519	[VM_EXITCODE_VMX]    = vmexit_vmx,
520	[VM_EXITCODE_BOGUS]  = vmexit_bogus,
521	[VM_EXITCODE_RDMSR]  = vmexit_rdmsr,
522	[VM_EXITCODE_WRMSR]  = vmexit_wrmsr,
523	[VM_EXITCODE_MTRAP]  = vmexit_mtrap,
524	[VM_EXITCODE_INST_EMUL] = vmexit_inst_emul,
525	[VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap,
526	[VM_EXITCODE_SUSPENDED] = vmexit_suspend
527};
528
529static void
530vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip)
531{
532	int error, rc, prevcpu;
533	enum vm_exitcode exitcode;
534	enum vm_suspend_how how;
535
536	if (vcpumap[vcpu] != NULL) {
537		error = pthread_setaffinity_np(pthread_self(),
538		    sizeof(cpuset_t), vcpumap[vcpu]);
539		assert(error == 0);
540	}
541
542	while (1) {
543		error = vm_run(ctx, vcpu, rip, &vmexit[vcpu]);
544		if (error != 0)
545			break;
546
547		prevcpu = vcpu;
548
549		exitcode = vmexit[vcpu].exitcode;
550		if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) {
551			fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n",
552			    exitcode);
553			exit(1);
554		}
555
556                rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu);
557
558		switch (rc) {
559		case VMEXIT_CONTINUE:
560                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
561			break;
562		case VMEXIT_RESTART:
563                        rip = vmexit[vcpu].rip;
564			break;
565		case VMEXIT_RESET:
566		case VMEXIT_POWEROFF:
567			if (rc == VMEXIT_RESET)
568				how = VM_SUSPEND_RESET;
569			else
570				how = VM_SUSPEND_POWEROFF;
571			error = vm_suspend(ctx, how);
572			assert(error == 0 || errno == EALREADY);
573                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
574			break;
575		case VMEXIT_ABORT:
576			abort();
577		default:
578			exit(1);
579		}
580	}
581	fprintf(stderr, "vm_run error %d, errno %d\n", error, errno);
582}
583
584static int
585num_vcpus_allowed(struct vmctx *ctx)
586{
587	int tmp, error;
588
589	error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp);
590
591	/*
592	 * The guest is allowed to spinup more than one processor only if the
593	 * UNRESTRICTED_GUEST capability is available.
594	 */
595	if (error == 0)
596		return (VM_MAXCPU);
597	else
598		return (1);
599}
600
601void
602fbsdrun_set_capabilities(struct vmctx *ctx, int cpu)
603{
604	int err, tmp;
605
606	if (fbsdrun_vmexit_on_hlt()) {
607		err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp);
608		if (err < 0) {
609			fprintf(stderr, "VM exit on HLT not supported\n");
610			exit(1);
611		}
612		vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1);
613		if (cpu == BSP)
614			handler[VM_EXITCODE_HLT] = vmexit_hlt;
615	}
616
617        if (fbsdrun_vmexit_on_pause()) {
618		/*
619		 * pause exit support required for this mode
620		 */
621		err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp);
622		if (err < 0) {
623			fprintf(stderr,
624			    "SMP mux requested, no pause support\n");
625			exit(1);
626		}
627		vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1);
628		if (cpu == BSP)
629			handler[VM_EXITCODE_PAUSE] = vmexit_pause;
630        }
631
632	if (x2apic_mode)
633		err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED);
634	else
635		err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED);
636
637	if (err) {
638		fprintf(stderr, "Unable to set x2apic state (%d)\n", err);
639		exit(1);
640	}
641
642	vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1);
643}
644
645int
646main(int argc, char *argv[])
647{
648	int c, error, gdb_port, err, bvmcons;
649	int dump_guest_memory, max_vcpus, mptgen;
650	struct vmctx *ctx;
651	uint64_t rip;
652	size_t memsize;
653
654	bvmcons = 0;
655	dump_guest_memory = 0;
656	progname = basename(argv[0]);
657	gdb_port = 0;
658	guest_ncpus = 1;
659	memsize = 256 * MB;
660	mptgen = 1;
661
662	while ((c = getopt(argc, argv, "abehwxACHIPWYp:g:c:s:m:l:U:")) != -1) {
663		switch (c) {
664		case 'a':
665			x2apic_mode = 0;
666			break;
667		case 'A':
668			acpi = 1;
669			break;
670		case 'b':
671			bvmcons = 1;
672			break;
673		case 'p':
674                        if (pincpu_parse(optarg) != 0) {
675                            errx(EX_USAGE, "invalid vcpu pinning "
676                                 "configuration '%s'", optarg);
677                        }
678			break;
679                case 'c':
680			guest_ncpus = atoi(optarg);
681			break;
682		case 'C':
683			dump_guest_memory = 1;
684			break;
685		case 'g':
686			gdb_port = atoi(optarg);
687			break;
688		case 'l':
689			if (lpc_device_parse(optarg) != 0) {
690				errx(EX_USAGE, "invalid lpc device "
691				    "configuration '%s'", optarg);
692			}
693			break;
694		case 's':
695			if (pci_parse_slot(optarg) != 0)
696				exit(1);
697			else
698				break;
699                case 'm':
700			error = vm_parse_memsize(optarg, &memsize);
701			if (error)
702				errx(EX_USAGE, "invalid memsize '%s'", optarg);
703			break;
704		case 'H':
705			guest_vmexit_on_hlt = 1;
706			break;
707		case 'I':
708			/*
709			 * The "-I" option was used to add an ioapic to the
710			 * virtual machine.
711			 *
712			 * An ioapic is now provided unconditionally for each
713			 * virtual machine and this option is now deprecated.
714			 */
715			break;
716		case 'P':
717			guest_vmexit_on_pause = 1;
718			break;
719		case 'e':
720			strictio = 1;
721			break;
722		case 'U':
723			guest_uuid_str = optarg;
724			break;
725		case 'w':
726			strictmsr = 0;
727			break;
728		case 'W':
729			virtio_msix = 0;
730			break;
731		case 'x':
732			x2apic_mode = 1;
733			break;
734		case 'Y':
735			mptgen = 0;
736			break;
737		case 'h':
738			usage(0);
739		default:
740			usage(1);
741		}
742	}
743	argc -= optind;
744	argv += optind;
745
746	if (argc != 1)
747		usage(1);
748
749	vmname = argv[0];
750
751	ctx = vm_open(vmname);
752	if (ctx == NULL) {
753		perror("vm_open");
754		exit(1);
755	}
756
757	max_vcpus = num_vcpus_allowed(ctx);
758	if (guest_ncpus > max_vcpus) {
759		fprintf(stderr, "%d vCPUs requested but only %d available\n",
760			guest_ncpus, max_vcpus);
761		exit(1);
762	}
763
764	fbsdrun_set_capabilities(ctx, BSP);
765
766	if (dump_guest_memory)
767		vm_set_memflags(ctx, VM_MEM_F_INCORE);
768	err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL);
769	if (err) {
770		fprintf(stderr, "Unable to setup memory (%d)\n", err);
771		exit(1);
772	}
773
774	init_mem();
775	init_inout();
776	pci_irq_init(ctx);
777	ioapic_init(ctx);
778
779	rtc_init(ctx);
780	sci_init(ctx);
781
782	/*
783	 * Exit if a device emulation finds an error in it's initilization
784	 */
785	if (init_pci(ctx) != 0)
786		exit(1);
787
788	if (gdb_port != 0)
789		init_dbgport(gdb_port);
790
791	if (bvmcons)
792		init_bvmcons();
793
794	error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip);
795	assert(error == 0);
796
797	/*
798	 * build the guest tables, MP etc.
799	 */
800	if (mptgen) {
801		error = mptable_build(ctx, guest_ncpus);
802		if (error)
803			exit(1);
804	}
805
806	error = smbios_build(ctx);
807	assert(error == 0);
808
809	if (acpi) {
810		error = acpi_build(ctx, guest_ncpus);
811		assert(error == 0);
812	}
813
814	/*
815	 * Change the proc title to include the VM name.
816	 */
817	setproctitle("%s", vmname);
818
819	/*
820	 * Add CPU 0
821	 */
822	fbsdrun_addcpu(ctx, BSP, BSP, rip);
823
824	/*
825	 * Head off to the main event dispatch loop
826	 */
827	mevent_dispatch();
828
829	exit(1);
830}
831