bhyverun.c revision 268935 
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 268935 2014-07-21 02:39:17Z jhb $
27 */
28
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD: stable/10/usr.sbin/bhyve/bhyverun.c 268935 2014-07-21 02:39:17Z 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_SWITCH		0	/* force vcpu switch in mux mode */
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		"       -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_catch_inout(void)
276{
277	return (VMEXIT_ABORT);
278}
279
280static int
281vmexit_handle_notify(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu,
282		     uint32_t eax)
283{
284#if BHYVE_DEBUG
285	/*
286	 * put guest-driven debug here
287	 */
288#endif
289        return (VMEXIT_CONTINUE);
290}
291
292static int
293vmexit_inout(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
294{
295	int error;
296	int bytes, port, in, out;
297	uint32_t eax;
298	int vcpu;
299
300	vcpu = *pvcpu;
301
302	port = vme->u.inout.port;
303	bytes = vme->u.inout.bytes;
304	eax = vme->u.inout.eax;
305	in = vme->u.inout.in;
306	out = !in;
307
308	/* We don't deal with these */
309	if (vme->u.inout.string || vme->u.inout.rep)
310		return (VMEXIT_ABORT);
311
312        /* Extra-special case of host notifications */
313        if (out && port == GUEST_NIO_PORT)
314                return (vmexit_handle_notify(ctx, vme, pvcpu, eax));
315
316	error = emulate_inout(ctx, vcpu, in, port, bytes, &eax, strictio);
317	if (error == INOUT_OK && in)
318		error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RAX, eax);
319
320	switch (error) {
321	case INOUT_OK:
322		return (VMEXIT_CONTINUE);
323	case INOUT_RESET:
324		stats.io_reset++;
325		return (VMEXIT_RESET);
326	case INOUT_POWEROFF:
327		stats.io_poweroff++;
328		return (VMEXIT_POWEROFF);
329	default:
330		fprintf(stderr, "Unhandled %s%c 0x%04x\n",
331			in ? "in" : "out",
332			bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), port);
333		return (vmexit_catch_inout());
334	}
335}
336
337static int
338vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
339{
340	uint64_t val;
341	uint32_t eax, edx;
342	int error;
343
344	val = 0;
345	error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val);
346	if (error != 0) {
347		fprintf(stderr, "rdmsr to register %#x on vcpu %d\n",
348		    vme->u.msr.code, *pvcpu);
349		if (strictmsr) {
350			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
351			assert(error == 0);
352			return (VMEXIT_RESTART);
353		}
354	}
355
356	eax = val;
357	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax);
358	assert(error == 0);
359
360	edx = val >> 32;
361	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx);
362	assert(error == 0);
363
364	return (VMEXIT_CONTINUE);
365}
366
367static int
368vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
369{
370	int error;
371
372	error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval);
373	if (error != 0) {
374		fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n",
375		    vme->u.msr.code, vme->u.msr.wval, *pvcpu);
376		if (strictmsr) {
377			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
378			assert(error == 0);
379			return (VMEXIT_RESTART);
380		}
381	}
382	return (VMEXIT_CONTINUE);
383}
384
385static int
386vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
387{
388	int newcpu;
389	int retval = VMEXIT_CONTINUE;
390
391	newcpu = spinup_ap(ctx, *pvcpu,
392			   vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip);
393
394	return (retval);
395}
396
397static int
398vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
399{
400
401	fprintf(stderr, "vm exit[%d]\n", *pvcpu);
402	fprintf(stderr, "\treason\t\tVMX\n");
403	fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip);
404	fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length);
405	fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status);
406	fprintf(stderr, "\texit_reason\t%u\n", vmexit->u.vmx.exit_reason);
407	fprintf(stderr, "\tqualification\t0x%016lx\n",
408	    vmexit->u.vmx.exit_qualification);
409	fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type);
410	fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error);
411
412	return (VMEXIT_ABORT);
413}
414
415static int
416vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
417{
418
419	stats.vmexit_bogus++;
420
421	return (VMEXIT_RESTART);
422}
423
424static int
425vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
426{
427
428	stats.vmexit_hlt++;
429
430	/*
431	 * Just continue execution with the next instruction. We use
432	 * the HLT VM exit as a way to be friendly with the host
433	 * scheduler.
434	 */
435	return (VMEXIT_CONTINUE);
436}
437
438static int
439vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
440{
441
442	stats.vmexit_pause++;
443
444	return (VMEXIT_CONTINUE);
445}
446
447static int
448vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
449{
450
451	stats.vmexit_mtrap++;
452
453	return (VMEXIT_RESTART);
454}
455
456static int
457vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
458{
459	int err;
460	stats.vmexit_inst_emul++;
461
462	err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa,
463			  &vmexit->u.inst_emul.vie);
464
465	if (err) {
466		if (err == EINVAL) {
467			fprintf(stderr,
468			    "Failed to emulate instruction at 0x%lx\n",
469			    vmexit->rip);
470		} else if (err == ESRCH) {
471			fprintf(stderr, "Unhandled memory access to 0x%lx\n",
472			    vmexit->u.inst_emul.gpa);
473		}
474
475		return (VMEXIT_ABORT);
476	}
477
478	return (VMEXIT_CONTINUE);
479}
480
481static pthread_mutex_t resetcpu_mtx = PTHREAD_MUTEX_INITIALIZER;
482static pthread_cond_t resetcpu_cond = PTHREAD_COND_INITIALIZER;
483
484static int
485vmexit_suspend(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
486{
487	enum vm_suspend_how how;
488
489	how = vmexit->u.suspended.how;
490
491	fbsdrun_deletecpu(ctx, *pvcpu);
492
493	if (*pvcpu != BSP) {
494		pthread_mutex_lock(&resetcpu_mtx);
495		pthread_cond_signal(&resetcpu_cond);
496		pthread_mutex_unlock(&resetcpu_mtx);
497		pthread_exit(NULL);
498	}
499
500	pthread_mutex_lock(&resetcpu_mtx);
501	while (!CPU_EMPTY(&cpumask)) {
502		pthread_cond_wait(&resetcpu_cond, &resetcpu_mtx);
503	}
504	pthread_mutex_unlock(&resetcpu_mtx);
505
506	switch (how) {
507	case VM_SUSPEND_RESET:
508		exit(0);
509	case VM_SUSPEND_POWEROFF:
510		exit(1);
511	case VM_SUSPEND_HALT:
512		exit(2);
513	default:
514		fprintf(stderr, "vmexit_suspend: invalid reason %d\n", how);
515		exit(100);
516	}
517	return (0);	/* NOTREACHED */
518}
519
520static vmexit_handler_t handler[VM_EXITCODE_MAX] = {
521	[VM_EXITCODE_INOUT]  = vmexit_inout,
522	[VM_EXITCODE_VMX]    = vmexit_vmx,
523	[VM_EXITCODE_BOGUS]  = vmexit_bogus,
524	[VM_EXITCODE_RDMSR]  = vmexit_rdmsr,
525	[VM_EXITCODE_WRMSR]  = vmexit_wrmsr,
526	[VM_EXITCODE_MTRAP]  = vmexit_mtrap,
527	[VM_EXITCODE_INST_EMUL] = vmexit_inst_emul,
528	[VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap,
529	[VM_EXITCODE_SUSPENDED] = vmexit_suspend
530};
531
532static void
533vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip)
534{
535	int error, rc, prevcpu;
536	enum vm_exitcode exitcode;
537	enum vm_suspend_how how;
538
539	if (vcpumap[vcpu] != NULL) {
540		error = pthread_setaffinity_np(pthread_self(),
541		    sizeof(cpuset_t), vcpumap[vcpu]);
542		assert(error == 0);
543	}
544
545	while (1) {
546		error = vm_run(ctx, vcpu, rip, &vmexit[vcpu]);
547		if (error != 0)
548			break;
549
550		prevcpu = vcpu;
551
552		exitcode = vmexit[vcpu].exitcode;
553		if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) {
554			fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n",
555			    exitcode);
556			exit(1);
557		}
558
559                rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu);
560
561		switch (rc) {
562		case VMEXIT_CONTINUE:
563                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
564			break;
565		case VMEXIT_RESTART:
566                        rip = vmexit[vcpu].rip;
567			break;
568		case VMEXIT_RESET:
569		case VMEXIT_POWEROFF:
570			if (rc == VMEXIT_RESET)
571				how = VM_SUSPEND_RESET;
572			else
573				how = VM_SUSPEND_POWEROFF;
574			error = vm_suspend(ctx, how);
575			assert(error == 0 || errno == EALREADY);
576                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
577			break;
578		default:
579			exit(1);
580		}
581	}
582	fprintf(stderr, "vm_run error %d, errno %d\n", error, errno);
583}
584
585static int
586num_vcpus_allowed(struct vmctx *ctx)
587{
588	int tmp, error;
589
590	error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp);
591
592	/*
593	 * The guest is allowed to spinup more than one processor only if the
594	 * UNRESTRICTED_GUEST capability is available.
595	 */
596	if (error == 0)
597		return (VM_MAXCPU);
598	else
599		return (1);
600}
601
602void
603fbsdrun_set_capabilities(struct vmctx *ctx, int cpu)
604{
605	int err, tmp;
606
607	if (fbsdrun_vmexit_on_hlt()) {
608		err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp);
609		if (err < 0) {
610			fprintf(stderr, "VM exit on HLT not supported\n");
611			exit(1);
612		}
613		vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1);
614		if (cpu == BSP)
615			handler[VM_EXITCODE_HLT] = vmexit_hlt;
616	}
617
618        if (fbsdrun_vmexit_on_pause()) {
619		/*
620		 * pause exit support required for this mode
621		 */
622		err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp);
623		if (err < 0) {
624			fprintf(stderr,
625			    "SMP mux requested, no pause support\n");
626			exit(1);
627		}
628		vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1);
629		if (cpu == BSP)
630			handler[VM_EXITCODE_PAUSE] = vmexit_pause;
631        }
632
633	if (x2apic_mode)
634		err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED);
635	else
636		err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED);
637
638	if (err) {
639		fprintf(stderr, "Unable to set x2apic state (%d)\n", err);
640		exit(1);
641	}
642
643	vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1);
644}
645
646int
647main(int argc, char *argv[])
648{
649	int c, error, gdb_port, err, bvmcons;
650	int max_vcpus, mptgen;
651	struct vmctx *ctx;
652	uint64_t rip;
653	size_t memsize;
654
655	bvmcons = 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, "abehwxAHIPWYp: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 '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	err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL);
764	if (err) {
765		fprintf(stderr, "Unable to setup memory (%d)\n", err);
766		exit(1);
767	}
768
769	init_mem();
770	init_inout();
771	ioapic_init(ctx);
772
773	rtc_init(ctx);
774
775	/*
776	 * Exit if a device emulation finds an error in it's initilization
777	 */
778	if (init_pci(ctx) != 0)
779		exit(1);
780
781	if (gdb_port != 0)
782		init_dbgport(gdb_port);
783
784	if (bvmcons)
785		init_bvmcons();
786
787	error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip);
788	assert(error == 0);
789
790	/*
791	 * build the guest tables, MP etc.
792	 */
793	if (mptgen) {
794		error = mptable_build(ctx, guest_ncpus);
795		if (error)
796			exit(1);
797	}
798
799	error = smbios_build(ctx);
800	assert(error == 0);
801
802	if (acpi) {
803		error = acpi_build(ctx, guest_ncpus);
804		assert(error == 0);
805	}
806
807	/*
808	 * Change the proc title to include the VM name.
809	 */
810	setproctitle("%s", vmname);
811
812	/*
813	 * Add CPU 0
814	 */
815	fbsdrun_addcpu(ctx, BSP, BSP, rip);
816
817	/*
818	 * Head off to the main event dispatch loop
819	 */
820	mevent_dispatch();
821
822	exit(1);
823}
824