bhyverun.c revision 268934 
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 268934 2014-07-21 02:17:28Z jhb $
27 */
28
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD: stable/10/usr.sbin/bhyve/bhyverun.c 268934 2014-07-21 02:17:28Z 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} 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		"       -p: pin 'vcpu' to 'hostcpu'\n"
138		"       -H: vmexit from the guest on hlt\n"
139		"       -P: vmexit from the guest on pause\n"
140		"       -W: force virtio to use single-vector MSI\n"
141		"       -e: exit on unhandled I/O access\n"
142		"       -h: help\n"
143		"       -s: <slot,driver,configinfo> PCI slot config\n"
144		"       -l: LPC device configuration\n"
145		"       -m: memory size in MB\n"
146		"       -w: ignore unimplemented MSRs\n"
147		"       -x: local apic is in x2APIC mode\n"
148		"       -Y: disable MPtable generation\n"
149		"       -U: uuid\n",
150		progname, (int)strlen(progname), "");
151
152	exit(code);
153}
154
155static int
156pincpu_parse(const char *opt)
157{
158	int vcpu, pcpu;
159
160	if (sscanf(opt, "%d:%d", &vcpu, &pcpu) != 2) {
161		fprintf(stderr, "invalid format: %s\n", opt);
162		return (-1);
163	}
164
165	if (vcpu < 0 || vcpu >= VM_MAXCPU) {
166		fprintf(stderr, "vcpu '%d' outside valid range from 0 to %d\n",
167		    vcpu, VM_MAXCPU - 1);
168		return (-1);
169	}
170
171	if (pcpu < 0 || pcpu >= CPU_SETSIZE) {
172		fprintf(stderr, "hostcpu '%d' outside valid range from "
173		    "0 to %d\n", pcpu, CPU_SETSIZE - 1);
174		return (-1);
175	}
176
177	if (vcpumap[vcpu] == NULL) {
178		if ((vcpumap[vcpu] = malloc(sizeof(cpuset_t))) == NULL) {
179			perror("malloc");
180			return (-1);
181		}
182		CPU_ZERO(vcpumap[vcpu]);
183	}
184	CPU_SET(pcpu, vcpumap[vcpu]);
185	return (0);
186}
187
188void *
189paddr_guest2host(struct vmctx *ctx, uintptr_t gaddr, size_t len)
190{
191
192	return (vm_map_gpa(ctx, gaddr, len));
193}
194
195int
196fbsdrun_vmexit_on_pause(void)
197{
198
199	return (guest_vmexit_on_pause);
200}
201
202int
203fbsdrun_vmexit_on_hlt(void)
204{
205
206	return (guest_vmexit_on_hlt);
207}
208
209int
210fbsdrun_virtio_msix(void)
211{
212
213	return (virtio_msix);
214}
215
216static void *
217fbsdrun_start_thread(void *param)
218{
219	char tname[MAXCOMLEN + 1];
220	struct mt_vmm_info *mtp;
221	int vcpu;
222
223	mtp = param;
224	vcpu = mtp->mt_vcpu;
225
226	snprintf(tname, sizeof(tname), "vcpu %d", vcpu);
227	pthread_set_name_np(mtp->mt_thr, tname);
228
229	vm_loop(mtp->mt_ctx, vcpu, vmexit[vcpu].rip);
230
231	/* not reached */
232	exit(1);
233	return (NULL);
234}
235
236void
237fbsdrun_addcpu(struct vmctx *ctx, int fromcpu, int newcpu, uint64_t rip)
238{
239	int error;
240
241	assert(fromcpu == BSP);
242
243	CPU_SET_ATOMIC(newcpu, &cpumask);
244
245	/*
246	 * Set up the vmexit struct to allow execution to start
247	 * at the given RIP
248	 */
249	vmexit[newcpu].rip = rip;
250	vmexit[newcpu].inst_length = 0;
251
252	mt_vmm_info[newcpu].mt_ctx = ctx;
253	mt_vmm_info[newcpu].mt_vcpu = newcpu;
254
255	error = pthread_create(&mt_vmm_info[newcpu].mt_thr, NULL,
256	    fbsdrun_start_thread, &mt_vmm_info[newcpu]);
257	assert(error == 0);
258}
259
260static int
261fbsdrun_deletecpu(struct vmctx *ctx, int vcpu)
262{
263
264	if (!CPU_ISSET(vcpu, &cpumask)) {
265		fprintf(stderr, "Attempting to delete unknown cpu %d\n", vcpu);
266		exit(1);
267	}
268
269	CPU_CLR_ATOMIC(vcpu, &cpumask);
270	return (CPU_EMPTY(&cpumask));
271}
272
273static int
274vmexit_catch_reset(void)
275{
276        stats.io_reset++;
277        return (VMEXIT_RESET);
278}
279
280static int
281vmexit_catch_inout(void)
282{
283	return (VMEXIT_ABORT);
284}
285
286static int
287vmexit_handle_notify(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu,
288		     uint32_t eax)
289{
290#if BHYVE_DEBUG
291	/*
292	 * put guest-driven debug here
293	 */
294#endif
295        return (VMEXIT_CONTINUE);
296}
297
298static int
299vmexit_inout(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
300{
301	int error;
302	int bytes, port, in, out;
303	uint32_t eax;
304	int vcpu;
305
306	vcpu = *pvcpu;
307
308	port = vme->u.inout.port;
309	bytes = vme->u.inout.bytes;
310	eax = vme->u.inout.eax;
311	in = vme->u.inout.in;
312	out = !in;
313
314	/* We don't deal with these */
315	if (vme->u.inout.string || vme->u.inout.rep)
316		return (VMEXIT_ABORT);
317
318        /* Extra-special case of host notifications */
319        if (out && port == GUEST_NIO_PORT)
320                return (vmexit_handle_notify(ctx, vme, pvcpu, eax));
321
322	error = emulate_inout(ctx, vcpu, in, port, bytes, &eax, strictio);
323	if (error == INOUT_OK && in)
324		error = vm_set_register(ctx, vcpu, VM_REG_GUEST_RAX, eax);
325
326	switch (error) {
327	case INOUT_OK:
328		return (VMEXIT_CONTINUE);
329	case INOUT_RESET:
330		return (VMEXIT_RESET);
331	case INOUT_POWEROFF:
332		return (VMEXIT_POWEROFF);
333	default:
334		fprintf(stderr, "Unhandled %s%c 0x%04x\n",
335			in ? "in" : "out",
336			bytes == 1 ? 'b' : (bytes == 2 ? 'w' : 'l'), port);
337		return (vmexit_catch_inout());
338	}
339}
340
341static int
342vmexit_rdmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
343{
344	uint64_t val;
345	uint32_t eax, edx;
346	int error;
347
348	val = 0;
349	error = emulate_rdmsr(ctx, *pvcpu, vme->u.msr.code, &val);
350	if (error != 0) {
351		fprintf(stderr, "rdmsr to register %#x on vcpu %d\n",
352		    vme->u.msr.code, *pvcpu);
353		if (strictmsr) {
354			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
355			assert(error == 0);
356			return (VMEXIT_RESTART);
357		}
358	}
359
360	eax = val;
361	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RAX, eax);
362	assert(error == 0);
363
364	edx = val >> 32;
365	error = vm_set_register(ctx, *pvcpu, VM_REG_GUEST_RDX, edx);
366	assert(error == 0);
367
368	return (VMEXIT_CONTINUE);
369}
370
371static int
372vmexit_wrmsr(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
373{
374	int error;
375
376	error = emulate_wrmsr(ctx, *pvcpu, vme->u.msr.code, vme->u.msr.wval);
377	if (error != 0) {
378		fprintf(stderr, "wrmsr to register %#x(%#lx) on vcpu %d\n",
379		    vme->u.msr.code, vme->u.msr.wval, *pvcpu);
380		if (strictmsr) {
381			error = vm_inject_exception2(ctx, *pvcpu, IDT_GP, 0);
382			assert(error == 0);
383			return (VMEXIT_RESTART);
384		}
385	}
386	return (VMEXIT_CONTINUE);
387}
388
389static int
390vmexit_spinup_ap(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
391{
392	int newcpu;
393	int retval = VMEXIT_CONTINUE;
394
395	newcpu = spinup_ap(ctx, *pvcpu,
396			   vme->u.spinup_ap.vcpu, vme->u.spinup_ap.rip);
397
398	return (retval);
399}
400
401static int
402vmexit_spindown_cpu(struct vmctx *ctx, struct vm_exit *vme, int *pvcpu)
403{
404	int lastcpu;
405
406	lastcpu = fbsdrun_deletecpu(ctx, *pvcpu);
407	if (!lastcpu)
408		pthread_exit(NULL);
409	return (vmexit_catch_reset());
410}
411
412static int
413vmexit_vmx(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
414{
415
416	fprintf(stderr, "vm exit[%d]\n", *pvcpu);
417	fprintf(stderr, "\treason\t\tVMX\n");
418	fprintf(stderr, "\trip\t\t0x%016lx\n", vmexit->rip);
419	fprintf(stderr, "\tinst_length\t%d\n", vmexit->inst_length);
420	fprintf(stderr, "\tstatus\t\t%d\n", vmexit->u.vmx.status);
421	fprintf(stderr, "\texit_reason\t%u\n", vmexit->u.vmx.exit_reason);
422	fprintf(stderr, "\tqualification\t0x%016lx\n",
423	    vmexit->u.vmx.exit_qualification);
424	fprintf(stderr, "\tinst_type\t\t%d\n", vmexit->u.vmx.inst_type);
425	fprintf(stderr, "\tinst_error\t\t%d\n", vmexit->u.vmx.inst_error);
426
427	return (VMEXIT_ABORT);
428}
429
430static int
431vmexit_bogus(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
432{
433
434	stats.vmexit_bogus++;
435
436	return (VMEXIT_RESTART);
437}
438
439static int
440vmexit_hlt(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
441{
442
443	stats.vmexit_hlt++;
444
445	/*
446	 * Just continue execution with the next instruction. We use
447	 * the HLT VM exit as a way to be friendly with the host
448	 * scheduler.
449	 */
450	return (VMEXIT_CONTINUE);
451}
452
453static int
454vmexit_pause(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
455{
456
457	stats.vmexit_pause++;
458
459	return (VMEXIT_CONTINUE);
460}
461
462static int
463vmexit_mtrap(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
464{
465
466	stats.vmexit_mtrap++;
467
468	return (VMEXIT_RESTART);
469}
470
471static int
472vmexit_inst_emul(struct vmctx *ctx, struct vm_exit *vmexit, int *pvcpu)
473{
474	int err;
475	stats.vmexit_inst_emul++;
476
477	err = emulate_mem(ctx, *pvcpu, vmexit->u.inst_emul.gpa,
478			  &vmexit->u.inst_emul.vie);
479
480	if (err) {
481		if (err == EINVAL) {
482			fprintf(stderr,
483			    "Failed to emulate instruction at 0x%lx\n",
484			    vmexit->rip);
485		} else if (err == ESRCH) {
486			fprintf(stderr, "Unhandled memory access to 0x%lx\n",
487			    vmexit->u.inst_emul.gpa);
488		}
489
490		return (VMEXIT_ABORT);
491	}
492
493	return (VMEXIT_CONTINUE);
494}
495
496static vmexit_handler_t handler[VM_EXITCODE_MAX] = {
497	[VM_EXITCODE_INOUT]  = vmexit_inout,
498	[VM_EXITCODE_VMX]    = vmexit_vmx,
499	[VM_EXITCODE_BOGUS]  = vmexit_bogus,
500	[VM_EXITCODE_RDMSR]  = vmexit_rdmsr,
501	[VM_EXITCODE_WRMSR]  = vmexit_wrmsr,
502	[VM_EXITCODE_MTRAP]  = vmexit_mtrap,
503	[VM_EXITCODE_INST_EMUL] = vmexit_inst_emul,
504	[VM_EXITCODE_SPINUP_AP] = vmexit_spinup_ap,
505	[VM_EXITCODE_SPINDOWN_CPU] = vmexit_spindown_cpu,
506};
507
508static void
509vm_loop(struct vmctx *ctx, int vcpu, uint64_t rip)
510{
511	int error, rc, prevcpu;
512	enum vm_exitcode exitcode;
513
514	if (vcpumap[vcpu] != NULL) {
515		error = pthread_setaffinity_np(pthread_self(),
516		    sizeof(cpuset_t), vcpumap[vcpu]);
517		assert(error == 0);
518	}
519
520	while (1) {
521		error = vm_run(ctx, vcpu, rip, &vmexit[vcpu]);
522		if (error != 0)
523			break;
524
525		prevcpu = vcpu;
526
527		exitcode = vmexit[vcpu].exitcode;
528		if (exitcode >= VM_EXITCODE_MAX || handler[exitcode] == NULL) {
529			fprintf(stderr, "vm_loop: unexpected exitcode 0x%x\n",
530			    exitcode);
531			exit(1);
532		}
533
534                rc = (*handler[exitcode])(ctx, &vmexit[vcpu], &vcpu);
535
536		switch (rc) {
537		case VMEXIT_CONTINUE:
538                        rip = vmexit[vcpu].rip + vmexit[vcpu].inst_length;
539			break;
540		case VMEXIT_RESTART:
541                        rip = vmexit[vcpu].rip;
542			break;
543		case VMEXIT_RESET:
544			exit(0);
545		default:
546			exit(1);
547		}
548	}
549	fprintf(stderr, "vm_run error %d, errno %d\n", error, errno);
550}
551
552static int
553num_vcpus_allowed(struct vmctx *ctx)
554{
555	int tmp, error;
556
557	error = vm_get_capability(ctx, BSP, VM_CAP_UNRESTRICTED_GUEST, &tmp);
558
559	/*
560	 * The guest is allowed to spinup more than one processor only if the
561	 * UNRESTRICTED_GUEST capability is available.
562	 */
563	if (error == 0)
564		return (VM_MAXCPU);
565	else
566		return (1);
567}
568
569void
570fbsdrun_set_capabilities(struct vmctx *ctx, int cpu)
571{
572	int err, tmp;
573
574	if (fbsdrun_vmexit_on_hlt()) {
575		err = vm_get_capability(ctx, cpu, VM_CAP_HALT_EXIT, &tmp);
576		if (err < 0) {
577			fprintf(stderr, "VM exit on HLT not supported\n");
578			exit(1);
579		}
580		vm_set_capability(ctx, cpu, VM_CAP_HALT_EXIT, 1);
581		if (cpu == BSP)
582			handler[VM_EXITCODE_HLT] = vmexit_hlt;
583	}
584
585        if (fbsdrun_vmexit_on_pause()) {
586		/*
587		 * pause exit support required for this mode
588		 */
589		err = vm_get_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, &tmp);
590		if (err < 0) {
591			fprintf(stderr,
592			    "SMP mux requested, no pause support\n");
593			exit(1);
594		}
595		vm_set_capability(ctx, cpu, VM_CAP_PAUSE_EXIT, 1);
596		if (cpu == BSP)
597			handler[VM_EXITCODE_PAUSE] = vmexit_pause;
598        }
599
600	if (x2apic_mode)
601		err = vm_set_x2apic_state(ctx, cpu, X2APIC_ENABLED);
602	else
603		err = vm_set_x2apic_state(ctx, cpu, X2APIC_DISABLED);
604
605	if (err) {
606		fprintf(stderr, "Unable to set x2apic state (%d)\n", err);
607		exit(1);
608	}
609
610	vm_set_capability(ctx, cpu, VM_CAP_ENABLE_INVPCID, 1);
611}
612
613int
614main(int argc, char *argv[])
615{
616	int c, error, gdb_port, err, bvmcons;
617	int max_vcpus, mptgen;
618	struct vmctx *ctx;
619	uint64_t rip;
620	size_t memsize;
621
622	bvmcons = 0;
623	progname = basename(argv[0]);
624	gdb_port = 0;
625	guest_ncpus = 1;
626	memsize = 256 * MB;
627	mptgen = 1;
628
629	while ((c = getopt(argc, argv, "abehwxAHIPWYp:g:c:s:m:l:U:")) != -1) {
630		switch (c) {
631		case 'a':
632			x2apic_mode = 0;
633			break;
634		case 'A':
635			acpi = 1;
636			break;
637		case 'b':
638			bvmcons = 1;
639			break;
640		case 'p':
641                        if (pincpu_parse(optarg) != 0) {
642                            errx(EX_USAGE, "invalid vcpu pinning "
643                                 "configuration '%s'", optarg);
644                        }
645			break;
646                case 'c':
647			guest_ncpus = atoi(optarg);
648			break;
649		case 'g':
650			gdb_port = atoi(optarg);
651			break;
652		case 'l':
653			if (lpc_device_parse(optarg) != 0) {
654				errx(EX_USAGE, "invalid lpc device "
655				    "configuration '%s'", optarg);
656			}
657			break;
658		case 's':
659			if (pci_parse_slot(optarg) != 0)
660				exit(1);
661			else
662				break;
663                case 'm':
664			error = vm_parse_memsize(optarg, &memsize);
665			if (error)
666				errx(EX_USAGE, "invalid memsize '%s'", optarg);
667			break;
668		case 'H':
669			guest_vmexit_on_hlt = 1;
670			break;
671		case 'I':
672			/*
673			 * The "-I" option was used to add an ioapic to the
674			 * virtual machine.
675			 *
676			 * An ioapic is now provided unconditionally for each
677			 * virtual machine and this option is now deprecated.
678			 */
679			break;
680		case 'P':
681			guest_vmexit_on_pause = 1;
682			break;
683		case 'e':
684			strictio = 1;
685			break;
686		case 'U':
687			guest_uuid_str = optarg;
688			break;
689		case 'w':
690			strictmsr = 0;
691			break;
692		case 'W':
693			virtio_msix = 0;
694			break;
695		case 'x':
696			x2apic_mode = 1;
697			break;
698		case 'Y':
699			mptgen = 0;
700			break;
701		case 'h':
702			usage(0);
703		default:
704			usage(1);
705		}
706	}
707	argc -= optind;
708	argv += optind;
709
710	if (argc != 1)
711		usage(1);
712
713	vmname = argv[0];
714
715	ctx = vm_open(vmname);
716	if (ctx == NULL) {
717		perror("vm_open");
718		exit(1);
719	}
720
721	max_vcpus = num_vcpus_allowed(ctx);
722	if (guest_ncpus > max_vcpus) {
723		fprintf(stderr, "%d vCPUs requested but only %d available\n",
724			guest_ncpus, max_vcpus);
725		exit(1);
726	}
727
728	fbsdrun_set_capabilities(ctx, BSP);
729
730	err = vm_setup_memory(ctx, memsize, VM_MMAP_ALL);
731	if (err) {
732		fprintf(stderr, "Unable to setup memory (%d)\n", err);
733		exit(1);
734	}
735
736	init_mem();
737	init_inout();
738	ioapic_init(ctx);
739
740	rtc_init(ctx);
741
742	/*
743	 * Exit if a device emulation finds an error in it's initilization
744	 */
745	if (init_pci(ctx) != 0)
746		exit(1);
747
748	if (gdb_port != 0)
749		init_dbgport(gdb_port);
750
751	if (bvmcons)
752		init_bvmcons();
753
754	error = vm_get_register(ctx, BSP, VM_REG_GUEST_RIP, &rip);
755	assert(error == 0);
756
757	/*
758	 * build the guest tables, MP etc.
759	 */
760	if (mptgen) {
761		error = mptable_build(ctx, guest_ncpus);
762		if (error)
763			exit(1);
764	}
765
766	error = smbios_build(ctx);
767	assert(error == 0);
768
769	if (acpi) {
770		error = acpi_build(ctx, guest_ncpus);
771		assert(error == 0);
772	}
773
774	/*
775	 * Change the proc title to include the VM name.
776	 */
777	setproctitle("%s", vmname);
778
779	/*
780	 * Add CPU 0
781	 */
782	fbsdrun_addcpu(ctx, BSP, BSP, rip);
783
784	/*
785	 * Head off to the main event dispatch loop
786	 */
787	mevent_dispatch();
788
789	exit(1);
790}
791