vmmapi.c revision 267393
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/lib/libvmmapi/vmmapi.c 267393 2014-06-12 13:13:15Z jhb $
27 */
28
29#include <sys/cdefs.h>
30__FBSDID("$FreeBSD: stable/10/lib/libvmmapi/vmmapi.c 267393 2014-06-12 13:13:15Z jhb $");
31
32#include <sys/types.h>
33#include <sys/sysctl.h>
34#include <sys/ioctl.h>
35#include <sys/mman.h>
36
37#include <machine/specialreg.h>
38
39#include <stdio.h>
40#include <stdlib.h>
41#include <assert.h>
42#include <string.h>
43#include <fcntl.h>
44#include <unistd.h>
45
46#include <libutil.h>
47
48#include <machine/vmm.h>
49#include <machine/vmm_dev.h>
50
51#include "vmmapi.h"
52
53#define	MB	(1024 * 1024UL)
54#define	GB	(1024 * 1024 * 1024UL)
55
56struct vmctx {
57	int	fd;
58	uint32_t lowmem_limit;
59	enum vm_mmap_style vms;
60	size_t	lowmem;
61	char	*lowmem_addr;
62	size_t	highmem;
63	char	*highmem_addr;
64	char	*name;
65};
66
67#define	CREATE(x)  sysctlbyname("hw.vmm.create", NULL, NULL, (x), strlen((x)))
68#define	DESTROY(x) sysctlbyname("hw.vmm.destroy", NULL, NULL, (x), strlen((x)))
69
70static int
71vm_device_open(const char *name)
72{
73        int fd, len;
74        char *vmfile;
75
76	len = strlen("/dev/vmm/") + strlen(name) + 1;
77	vmfile = malloc(len);
78	assert(vmfile != NULL);
79	snprintf(vmfile, len, "/dev/vmm/%s", name);
80
81        /* Open the device file */
82        fd = open(vmfile, O_RDWR, 0);
83
84	free(vmfile);
85        return (fd);
86}
87
88int
89vm_create(const char *name)
90{
91
92	return (CREATE((char *)name));
93}
94
95struct vmctx *
96vm_open(const char *name)
97{
98	struct vmctx *vm;
99
100	vm = malloc(sizeof(struct vmctx) + strlen(name) + 1);
101	assert(vm != NULL);
102
103	vm->fd = -1;
104	vm->lowmem_limit = 3 * GB;
105	vm->name = (char *)(vm + 1);
106	strcpy(vm->name, name);
107
108	if ((vm->fd = vm_device_open(vm->name)) < 0)
109		goto err;
110
111	return (vm);
112err:
113	vm_destroy(vm);
114	return (NULL);
115}
116
117void
118vm_destroy(struct vmctx *vm)
119{
120	assert(vm != NULL);
121
122	if (vm->fd >= 0)
123		close(vm->fd);
124	DESTROY(vm->name);
125
126	free(vm);
127}
128
129int
130vm_parse_memsize(const char *optarg, size_t *ret_memsize)
131{
132	char *endptr;
133	size_t optval;
134	int error;
135
136	optval = strtoul(optarg, &endptr, 0);
137	if (*optarg != '\0' && *endptr == '\0') {
138		/*
139		 * For the sake of backward compatibility if the memory size
140		 * specified on the command line is less than a megabyte then
141		 * it is interpreted as being in units of MB.
142		 */
143		if (optval < MB)
144			optval *= MB;
145		*ret_memsize = optval;
146		error = 0;
147	} else
148		error = expand_number(optarg, ret_memsize);
149
150	return (error);
151}
152
153int
154vm_get_memory_seg(struct vmctx *ctx, vm_paddr_t gpa, size_t *ret_len,
155		  int *wired)
156{
157	int error;
158	struct vm_memory_segment seg;
159
160	bzero(&seg, sizeof(seg));
161	seg.gpa = gpa;
162	error = ioctl(ctx->fd, VM_GET_MEMORY_SEG, &seg);
163	*ret_len = seg.len;
164	if (wired != NULL)
165		*wired = seg.wired;
166	return (error);
167}
168
169uint32_t
170vm_get_lowmem_limit(struct vmctx *ctx)
171{
172
173	return (ctx->lowmem_limit);
174}
175
176void
177vm_set_lowmem_limit(struct vmctx *ctx, uint32_t limit)
178{
179
180	ctx->lowmem_limit = limit;
181}
182
183static int
184setup_memory_segment(struct vmctx *ctx, vm_paddr_t gpa, size_t len, char **addr)
185{
186	int error;
187	struct vm_memory_segment seg;
188
189	/*
190	 * Create and optionally map 'len' bytes of memory at guest
191	 * physical address 'gpa'
192	 */
193	bzero(&seg, sizeof(seg));
194	seg.gpa = gpa;
195	seg.len = len;
196	error = ioctl(ctx->fd, VM_MAP_MEMORY, &seg);
197	if (error == 0 && addr != NULL) {
198		*addr = mmap(NULL, len, PROT_READ | PROT_WRITE, MAP_SHARED,
199				ctx->fd, gpa);
200	}
201	return (error);
202}
203
204int
205vm_setup_memory(struct vmctx *ctx, size_t memsize, enum vm_mmap_style vms)
206{
207	char **addr;
208	int error;
209
210	/* XXX VM_MMAP_SPARSE not implemented yet */
211	assert(vms == VM_MMAP_NONE || vms == VM_MMAP_ALL);
212	ctx->vms = vms;
213
214	/*
215	 * If 'memsize' cannot fit entirely in the 'lowmem' segment then
216	 * create another 'highmem' segment above 4GB for the remainder.
217	 */
218	if (memsize > ctx->lowmem_limit) {
219		ctx->lowmem = ctx->lowmem_limit;
220		ctx->highmem = memsize - ctx->lowmem;
221	} else {
222		ctx->lowmem = memsize;
223		ctx->highmem = 0;
224	}
225
226	if (ctx->lowmem > 0) {
227		addr = (vms == VM_MMAP_ALL) ? &ctx->lowmem_addr : NULL;
228		error = setup_memory_segment(ctx, 0, ctx->lowmem, addr);
229		if (error)
230			return (error);
231	}
232
233	if (ctx->highmem > 0) {
234		addr = (vms == VM_MMAP_ALL) ? &ctx->highmem_addr : NULL;
235		error = setup_memory_segment(ctx, 4*GB, ctx->highmem, addr);
236		if (error)
237			return (error);
238	}
239
240	return (0);
241}
242
243void *
244vm_map_gpa(struct vmctx *ctx, vm_paddr_t gaddr, size_t len)
245{
246
247	/* XXX VM_MMAP_SPARSE not implemented yet */
248	assert(ctx->vms == VM_MMAP_ALL);
249
250	if (gaddr < ctx->lowmem && gaddr + len <= ctx->lowmem)
251		return ((void *)(ctx->lowmem_addr + gaddr));
252
253	if (gaddr >= 4*GB) {
254		gaddr -= 4*GB;
255		if (gaddr < ctx->highmem && gaddr + len <= ctx->highmem)
256			return ((void *)(ctx->highmem_addr + gaddr));
257	}
258
259	return (NULL);
260}
261
262int
263vm_set_desc(struct vmctx *ctx, int vcpu, int reg,
264	    uint64_t base, uint32_t limit, uint32_t access)
265{
266	int error;
267	struct vm_seg_desc vmsegdesc;
268
269	bzero(&vmsegdesc, sizeof(vmsegdesc));
270	vmsegdesc.cpuid = vcpu;
271	vmsegdesc.regnum = reg;
272	vmsegdesc.desc.base = base;
273	vmsegdesc.desc.limit = limit;
274	vmsegdesc.desc.access = access;
275
276	error = ioctl(ctx->fd, VM_SET_SEGMENT_DESCRIPTOR, &vmsegdesc);
277	return (error);
278}
279
280int
281vm_get_desc(struct vmctx *ctx, int vcpu, int reg,
282	    uint64_t *base, uint32_t *limit, uint32_t *access)
283{
284	int error;
285	struct vm_seg_desc vmsegdesc;
286
287	bzero(&vmsegdesc, sizeof(vmsegdesc));
288	vmsegdesc.cpuid = vcpu;
289	vmsegdesc.regnum = reg;
290
291	error = ioctl(ctx->fd, VM_GET_SEGMENT_DESCRIPTOR, &vmsegdesc);
292	if (error == 0) {
293		*base = vmsegdesc.desc.base;
294		*limit = vmsegdesc.desc.limit;
295		*access = vmsegdesc.desc.access;
296	}
297	return (error);
298}
299
300int
301vm_set_register(struct vmctx *ctx, int vcpu, int reg, uint64_t val)
302{
303	int error;
304	struct vm_register vmreg;
305
306	bzero(&vmreg, sizeof(vmreg));
307	vmreg.cpuid = vcpu;
308	vmreg.regnum = reg;
309	vmreg.regval = val;
310
311	error = ioctl(ctx->fd, VM_SET_REGISTER, &vmreg);
312	return (error);
313}
314
315int
316vm_get_register(struct vmctx *ctx, int vcpu, int reg, uint64_t *ret_val)
317{
318	int error;
319	struct vm_register vmreg;
320
321	bzero(&vmreg, sizeof(vmreg));
322	vmreg.cpuid = vcpu;
323	vmreg.regnum = reg;
324
325	error = ioctl(ctx->fd, VM_GET_REGISTER, &vmreg);
326	*ret_val = vmreg.regval;
327	return (error);
328}
329
330int
331vm_run(struct vmctx *ctx, int vcpu, uint64_t rip, struct vm_exit *vmexit)
332{
333	int error;
334	struct vm_run vmrun;
335
336	bzero(&vmrun, sizeof(vmrun));
337	vmrun.cpuid = vcpu;
338	vmrun.rip = rip;
339
340	error = ioctl(ctx->fd, VM_RUN, &vmrun);
341	bcopy(&vmrun.vm_exit, vmexit, sizeof(struct vm_exit));
342	return (error);
343}
344
345static int
346vm_inject_event_real(struct vmctx *ctx, int vcpu, enum vm_event_type type,
347		     int vector, int error_code, int error_code_valid)
348{
349	struct vm_event ev;
350
351	bzero(&ev, sizeof(ev));
352	ev.cpuid = vcpu;
353	ev.type = type;
354	ev.vector = vector;
355	ev.error_code = error_code;
356	ev.error_code_valid = error_code_valid;
357
358	return (ioctl(ctx->fd, VM_INJECT_EVENT, &ev));
359}
360
361int
362vm_inject_event(struct vmctx *ctx, int vcpu, enum vm_event_type type,
363		int vector)
364{
365
366	return (vm_inject_event_real(ctx, vcpu, type, vector, 0, 0));
367}
368
369int
370vm_inject_event2(struct vmctx *ctx, int vcpu, enum vm_event_type type,
371		 int vector, int error_code)
372{
373
374	return (vm_inject_event_real(ctx, vcpu, type, vector, error_code, 1));
375}
376
377int
378vm_apicid2vcpu(struct vmctx *ctx, int apicid)
379{
380	/*
381	 * The apic id associated with the 'vcpu' has the same numerical value
382	 * as the 'vcpu' itself.
383	 */
384	return (apicid);
385}
386
387int
388vm_lapic_irq(struct vmctx *ctx, int vcpu, int vector)
389{
390	struct vm_lapic_irq vmirq;
391
392	bzero(&vmirq, sizeof(vmirq));
393	vmirq.cpuid = vcpu;
394	vmirq.vector = vector;
395
396	return (ioctl(ctx->fd, VM_LAPIC_IRQ, &vmirq));
397}
398
399int
400vm_lapic_local_irq(struct vmctx *ctx, int vcpu, int vector)
401{
402	struct vm_lapic_irq vmirq;
403
404	bzero(&vmirq, sizeof(vmirq));
405	vmirq.cpuid = vcpu;
406	vmirq.vector = vector;
407
408	return (ioctl(ctx->fd, VM_LAPIC_LOCAL_IRQ, &vmirq));
409}
410
411int
412vm_lapic_msi(struct vmctx *ctx, uint64_t addr, uint64_t msg)
413{
414	struct vm_lapic_msi vmmsi;
415
416	bzero(&vmmsi, sizeof(vmmsi));
417	vmmsi.addr = addr;
418	vmmsi.msg = msg;
419
420	return (ioctl(ctx->fd, VM_LAPIC_MSI, &vmmsi));
421}
422
423int
424vm_ioapic_assert_irq(struct vmctx *ctx, int irq)
425{
426	struct vm_ioapic_irq ioapic_irq;
427
428	bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq));
429	ioapic_irq.irq = irq;
430
431	return (ioctl(ctx->fd, VM_IOAPIC_ASSERT_IRQ, &ioapic_irq));
432}
433
434int
435vm_ioapic_deassert_irq(struct vmctx *ctx, int irq)
436{
437	struct vm_ioapic_irq ioapic_irq;
438
439	bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq));
440	ioapic_irq.irq = irq;
441
442	return (ioctl(ctx->fd, VM_IOAPIC_DEASSERT_IRQ, &ioapic_irq));
443}
444
445int
446vm_ioapic_pulse_irq(struct vmctx *ctx, int irq)
447{
448	struct vm_ioapic_irq ioapic_irq;
449
450	bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq));
451	ioapic_irq.irq = irq;
452
453	return (ioctl(ctx->fd, VM_IOAPIC_PULSE_IRQ, &ioapic_irq));
454}
455
456int
457vm_ioapic_pincount(struct vmctx *ctx, int *pincount)
458{
459
460	return (ioctl(ctx->fd, VM_IOAPIC_PINCOUNT, pincount));
461}
462
463int
464vm_inject_nmi(struct vmctx *ctx, int vcpu)
465{
466	struct vm_nmi vmnmi;
467
468	bzero(&vmnmi, sizeof(vmnmi));
469	vmnmi.cpuid = vcpu;
470
471	return (ioctl(ctx->fd, VM_INJECT_NMI, &vmnmi));
472}
473
474static struct {
475	const char	*name;
476	int		type;
477} capstrmap[] = {
478	{ "hlt_exit",		VM_CAP_HALT_EXIT },
479	{ "mtrap_exit",		VM_CAP_MTRAP_EXIT },
480	{ "pause_exit",		VM_CAP_PAUSE_EXIT },
481	{ "unrestricted_guest",	VM_CAP_UNRESTRICTED_GUEST },
482	{ "enable_invpcid",	VM_CAP_ENABLE_INVPCID },
483	{ 0 }
484};
485
486int
487vm_capability_name2type(const char *capname)
488{
489	int i;
490
491	for (i = 0; capstrmap[i].name != NULL && capname != NULL; i++) {
492		if (strcmp(capstrmap[i].name, capname) == 0)
493			return (capstrmap[i].type);
494	}
495
496	return (-1);
497}
498
499const char *
500vm_capability_type2name(int type)
501{
502	int i;
503
504	for (i = 0; capstrmap[i].name != NULL; i++) {
505		if (capstrmap[i].type == type)
506			return (capstrmap[i].name);
507	}
508
509	return (NULL);
510}
511
512int
513vm_get_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap,
514		  int *retval)
515{
516	int error;
517	struct vm_capability vmcap;
518
519	bzero(&vmcap, sizeof(vmcap));
520	vmcap.cpuid = vcpu;
521	vmcap.captype = cap;
522
523	error = ioctl(ctx->fd, VM_GET_CAPABILITY, &vmcap);
524	*retval = vmcap.capval;
525	return (error);
526}
527
528int
529vm_set_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap, int val)
530{
531	struct vm_capability vmcap;
532
533	bzero(&vmcap, sizeof(vmcap));
534	vmcap.cpuid = vcpu;
535	vmcap.captype = cap;
536	vmcap.capval = val;
537
538	return (ioctl(ctx->fd, VM_SET_CAPABILITY, &vmcap));
539}
540
541int
542vm_assign_pptdev(struct vmctx *ctx, int bus, int slot, int func)
543{
544	struct vm_pptdev pptdev;
545
546	bzero(&pptdev, sizeof(pptdev));
547	pptdev.bus = bus;
548	pptdev.slot = slot;
549	pptdev.func = func;
550
551	return (ioctl(ctx->fd, VM_BIND_PPTDEV, &pptdev));
552}
553
554int
555vm_unassign_pptdev(struct vmctx *ctx, int bus, int slot, int func)
556{
557	struct vm_pptdev pptdev;
558
559	bzero(&pptdev, sizeof(pptdev));
560	pptdev.bus = bus;
561	pptdev.slot = slot;
562	pptdev.func = func;
563
564	return (ioctl(ctx->fd, VM_UNBIND_PPTDEV, &pptdev));
565}
566
567int
568vm_map_pptdev_mmio(struct vmctx *ctx, int bus, int slot, int func,
569		   vm_paddr_t gpa, size_t len, vm_paddr_t hpa)
570{
571	struct vm_pptdev_mmio pptmmio;
572
573	bzero(&pptmmio, sizeof(pptmmio));
574	pptmmio.bus = bus;
575	pptmmio.slot = slot;
576	pptmmio.func = func;
577	pptmmio.gpa = gpa;
578	pptmmio.len = len;
579	pptmmio.hpa = hpa;
580
581	return (ioctl(ctx->fd, VM_MAP_PPTDEV_MMIO, &pptmmio));
582}
583
584int
585vm_setup_pptdev_msi(struct vmctx *ctx, int vcpu, int bus, int slot, int func,
586    uint64_t addr, uint64_t msg, int numvec)
587{
588	struct vm_pptdev_msi pptmsi;
589
590	bzero(&pptmsi, sizeof(pptmsi));
591	pptmsi.vcpu = vcpu;
592	pptmsi.bus = bus;
593	pptmsi.slot = slot;
594	pptmsi.func = func;
595	pptmsi.msg = msg;
596	pptmsi.addr = addr;
597	pptmsi.numvec = numvec;
598
599	return (ioctl(ctx->fd, VM_PPTDEV_MSI, &pptmsi));
600}
601
602int
603vm_setup_pptdev_msix(struct vmctx *ctx, int vcpu, int bus, int slot, int func,
604    int idx, uint64_t addr, uint64_t msg, uint32_t vector_control)
605{
606	struct vm_pptdev_msix pptmsix;
607
608	bzero(&pptmsix, sizeof(pptmsix));
609	pptmsix.vcpu = vcpu;
610	pptmsix.bus = bus;
611	pptmsix.slot = slot;
612	pptmsix.func = func;
613	pptmsix.idx = idx;
614	pptmsix.msg = msg;
615	pptmsix.addr = addr;
616	pptmsix.vector_control = vector_control;
617
618	return ioctl(ctx->fd, VM_PPTDEV_MSIX, &pptmsix);
619}
620
621uint64_t *
622vm_get_stats(struct vmctx *ctx, int vcpu, struct timeval *ret_tv,
623	     int *ret_entries)
624{
625	int error;
626
627	static struct vm_stats vmstats;
628
629	vmstats.cpuid = vcpu;
630
631	error = ioctl(ctx->fd, VM_STATS, &vmstats);
632	if (error == 0) {
633		if (ret_entries)
634			*ret_entries = vmstats.num_entries;
635		if (ret_tv)
636			*ret_tv = vmstats.tv;
637		return (vmstats.statbuf);
638	} else
639		return (NULL);
640}
641
642const char *
643vm_get_stat_desc(struct vmctx *ctx, int index)
644{
645	static struct vm_stat_desc statdesc;
646
647	statdesc.index = index;
648	if (ioctl(ctx->fd, VM_STAT_DESC, &statdesc) == 0)
649		return (statdesc.desc);
650	else
651		return (NULL);
652}
653
654int
655vm_get_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state *state)
656{
657	int error;
658	struct vm_x2apic x2apic;
659
660	bzero(&x2apic, sizeof(x2apic));
661	x2apic.cpuid = vcpu;
662
663	error = ioctl(ctx->fd, VM_GET_X2APIC_STATE, &x2apic);
664	*state = x2apic.state;
665	return (error);
666}
667
668int
669vm_set_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state state)
670{
671	int error;
672	struct vm_x2apic x2apic;
673
674	bzero(&x2apic, sizeof(x2apic));
675	x2apic.cpuid = vcpu;
676	x2apic.state = state;
677
678	error = ioctl(ctx->fd, VM_SET_X2APIC_STATE, &x2apic);
679
680	return (error);
681}
682
683/*
684 * From Intel Vol 3a:
685 * Table 9-1. IA-32 Processor States Following Power-up, Reset or INIT
686 */
687int
688vcpu_reset(struct vmctx *vmctx, int vcpu)
689{
690	int error;
691	uint64_t rflags, rip, cr0, cr4, zero, desc_base, rdx;
692	uint32_t desc_access, desc_limit;
693	uint16_t sel;
694
695	zero = 0;
696
697	rflags = 0x2;
698	error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RFLAGS, rflags);
699	if (error)
700		goto done;
701
702	rip = 0xfff0;
703	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RIP, rip)) != 0)
704		goto done;
705
706	cr0 = CR0_NE;
707	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR0, cr0)) != 0)
708		goto done;
709
710	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR3, zero)) != 0)
711		goto done;
712
713	cr4 = 0;
714	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR4, cr4)) != 0)
715		goto done;
716
717	/*
718	 * CS: present, r/w, accessed, 16-bit, byte granularity, usable
719	 */
720	desc_base = 0xffff0000;
721	desc_limit = 0xffff;
722	desc_access = 0x0093;
723	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_CS,
724			    desc_base, desc_limit, desc_access);
725	if (error)
726		goto done;
727
728	sel = 0xf000;
729	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CS, sel)) != 0)
730		goto done;
731
732	/*
733	 * SS,DS,ES,FS,GS: present, r/w, accessed, 16-bit, byte granularity
734	 */
735	desc_base = 0;
736	desc_limit = 0xffff;
737	desc_access = 0x0093;
738	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_SS,
739			    desc_base, desc_limit, desc_access);
740	if (error)
741		goto done;
742
743	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_DS,
744			    desc_base, desc_limit, desc_access);
745	if (error)
746		goto done;
747
748	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_ES,
749			    desc_base, desc_limit, desc_access);
750	if (error)
751		goto done;
752
753	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_FS,
754			    desc_base, desc_limit, desc_access);
755	if (error)
756		goto done;
757
758	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_GS,
759			    desc_base, desc_limit, desc_access);
760	if (error)
761		goto done;
762
763	sel = 0;
764	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_SS, sel)) != 0)
765		goto done;
766	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_DS, sel)) != 0)
767		goto done;
768	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_ES, sel)) != 0)
769		goto done;
770	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_FS, sel)) != 0)
771		goto done;
772	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_GS, sel)) != 0)
773		goto done;
774
775	/* General purpose registers */
776	rdx = 0xf00;
777	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RAX, zero)) != 0)
778		goto done;
779	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RBX, zero)) != 0)
780		goto done;
781	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RCX, zero)) != 0)
782		goto done;
783	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RDX, rdx)) != 0)
784		goto done;
785	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RSI, zero)) != 0)
786		goto done;
787	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RDI, zero)) != 0)
788		goto done;
789	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RBP, zero)) != 0)
790		goto done;
791	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RSP, zero)) != 0)
792		goto done;
793
794	/* GDTR, IDTR */
795	desc_base = 0;
796	desc_limit = 0xffff;
797	desc_access = 0;
798	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_GDTR,
799			    desc_base, desc_limit, desc_access);
800	if (error != 0)
801		goto done;
802
803	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_IDTR,
804			    desc_base, desc_limit, desc_access);
805	if (error != 0)
806		goto done;
807
808	/* TR */
809	desc_base = 0;
810	desc_limit = 0xffff;
811	desc_access = 0x0000008b;
812	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_TR, 0, 0, desc_access);
813	if (error)
814		goto done;
815
816	sel = 0;
817	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_TR, sel)) != 0)
818		goto done;
819
820	/* LDTR */
821	desc_base = 0;
822	desc_limit = 0xffff;
823	desc_access = 0x00000082;
824	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_LDTR, desc_base,
825			    desc_limit, desc_access);
826	if (error)
827		goto done;
828
829	sel = 0;
830	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_LDTR, 0)) != 0)
831		goto done;
832
833	/* XXX cr2, debug registers */
834
835	error = 0;
836done:
837	return (error);
838}
839
840int
841vm_get_gpa_pmap(struct vmctx *ctx, uint64_t gpa, uint64_t *pte, int *num)
842{
843	int error, i;
844	struct vm_gpa_pte gpapte;
845
846	bzero(&gpapte, sizeof(gpapte));
847	gpapte.gpa = gpa;
848
849	error = ioctl(ctx->fd, VM_GET_GPA_PMAP, &gpapte);
850
851	if (error == 0) {
852		*num = gpapte.ptenum;
853		for (i = 0; i < gpapte.ptenum; i++)
854			pte[i] = gpapte.pte[i];
855	}
856
857	return (error);
858}
859
860int
861vm_get_hpet_capabilities(struct vmctx *ctx, uint32_t *capabilities)
862{
863	int error;
864	struct vm_hpet_cap cap;
865
866	bzero(&cap, sizeof(struct vm_hpet_cap));
867	error = ioctl(ctx->fd, VM_GET_HPET_CAPABILITIES, &cap);
868	if (capabilities != NULL)
869		*capabilities = cap.capabilities;
870	return (error);
871}
872