xref: /freebsd/lib/libvmmapi/vmmapi.c (revision 9ecd54f24fe9fa373e07c9fd7c052deb2188f545)
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  * $FreeBSD$
27  */
28 
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
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 
56 struct 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 
70 static int
71 vm_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 
88 int
89 vm_create(const char *name)
90 {
91 
92 	return (CREATE((char *)name));
93 }
94 
95 struct vmctx *
96 vm_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);
112 err:
113 	vm_destroy(vm);
114 	return (NULL);
115 }
116 
117 void
118 vm_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 
129 int
130 vm_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 
153 int
154 vm_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 
169 uint32_t
170 vm_get_lowmem_limit(struct vmctx *ctx)
171 {
172 
173 	return (ctx->lowmem_limit);
174 }
175 
176 void
177 vm_set_lowmem_limit(struct vmctx *ctx, uint32_t limit)
178 {
179 
180 	ctx->lowmem_limit = limit;
181 }
182 
183 static int
184 setup_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 
204 int
205 vm_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 
243 void *
244 vm_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 
262 int
263 vm_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 
280 int
281 vm_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 
300 int
301 vm_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 
315 int
316 vm_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 
330 int
331 vm_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 
345 int
346 vm_suspend(struct vmctx *ctx)
347 {
348 
349 	return (ioctl(ctx->fd, VM_SUSPEND, 0));
350 }
351 
352 static int
353 vm_inject_exception_real(struct vmctx *ctx, int vcpu, int vector,
354     int error_code, int error_code_valid)
355 {
356 	struct vm_exception exc;
357 
358 	bzero(&exc, sizeof(exc));
359 	exc.cpuid = vcpu;
360 	exc.vector = vector;
361 	exc.error_code = error_code;
362 	exc.error_code_valid = error_code_valid;
363 
364 	return (ioctl(ctx->fd, VM_INJECT_EXCEPTION, &exc));
365 }
366 
367 int
368 vm_inject_exception(struct vmctx *ctx, int vcpu, int vector)
369 {
370 
371 	return (vm_inject_exception_real(ctx, vcpu, vector, 0, 0));
372 }
373 
374 int
375 vm_inject_exception2(struct vmctx *ctx, int vcpu, int vector, int errcode)
376 {
377 
378 	return (vm_inject_exception_real(ctx, vcpu, vector, errcode, 1));
379 }
380 
381 int
382 vm_apicid2vcpu(struct vmctx *ctx, int apicid)
383 {
384 	/*
385 	 * The apic id associated with the 'vcpu' has the same numerical value
386 	 * as the 'vcpu' itself.
387 	 */
388 	return (apicid);
389 }
390 
391 int
392 vm_lapic_irq(struct vmctx *ctx, int vcpu, int vector)
393 {
394 	struct vm_lapic_irq vmirq;
395 
396 	bzero(&vmirq, sizeof(vmirq));
397 	vmirq.cpuid = vcpu;
398 	vmirq.vector = vector;
399 
400 	return (ioctl(ctx->fd, VM_LAPIC_IRQ, &vmirq));
401 }
402 
403 int
404 vm_lapic_local_irq(struct vmctx *ctx, int vcpu, int vector)
405 {
406 	struct vm_lapic_irq vmirq;
407 
408 	bzero(&vmirq, sizeof(vmirq));
409 	vmirq.cpuid = vcpu;
410 	vmirq.vector = vector;
411 
412 	return (ioctl(ctx->fd, VM_LAPIC_LOCAL_IRQ, &vmirq));
413 }
414 
415 int
416 vm_lapic_msi(struct vmctx *ctx, uint64_t addr, uint64_t msg)
417 {
418 	struct vm_lapic_msi vmmsi;
419 
420 	bzero(&vmmsi, sizeof(vmmsi));
421 	vmmsi.addr = addr;
422 	vmmsi.msg = msg;
423 
424 	return (ioctl(ctx->fd, VM_LAPIC_MSI, &vmmsi));
425 }
426 
427 int
428 vm_ioapic_assert_irq(struct vmctx *ctx, int irq)
429 {
430 	struct vm_ioapic_irq ioapic_irq;
431 
432 	bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq));
433 	ioapic_irq.irq = irq;
434 
435 	return (ioctl(ctx->fd, VM_IOAPIC_ASSERT_IRQ, &ioapic_irq));
436 }
437 
438 int
439 vm_ioapic_deassert_irq(struct vmctx *ctx, int irq)
440 {
441 	struct vm_ioapic_irq ioapic_irq;
442 
443 	bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq));
444 	ioapic_irq.irq = irq;
445 
446 	return (ioctl(ctx->fd, VM_IOAPIC_DEASSERT_IRQ, &ioapic_irq));
447 }
448 
449 int
450 vm_ioapic_pulse_irq(struct vmctx *ctx, int irq)
451 {
452 	struct vm_ioapic_irq ioapic_irq;
453 
454 	bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq));
455 	ioapic_irq.irq = irq;
456 
457 	return (ioctl(ctx->fd, VM_IOAPIC_PULSE_IRQ, &ioapic_irq));
458 }
459 
460 int
461 vm_ioapic_pincount(struct vmctx *ctx, int *pincount)
462 {
463 
464 	return (ioctl(ctx->fd, VM_IOAPIC_PINCOUNT, pincount));
465 }
466 
467 int
468 vm_isa_assert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq)
469 {
470 	struct vm_isa_irq isa_irq;
471 
472 	bzero(&isa_irq, sizeof(struct vm_isa_irq));
473 	isa_irq.atpic_irq = atpic_irq;
474 	isa_irq.ioapic_irq = ioapic_irq;
475 
476 	return (ioctl(ctx->fd, VM_ISA_ASSERT_IRQ, &isa_irq));
477 }
478 
479 int
480 vm_isa_deassert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq)
481 {
482 	struct vm_isa_irq isa_irq;
483 
484 	bzero(&isa_irq, sizeof(struct vm_isa_irq));
485 	isa_irq.atpic_irq = atpic_irq;
486 	isa_irq.ioapic_irq = ioapic_irq;
487 
488 	return (ioctl(ctx->fd, VM_ISA_DEASSERT_IRQ, &isa_irq));
489 }
490 
491 int
492 vm_isa_pulse_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq)
493 {
494 	struct vm_isa_irq isa_irq;
495 	bzero(&isa_irq, sizeof(struct vm_isa_irq));
496 	isa_irq.atpic_irq = atpic_irq;
497 	isa_irq.ioapic_irq = ioapic_irq;
498 
499 	return (ioctl(ctx->fd, VM_ISA_PULSE_IRQ, &isa_irq));
500 }
501 
502 int
503 vm_inject_nmi(struct vmctx *ctx, int vcpu)
504 {
505 	struct vm_nmi vmnmi;
506 
507 	bzero(&vmnmi, sizeof(vmnmi));
508 	vmnmi.cpuid = vcpu;
509 
510 	return (ioctl(ctx->fd, VM_INJECT_NMI, &vmnmi));
511 }
512 
513 static struct {
514 	const char	*name;
515 	int		type;
516 } capstrmap[] = {
517 	{ "hlt_exit",		VM_CAP_HALT_EXIT },
518 	{ "mtrap_exit",		VM_CAP_MTRAP_EXIT },
519 	{ "pause_exit",		VM_CAP_PAUSE_EXIT },
520 	{ "unrestricted_guest",	VM_CAP_UNRESTRICTED_GUEST },
521 	{ "enable_invpcid",	VM_CAP_ENABLE_INVPCID },
522 	{ 0 }
523 };
524 
525 int
526 vm_capability_name2type(const char *capname)
527 {
528 	int i;
529 
530 	for (i = 0; capstrmap[i].name != NULL && capname != NULL; i++) {
531 		if (strcmp(capstrmap[i].name, capname) == 0)
532 			return (capstrmap[i].type);
533 	}
534 
535 	return (-1);
536 }
537 
538 const char *
539 vm_capability_type2name(int type)
540 {
541 	int i;
542 
543 	for (i = 0; capstrmap[i].name != NULL; i++) {
544 		if (capstrmap[i].type == type)
545 			return (capstrmap[i].name);
546 	}
547 
548 	return (NULL);
549 }
550 
551 int
552 vm_get_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap,
553 		  int *retval)
554 {
555 	int error;
556 	struct vm_capability vmcap;
557 
558 	bzero(&vmcap, sizeof(vmcap));
559 	vmcap.cpuid = vcpu;
560 	vmcap.captype = cap;
561 
562 	error = ioctl(ctx->fd, VM_GET_CAPABILITY, &vmcap);
563 	*retval = vmcap.capval;
564 	return (error);
565 }
566 
567 int
568 vm_set_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap, int val)
569 {
570 	struct vm_capability vmcap;
571 
572 	bzero(&vmcap, sizeof(vmcap));
573 	vmcap.cpuid = vcpu;
574 	vmcap.captype = cap;
575 	vmcap.capval = val;
576 
577 	return (ioctl(ctx->fd, VM_SET_CAPABILITY, &vmcap));
578 }
579 
580 int
581 vm_assign_pptdev(struct vmctx *ctx, int bus, int slot, int func)
582 {
583 	struct vm_pptdev pptdev;
584 
585 	bzero(&pptdev, sizeof(pptdev));
586 	pptdev.bus = bus;
587 	pptdev.slot = slot;
588 	pptdev.func = func;
589 
590 	return (ioctl(ctx->fd, VM_BIND_PPTDEV, &pptdev));
591 }
592 
593 int
594 vm_unassign_pptdev(struct vmctx *ctx, int bus, int slot, int func)
595 {
596 	struct vm_pptdev pptdev;
597 
598 	bzero(&pptdev, sizeof(pptdev));
599 	pptdev.bus = bus;
600 	pptdev.slot = slot;
601 	pptdev.func = func;
602 
603 	return (ioctl(ctx->fd, VM_UNBIND_PPTDEV, &pptdev));
604 }
605 
606 int
607 vm_map_pptdev_mmio(struct vmctx *ctx, int bus, int slot, int func,
608 		   vm_paddr_t gpa, size_t len, vm_paddr_t hpa)
609 {
610 	struct vm_pptdev_mmio pptmmio;
611 
612 	bzero(&pptmmio, sizeof(pptmmio));
613 	pptmmio.bus = bus;
614 	pptmmio.slot = slot;
615 	pptmmio.func = func;
616 	pptmmio.gpa = gpa;
617 	pptmmio.len = len;
618 	pptmmio.hpa = hpa;
619 
620 	return (ioctl(ctx->fd, VM_MAP_PPTDEV_MMIO, &pptmmio));
621 }
622 
623 int
624 vm_setup_pptdev_msi(struct vmctx *ctx, int vcpu, int bus, int slot, int func,
625     uint64_t addr, uint64_t msg, int numvec)
626 {
627 	struct vm_pptdev_msi pptmsi;
628 
629 	bzero(&pptmsi, sizeof(pptmsi));
630 	pptmsi.vcpu = vcpu;
631 	pptmsi.bus = bus;
632 	pptmsi.slot = slot;
633 	pptmsi.func = func;
634 	pptmsi.msg = msg;
635 	pptmsi.addr = addr;
636 	pptmsi.numvec = numvec;
637 
638 	return (ioctl(ctx->fd, VM_PPTDEV_MSI, &pptmsi));
639 }
640 
641 int
642 vm_setup_pptdev_msix(struct vmctx *ctx, int vcpu, int bus, int slot, int func,
643     int idx, uint64_t addr, uint64_t msg, uint32_t vector_control)
644 {
645 	struct vm_pptdev_msix pptmsix;
646 
647 	bzero(&pptmsix, sizeof(pptmsix));
648 	pptmsix.vcpu = vcpu;
649 	pptmsix.bus = bus;
650 	pptmsix.slot = slot;
651 	pptmsix.func = func;
652 	pptmsix.idx = idx;
653 	pptmsix.msg = msg;
654 	pptmsix.addr = addr;
655 	pptmsix.vector_control = vector_control;
656 
657 	return ioctl(ctx->fd, VM_PPTDEV_MSIX, &pptmsix);
658 }
659 
660 uint64_t *
661 vm_get_stats(struct vmctx *ctx, int vcpu, struct timeval *ret_tv,
662 	     int *ret_entries)
663 {
664 	int error;
665 
666 	static struct vm_stats vmstats;
667 
668 	vmstats.cpuid = vcpu;
669 
670 	error = ioctl(ctx->fd, VM_STATS, &vmstats);
671 	if (error == 0) {
672 		if (ret_entries)
673 			*ret_entries = vmstats.num_entries;
674 		if (ret_tv)
675 			*ret_tv = vmstats.tv;
676 		return (vmstats.statbuf);
677 	} else
678 		return (NULL);
679 }
680 
681 const char *
682 vm_get_stat_desc(struct vmctx *ctx, int index)
683 {
684 	static struct vm_stat_desc statdesc;
685 
686 	statdesc.index = index;
687 	if (ioctl(ctx->fd, VM_STAT_DESC, &statdesc) == 0)
688 		return (statdesc.desc);
689 	else
690 		return (NULL);
691 }
692 
693 int
694 vm_get_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state *state)
695 {
696 	int error;
697 	struct vm_x2apic x2apic;
698 
699 	bzero(&x2apic, sizeof(x2apic));
700 	x2apic.cpuid = vcpu;
701 
702 	error = ioctl(ctx->fd, VM_GET_X2APIC_STATE, &x2apic);
703 	*state = x2apic.state;
704 	return (error);
705 }
706 
707 int
708 vm_set_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state state)
709 {
710 	int error;
711 	struct vm_x2apic x2apic;
712 
713 	bzero(&x2apic, sizeof(x2apic));
714 	x2apic.cpuid = vcpu;
715 	x2apic.state = state;
716 
717 	error = ioctl(ctx->fd, VM_SET_X2APIC_STATE, &x2apic);
718 
719 	return (error);
720 }
721 
722 /*
723  * From Intel Vol 3a:
724  * Table 9-1. IA-32 Processor States Following Power-up, Reset or INIT
725  */
726 int
727 vcpu_reset(struct vmctx *vmctx, int vcpu)
728 {
729 	int error;
730 	uint64_t rflags, rip, cr0, cr4, zero, desc_base, rdx;
731 	uint32_t desc_access, desc_limit;
732 	uint16_t sel;
733 
734 	zero = 0;
735 
736 	rflags = 0x2;
737 	error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RFLAGS, rflags);
738 	if (error)
739 		goto done;
740 
741 	rip = 0xfff0;
742 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RIP, rip)) != 0)
743 		goto done;
744 
745 	cr0 = CR0_NE;
746 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR0, cr0)) != 0)
747 		goto done;
748 
749 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR3, zero)) != 0)
750 		goto done;
751 
752 	cr4 = 0;
753 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR4, cr4)) != 0)
754 		goto done;
755 
756 	/*
757 	 * CS: present, r/w, accessed, 16-bit, byte granularity, usable
758 	 */
759 	desc_base = 0xffff0000;
760 	desc_limit = 0xffff;
761 	desc_access = 0x0093;
762 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_CS,
763 			    desc_base, desc_limit, desc_access);
764 	if (error)
765 		goto done;
766 
767 	sel = 0xf000;
768 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CS, sel)) != 0)
769 		goto done;
770 
771 	/*
772 	 * SS,DS,ES,FS,GS: present, r/w, accessed, 16-bit, byte granularity
773 	 */
774 	desc_base = 0;
775 	desc_limit = 0xffff;
776 	desc_access = 0x0093;
777 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_SS,
778 			    desc_base, desc_limit, desc_access);
779 	if (error)
780 		goto done;
781 
782 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_DS,
783 			    desc_base, desc_limit, desc_access);
784 	if (error)
785 		goto done;
786 
787 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_ES,
788 			    desc_base, desc_limit, desc_access);
789 	if (error)
790 		goto done;
791 
792 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_FS,
793 			    desc_base, desc_limit, desc_access);
794 	if (error)
795 		goto done;
796 
797 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_GS,
798 			    desc_base, desc_limit, desc_access);
799 	if (error)
800 		goto done;
801 
802 	sel = 0;
803 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_SS, sel)) != 0)
804 		goto done;
805 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_DS, sel)) != 0)
806 		goto done;
807 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_ES, sel)) != 0)
808 		goto done;
809 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_FS, sel)) != 0)
810 		goto done;
811 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_GS, sel)) != 0)
812 		goto done;
813 
814 	/* General purpose registers */
815 	rdx = 0xf00;
816 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RAX, zero)) != 0)
817 		goto done;
818 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RBX, zero)) != 0)
819 		goto done;
820 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RCX, zero)) != 0)
821 		goto done;
822 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RDX, rdx)) != 0)
823 		goto done;
824 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RSI, zero)) != 0)
825 		goto done;
826 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RDI, zero)) != 0)
827 		goto done;
828 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RBP, zero)) != 0)
829 		goto done;
830 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RSP, zero)) != 0)
831 		goto done;
832 
833 	/* GDTR, IDTR */
834 	desc_base = 0;
835 	desc_limit = 0xffff;
836 	desc_access = 0;
837 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_GDTR,
838 			    desc_base, desc_limit, desc_access);
839 	if (error != 0)
840 		goto done;
841 
842 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_IDTR,
843 			    desc_base, desc_limit, desc_access);
844 	if (error != 0)
845 		goto done;
846 
847 	/* TR */
848 	desc_base = 0;
849 	desc_limit = 0xffff;
850 	desc_access = 0x0000008b;
851 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_TR, 0, 0, desc_access);
852 	if (error)
853 		goto done;
854 
855 	sel = 0;
856 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_TR, sel)) != 0)
857 		goto done;
858 
859 	/* LDTR */
860 	desc_base = 0;
861 	desc_limit = 0xffff;
862 	desc_access = 0x00000082;
863 	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_LDTR, desc_base,
864 			    desc_limit, desc_access);
865 	if (error)
866 		goto done;
867 
868 	sel = 0;
869 	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_LDTR, 0)) != 0)
870 		goto done;
871 
872 	/* XXX cr2, debug registers */
873 
874 	error = 0;
875 done:
876 	return (error);
877 }
878 
879 int
880 vm_get_gpa_pmap(struct vmctx *ctx, uint64_t gpa, uint64_t *pte, int *num)
881 {
882 	int error, i;
883 	struct vm_gpa_pte gpapte;
884 
885 	bzero(&gpapte, sizeof(gpapte));
886 	gpapte.gpa = gpa;
887 
888 	error = ioctl(ctx->fd, VM_GET_GPA_PMAP, &gpapte);
889 
890 	if (error == 0) {
891 		*num = gpapte.ptenum;
892 		for (i = 0; i < gpapte.ptenum; i++)
893 			pte[i] = gpapte.pte[i];
894 	}
895 
896 	return (error);
897 }
898 
899 int
900 vm_get_hpet_capabilities(struct vmctx *ctx, uint32_t *capabilities)
901 {
902 	int error;
903 	struct vm_hpet_cap cap;
904 
905 	bzero(&cap, sizeof(struct vm_hpet_cap));
906 	error = ioctl(ctx->fd, VM_GET_HPET_CAPABILITIES, &cap);
907 	if (capabilities != NULL)
908 		*capabilities = cap.capabilities;
909 	return (error);
910 }
911