Import of bhyve hypervisor and utilities, part 1.

  vmm.ko - kernel module for VT-x, VT-d and hypervisor control
  bhyve  - user-space sequencer and i/o emulation
  vmmctl - dump of hypervisor register state
  libvmm - front-end to vmm.ko chardev interface

bhyve was designed and implemented by Neel Natu.

Thanks to the following folk from NetApp who helped to make this available:
	Joe CaraDonna
	Peter Snyder
	Jeff Heller
	Sandeep Mann
	Steve Miller
	Brian Pawlowski

1 files changed, 647 insertions, 0 deletions

diff --git a/lib/libvmmapi/vmmapi.c b/lib/libvmmapi/vmmapi.c
new file mode 100644
index 0000000..7f95fea
--- /dev/null
+++ b/lib/libvmmapi/vmmapi.c
@@ -0,0 +1,647 @@
+/*-
+ * Copyright (c) 2011 NetApp, Inc.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * $FreeBSD$
+ */
+
+#include <sys/cdefs.h>
+__FBSDID("$FreeBSD$");
+
+#include <sys/types.h>
+#include <sys/sysctl.h>
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+
+#include <machine/specialreg.h>
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <assert.h>
+#include <string.h>
+#include <fcntl.h>
+#include <unistd.h>
+
+#include <machine/vmm.h>
+#include <machine/vmm_dev.h>
+
+#include "vmmapi.h"
+#include "mptable.h"
+
+#ifndef	CR4_VMXE
+#define	CR4_VMXE	(1UL << 13)
+#endif
+
+#define BIOS_ROM_BASE		(0xf0000)
+#define BIOS_ROM_SIZE		(0x10000)
+
+struct vmctx {
+	int	fd;
+	char	*name;
+};
+
+#define	CREATE(x)  sysctlbyname("hw.vmm.create", NULL, NULL, (x), strlen((x)))
+#define	DESTROY(x) sysctlbyname("hw.vmm.destroy", NULL, NULL, (x), strlen((x)))
+
+static int
+vm_device_open(const char *name)
+{
+        int fd, len;
+        char *vmfile;
+
+	len = strlen("/dev/vmm/") + strlen(name) + 1;
+	vmfile = malloc(len);
+	assert(vmfile != NULL);
+	snprintf(vmfile, len, "/dev/vmm/%s", name);
+
+        /* Open the device file */
+        fd = open(vmfile, O_RDWR, 0);
+
+	free(vmfile);
+        return (fd);
+}
+
+int
+vm_create(const char *name)
+{
+
+	return (CREATE((char *)name));
+}
+
+struct vmctx *
+vm_open(const char *name)
+{
+	struct vmctx *vm;
+
+	vm = malloc(sizeof(struct vmctx) + strlen(name) + 1);
+	assert(vm != NULL);
+
+	vm->fd = -1;
+	vm->name = (char *)(vm + 1);
+	strcpy(vm->name, name);
+
+	if ((vm->fd = vm_device_open(vm->name)) < 0)
+		goto err;
+
+	return (vm);
+err:
+	vm_destroy(vm);
+	return (NULL);
+}
+
+void
+vm_destroy(struct vmctx *vm)
+{
+	assert(vm != NULL);
+
+	DESTROY(vm->name);
+	if (vm->fd >= 0)
+		close(vm->fd);
+	free(vm);
+}
+
+int
+vm_get_memory_seg(struct vmctx *ctx, vm_paddr_t gpa,
+		  vm_paddr_t *ret_hpa, size_t *ret_len)
+{
+	int error;
+	struct vm_memory_segment seg;
+
+	bzero(&seg, sizeof(seg));
+	seg.gpa = gpa;
+	error = ioctl(ctx->fd, VM_GET_MEMORY_SEG, &seg);
+	*ret_hpa = seg.hpa;
+	*ret_len = seg.len;
+	return (error);
+}
+
+int
+vm_setup_memory(struct vmctx *ctx, vm_paddr_t gpa, size_t len, char **mapaddr)
+{
+	int error;
+	struct vm_memory_segment seg;
+
+	/*
+	 * Create and optionally map 'len' bytes of memory at guest
+	 * physical address 'gpa'
+	 */
+	bzero(&seg, sizeof(seg));
+	seg.gpa = gpa;
+	seg.len = len;
+	error = ioctl(ctx->fd, VM_MAP_MEMORY, &seg);
+	if (error == 0 && mapaddr != NULL) {
+		*mapaddr = mmap(NULL, len, PROT_READ | PROT_WRITE, MAP_SHARED,
+				ctx->fd, gpa);
+	}
+	return (error);
+}
+
+char *
+vm_map_memory(struct vmctx *ctx, vm_paddr_t gpa, size_t len)
+{
+
+	/* Map 'len' bytes of memory at guest physical address 'gpa' */
+	return ((char *)mmap(NULL, len, PROT_READ | PROT_WRITE, MAP_SHARED,
+		     ctx->fd, gpa));
+}
+
+int
+vm_set_desc(struct vmctx *ctx, int vcpu, int reg,
+	    uint64_t base, uint32_t limit, uint32_t access)
+{
+	int error;
+	struct vm_seg_desc vmsegdesc;
+
+	bzero(&vmsegdesc, sizeof(vmsegdesc));
+	vmsegdesc.cpuid = vcpu;
+	vmsegdesc.regnum = reg;
+	vmsegdesc.desc.base = base;
+	vmsegdesc.desc.limit = limit;
+	vmsegdesc.desc.access = access;
+
+	error = ioctl(ctx->fd, VM_SET_SEGMENT_DESCRIPTOR, &vmsegdesc);
+	return (error);
+}
+
+int
+vm_get_desc(struct vmctx *ctx, int vcpu, int reg,
+	    uint64_t *base, uint32_t *limit, uint32_t *access)
+{
+	int error;
+	struct vm_seg_desc vmsegdesc;
+
+	bzero(&vmsegdesc, sizeof(vmsegdesc));
+	vmsegdesc.cpuid = vcpu;
+	vmsegdesc.regnum = reg;
+
+	error = ioctl(ctx->fd, VM_GET_SEGMENT_DESCRIPTOR, &vmsegdesc);
+	if (error == 0) {
+		*base = vmsegdesc.desc.base;
+		*limit = vmsegdesc.desc.limit;
+		*access = vmsegdesc.desc.access;
+	}
+	return (error);
+}
+
+int
+vm_set_register(struct vmctx *ctx, int vcpu, int reg, uint64_t val)
+{
+	int error;
+	struct vm_register vmreg;
+
+	bzero(&vmreg, sizeof(vmreg));
+	vmreg.cpuid = vcpu;
+	vmreg.regnum = reg;
+	vmreg.regval = val;
+
+	error = ioctl(ctx->fd, VM_SET_REGISTER, &vmreg);
+	return (error);
+}
+
+int
+vm_get_register(struct vmctx *ctx, int vcpu, int reg, uint64_t *ret_val)
+{
+	int error;
+	struct vm_register vmreg;
+
+	bzero(&vmreg, sizeof(vmreg));
+	vmreg.cpuid = vcpu;
+	vmreg.regnum = reg;
+
+	error = ioctl(ctx->fd, VM_GET_REGISTER, &vmreg);
+	*ret_val = vmreg.regval;
+	return (error);
+}
+
+int
+vm_get_pinning(struct vmctx *ctx, int vcpu, int *host_cpuid)
+{
+	int error;
+	struct vm_pin vmpin;
+
+	bzero(&vmpin, sizeof(vmpin));
+	vmpin.vm_cpuid = vcpu;
+
+	error = ioctl(ctx->fd, VM_GET_PINNING, &vmpin);
+	*host_cpuid = vmpin.host_cpuid;
+	return (error);
+}
+
+int
+vm_set_pinning(struct vmctx *ctx, int vcpu, int host_cpuid)
+{
+	int error;
+	struct vm_pin vmpin;
+
+	bzero(&vmpin, sizeof(vmpin));
+	vmpin.vm_cpuid = vcpu;
+	vmpin.host_cpuid = host_cpuid;
+
+	error = ioctl(ctx->fd, VM_SET_PINNING, &vmpin);
+	return (error);
+}
+
+int
+vm_run(struct vmctx *ctx, int vcpu, uint64_t rip, struct vm_exit *vmexit)
+{
+	int error;
+	struct vm_run vmrun;
+
+	bzero(&vmrun, sizeof(vmrun));
+	vmrun.cpuid = vcpu;
+	vmrun.rip = rip;
+
+	error = ioctl(ctx->fd, VM_RUN, &vmrun);
+	bcopy(&vmrun.vm_exit, vmexit, sizeof(struct vm_exit));
+	return (error);
+}
+
+static int
+vm_inject_event_real(struct vmctx *ctx, int vcpu, enum vm_event_type type,
+		     int vector, int error_code, int error_code_valid)
+{
+	struct vm_event ev;
+
+	bzero(&ev, sizeof(ev));
+	ev.cpuid = vcpu;
+	ev.type = type;
+	ev.vector = vector;
+	ev.error_code = error_code;
+	ev.error_code_valid = error_code_valid;
+
+	return (ioctl(ctx->fd, VM_INJECT_EVENT, &ev));
+}
+
+int
+vm_inject_event(struct vmctx *ctx, int vcpu, enum vm_event_type type,
+		int vector)
+{
+
+	return (vm_inject_event_real(ctx, vcpu, type, vector, 0, 0));
+}
+
+int
+vm_inject_event2(struct vmctx *ctx, int vcpu, enum vm_event_type type,
+		 int vector, int error_code)
+{
+
+	return (vm_inject_event_real(ctx, vcpu, type, vector, error_code, 1));
+}
+
+int
+vm_build_tables(struct vmctx *ctxt, int ncpu, void *oemtbl, int oemtblsz)
+{
+
+	return (vm_build_mptable(ctxt, BIOS_ROM_BASE, BIOS_ROM_SIZE, ncpu,
+				 oemtbl, oemtblsz));
+}
+
+int
+vm_lapic_irq(struct vmctx *ctx, int vcpu, int vector)
+{
+	struct vm_lapic_irq vmirq;
+
+	bzero(&vmirq, sizeof(vmirq));
+	vmirq.cpuid = vcpu;
+	vmirq.vector = vector;
+
+	return (ioctl(ctx->fd, VM_LAPIC_IRQ, &vmirq));
+}
+
+int
+vm_inject_nmi(struct vmctx *ctx, int vcpu)
+{
+	struct vm_nmi vmnmi;
+
+	bzero(&vmnmi, sizeof(vmnmi));
+	vmnmi.cpuid = vcpu;
+
+	return (ioctl(ctx->fd, VM_INJECT_NMI, &vmnmi));
+}
+
+int
+vm_capability_name2type(const char *capname)
+{
+	int i;
+
+	static struct {
+		const char	*name;
+		int		type;
+	} capstrmap[] = {
+		{ "hlt_exit",		VM_CAP_HALT_EXIT },
+		{ "mtrap_exit",		VM_CAP_MTRAP_EXIT },
+		{ "pause_exit",		VM_CAP_PAUSE_EXIT },
+		{ "unrestricted_guest",	VM_CAP_UNRESTRICTED_GUEST },
+		{ 0 }
+	};
+
+	for (i = 0; capstrmap[i].name != NULL && capname != NULL; i++) {
+		if (strcmp(capstrmap[i].name, capname) == 0)
+			return (capstrmap[i].type);
+	}
+
+	return (-1);
+}
+
+int
+vm_get_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap,
+		  int *retval)
+{
+	int error;
+	struct vm_capability vmcap;
+
+	bzero(&vmcap, sizeof(vmcap));
+	vmcap.cpuid = vcpu;
+	vmcap.captype = cap;
+
+	error = ioctl(ctx->fd, VM_GET_CAPABILITY, &vmcap);
+	*retval = vmcap.capval;
+	return (error);
+}
+
+int
+vm_set_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap, int val)
+{
+	struct vm_capability vmcap;
+
+	bzero(&vmcap, sizeof(vmcap));
+	vmcap.cpuid = vcpu;
+	vmcap.captype = cap;
+	vmcap.capval = val;
+	
+	return (ioctl(ctx->fd, VM_SET_CAPABILITY, &vmcap));
+}
+
+int
+vm_assign_pptdev(struct vmctx *ctx, int bus, int slot, int func)
+{
+	struct vm_pptdev pptdev;
+
+	bzero(&pptdev, sizeof(pptdev));
+	pptdev.bus = bus;
+	pptdev.slot = slot;
+	pptdev.func = func;
+
+	return (ioctl(ctx->fd, VM_BIND_PPTDEV, &pptdev));
+}
+
+int
+vm_unassign_pptdev(struct vmctx *ctx, int bus, int slot, int func)
+{
+	struct vm_pptdev pptdev;
+
+	bzero(&pptdev, sizeof(pptdev));
+	pptdev.bus = bus;
+	pptdev.slot = slot;
+	pptdev.func = func;
+
+	return (ioctl(ctx->fd, VM_UNBIND_PPTDEV, &pptdev));
+}
+
+int
+vm_map_pptdev_mmio(struct vmctx *ctx, int bus, int slot, int func,
+		   vm_paddr_t gpa, size_t len, vm_paddr_t hpa)
+{
+	struct vm_pptdev_mmio pptmmio;
+
+	bzero(&pptmmio, sizeof(pptmmio));
+	pptmmio.bus = bus;
+	pptmmio.slot = slot;
+	pptmmio.func = func;
+	pptmmio.gpa = gpa;
+	pptmmio.len = len;
+	pptmmio.hpa = hpa;
+
+	return (ioctl(ctx->fd, VM_MAP_PPTDEV_MMIO, &pptmmio));
+}
+
+int
+vm_setup_msi(struct vmctx *ctx, int vcpu, int bus, int slot, int func,
+	     int destcpu, int vector, int numvec)
+{
+	struct vm_pptdev_msi pptmsi;
+
+	bzero(&pptmsi, sizeof(pptmsi));
+	pptmsi.vcpu = vcpu;
+	pptmsi.bus = bus;
+	pptmsi.slot = slot;
+	pptmsi.func = func;
+	pptmsi.destcpu = destcpu;
+	pptmsi.vector = vector;
+	pptmsi.numvec = numvec;
+
+	return (ioctl(ctx->fd, VM_PPTDEV_MSI, &pptmsi));
+}
+
+uint64_t *
+vm_get_stats(struct vmctx *ctx, int vcpu, struct timeval *ret_tv,
+	     int *ret_entries)
+{
+	int error;
+
+	static struct vm_stats vmstats;
+
+	vmstats.cpuid = vcpu;
+
+	error = ioctl(ctx->fd, VM_STATS, &vmstats);
+	if (error == 0) {
+		if (ret_entries)
+			*ret_entries = vmstats.num_entries;
+		if (ret_tv)
+			*ret_tv = vmstats.tv;
+		return (vmstats.statbuf);
+	} else
+		return (NULL);
+}
+
+const char *
+vm_get_stat_desc(struct vmctx *ctx, int index)
+{
+	int error;
+
+	static struct vm_stat_desc statdesc;
+
+	statdesc.index = index;
+	if (ioctl(ctx->fd, VM_STAT_DESC, &statdesc) == 0)
+		return (statdesc.desc);
+	else
+		return (NULL);
+}
+
+/*
+ * From Intel Vol 3a:
+ * Table 9-1. IA-32 Processor States Following Power-up, Reset or INIT
+ */
+int
+vcpu_reset(struct vmctx *vmctx, int vcpu)
+{
+	int error;
+	uint64_t rflags, rip, cr0, cr4, zero, desc_base, rdx;
+	uint32_t desc_access, desc_limit;
+	uint16_t sel;
+
+	zero = 0;
+
+	rflags = 0x2;
+	error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RFLAGS, rflags);
+	if (error)
+		goto done;
+
+	rip = 0xfff0;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RIP, rip)) != 0)
+		goto done;
+
+	cr0 = CR0_NE;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR0, cr0)) != 0)
+		goto done;
+
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR3, zero)) != 0)
+		goto done;
+	
+	cr4 = CR4_VMXE;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CR4, cr4)) != 0)
+		goto done;
+
+	/*
+	 * CS: present, r/w, accessed, 16-bit, byte granularity, usable
+	 */
+	desc_base = 0xffff0000;
+	desc_limit = 0xffff;
+	desc_access = 0x0093;
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_CS,
+			    desc_base, desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	sel = 0xf000;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_CS, sel)) != 0)
+		goto done;
+
+	/*
+	 * SS,DS,ES,FS,GS: present, r/w, accessed, 16-bit, byte granularity
+	 */
+	desc_base = 0;
+	desc_limit = 0xffff;
+	desc_access = 0x0093;
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_SS,
+			    desc_base, desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_DS,
+			    desc_base, desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_ES,
+			    desc_base, desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_FS,
+			    desc_base, desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_GS,
+			    desc_base, desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	sel = 0;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_SS, sel)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_DS, sel)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_ES, sel)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_FS, sel)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_GS, sel)) != 0)
+		goto done;
+
+	/* General purpose registers */
+	rdx = 0xf00;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RAX, zero)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RBX, zero)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RCX, zero)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RDX, rdx)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RSI, zero)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RDI, zero)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RBP, zero)) != 0)
+		goto done;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_RSP, zero)) != 0)
+		goto done;
+
+	/* GDTR, IDTR */
+	desc_base = 0;
+	desc_limit = 0xffff;
+	desc_access = 0;
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_GDTR,
+			    desc_base, desc_limit, desc_access);
+	if (error != 0)
+		goto done;
+
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_IDTR,
+			    desc_base, desc_limit, desc_access);
+	if (error != 0)
+		goto done;
+
+	/* TR */
+	desc_base = 0;
+	desc_limit = 0xffff;
+	desc_access = 0x0000008b;
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_TR, 0, 0, desc_access);
+	if (error)
+		goto done;
+
+	sel = 0;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_TR, sel)) != 0)
+		goto done;
+
+	/* LDTR */
+	desc_base = 0;
+	desc_limit = 0xffff;
+	desc_access = 0x00000082;
+	error = vm_set_desc(vmctx, vcpu, VM_REG_GUEST_LDTR, desc_base,
+			    desc_limit, desc_access);
+	if (error)
+		goto done;
+
+	sel = 0;
+	if ((error = vm_set_register(vmctx, vcpu, VM_REG_GUEST_LDTR, 0)) != 0)
+		goto done;
+
+	/* XXX cr2, debug registers */
+
+	error = 0;
+done:
+	return (error);
+}