Index: lib/libvmmapi/Makefile =================================================================== --- lib/libvmmapi/Makefile +++ lib/libvmmapi/Makefile @@ -2,13 +2,14 @@ PACKAGE=lib${LIB} LIB= vmmapi -SRCS= vmmapi.c vmmapi_freebsd.c -INCS= vmmapi.h +SRCS= vmmapi.c vmmapi_internal.c +INCS= vmmapi.h vmmapi_internal.h WARNS?= 2 LIBADD= util -CFLAGS+= -I${.CURDIR} +CFLAGS+= -I${.CURDIR} -I${.CURDIR}/${MACHINE} +.include "${MACHINE}/Makefile.inc" .include Index: lib/libvmmapi/amd64/Makefile.inc =================================================================== --- /dev/null +++ lib/libvmmapi/amd64/Makefile.inc @@ -0,0 +1,6 @@ +# $FreeBSD$ + +.PATH: ${.CURDIR}/amd64 + +SRCS+= vmmapi_amd64.c vmmapi_freebsd.c +INCS+= vmmapi_amd64.h Index: lib/libvmmapi/amd64/vmctx_amd64.h =================================================================== --- /dev/null +++ lib/libvmmapi/amd64/vmctx_amd64.h @@ -0,0 +1,43 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * 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. + * + * $FreeBSD$ + */ + +#ifndef _VMMAPI_AMD64_INTERNAL_H_ +#define _VMMAPI_AMD64_INTERNAL_H_ + +struct vmctx { + int fd; + uint32_t lowmem_limit; + int memflags; + size_t lowmem; + size_t highmem; + char *baseaddr; + char *name; +}; +#endif Index: lib/libvmmapi/amd64/vmmapi_amd64.h =================================================================== --- /dev/null +++ lib/libvmmapi/amd64/vmmapi_amd64.h @@ -0,0 +1,133 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * 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. + * + * $FreeBSD$ + */ + +#ifndef _VMMAPI_AMD64_H_ +#define _VMMAPI_AMD64_H_ + +enum x2apic_state; + +/* + * Identifiers for memory segments: + * - vm_setup_memory() uses VM_SYSMEM for the system memory segment. + * - the remaining identifiers can be used to create devmem segments. + */ +enum { + VM_SYSMEM, + VM_BOOTROM, + VM_FRAMEBUFFER, +}; + +/* + * Get the length and name of the memory segment identified by 'segid'. + * Note that system memory segments are identified with a nul name. + * + * Returns 0 on success and non-zero otherwise. + */ +int vm_get_memseg(struct vmctx *ctx, int ident, size_t *lenp, char *name, + size_t namesiz); + +/* + * Iterate over the guest address space. This function finds an address range + * that starts at an address >= *gpa. + * + * Returns 0 if the next address range was found and non-zero otherwise. + */ +int vm_mmap_getnext(struct vmctx *ctx, vm_paddr_t *gpa, int *segid, + vm_ooffset_t *segoff, size_t *len, int *prot, int *flags); +/* + * Create a device memory segment identified by 'segid'. + * + * Returns a pointer to the memory segment on success and MAP_FAILED otherwise. + */ +void *vm_create_devmem(struct vmctx *ctx, int segid, const char *name, + size_t len); + +/* + * Map the memory segment identified by 'segid' into the guest address space + * at [gpa,gpa+len) with protection 'prot'. + */ +int vm_mmap_memseg(struct vmctx *ctx, vm_paddr_t gpa, int segid, + vm_ooffset_t segoff, size_t len, int prot); + +uint32_t vm_get_lowmem_limit(struct vmctx *ctx); +void vm_set_lowmem_limit(struct vmctx *ctx, uint32_t limit); +size_t vm_get_lowmem_size(struct vmctx *ctx); +size_t vm_get_highmem_size(struct vmctx *ctx); +int vm_set_desc(struct vmctx *ctx, int vcpu, int reg, + uint64_t base, uint32_t limit, uint32_t access); +int vm_get_desc(struct vmctx *ctx, int vcpu, int reg, + uint64_t *base, uint32_t *limit, uint32_t *access); +int vm_get_seg_desc(struct vmctx *ctx, int vcpu, int reg, + struct seg_desc *seg_desc); +int vm_apicid2vcpu(struct vmctx *ctx, int apicid); +int vm_inject_exception(struct vmctx *ctx, int vcpu, int vector, + int errcode_valid, uint32_t errcode, int restart_instruction); +int vm_lapic_irq(struct vmctx *ctx, int vcpu, int vector); +int vm_lapic_local_irq(struct vmctx *ctx, int vcpu, int vector); +int vm_lapic_msi(struct vmctx *ctx, uint64_t addr, uint64_t msg); +int vm_ioapic_assert_irq(struct vmctx *ctx, int irq); +int vm_ioapic_deassert_irq(struct vmctx *ctx, int irq); +int vm_ioapic_pulse_irq(struct vmctx *ctx, int irq); +int vm_ioapic_pincount(struct vmctx *ctx, int *pincount); +int vm_isa_assert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq); +int vm_isa_deassert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq); +int vm_isa_pulse_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq); +int vm_isa_set_irq_trigger(struct vmctx *ctx, int atpic_irq, + enum vm_intr_trigger trigger); +int vm_inject_nmi(struct vmctx *ctx, int vcpu); + +int vm_get_intinfo(struct vmctx *ctx, int vcpu, uint64_t *i1, uint64_t *i2); +int vm_set_intinfo(struct vmctx *ctx, int vcpu, uint64_t exit_intinfo); + +int vm_get_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state *s); +int vm_set_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state s); + +int vm_get_hpet_capabilities(struct vmctx *ctx, uint32_t *capabilities); + +/* RTC */ +int vm_rtc_write(struct vmctx *ctx, int offset, uint8_t value); +int vm_rtc_read(struct vmctx *ctx, int offset, uint8_t *retval); +int vm_rtc_settime(struct vmctx *ctx, time_t secs); +int vm_rtc_gettime(struct vmctx *ctx, time_t *secs); + +/* Reset vcpu register state */ +int vcpu_reset(struct vmctx *ctx, int vcpu); + +/* + * FreeBSD specific APIs + */ +int vm_setup_freebsd_registers(struct vmctx *ctx, int vcpu, + uint64_t rip, uint64_t cr3, uint64_t gdtbase, + uint64_t rsp); +int vm_setup_freebsd_registers_i386(struct vmctx *vmctx, int vcpu, + uint32_t eip, uint32_t gdtbase, + uint32_t esp); +void vm_setup_freebsd_gdt(uint64_t *gdtr); +#endif /* _VMMAPI_AMD64_H_ */ Index: lib/libvmmapi/amd64/vmmapi_amd64.c =================================================================== --- /dev/null +++ lib/libvmmapi/amd64/vmmapi_amd64.c @@ -0,0 +1,1015 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * 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 +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include +#include + +#include +#include + +#include +#include +#include +#include +#include +#include +#include + +#include + +#include +#include + +#include + +#include "vmmapi_internal.h" +#include "vmctx_amd64.h" + +#define MB (1024 * 1024UL) +#define GB (1024 * 1024 * 1024UL) + +/* + * Size of the guard region before and after the virtual address space + * mapping the guest physical memory. This must be a multiple of the + * superpage size for performance reasons. + */ +#define VM_MMAP_GUARD_SIZE (4 * MB) + +#define PROT_RW (PROT_READ | PROT_WRITE) +#define PROT_ALL (PROT_READ | PROT_WRITE | PROT_EXEC) + +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->memflags = 0; + vm->lowmem_limit = 3 * GB; + 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); +} + +uint32_t +vm_get_lowmem_limit(struct vmctx *ctx) +{ + + return (ctx->lowmem_limit); +} + +void +vm_set_lowmem_limit(struct vmctx *ctx, uint32_t limit) +{ + + ctx->lowmem_limit = limit; +} + +void +vm_set_memflags(struct vmctx *ctx, int flags) +{ + + ctx->memflags = flags; +} + +int +vm_get_memflags(struct vmctx *ctx) +{ + + return (ctx->memflags); +} + +/* + * Map segment 'segid' starting at 'off' into guest address range [gpa,gpa+len). + */ +int +vm_mmap_memseg(struct vmctx *ctx, vm_paddr_t gpa, int segid, vm_ooffset_t off, + size_t len, int prot) +{ + struct vm_memmap memmap; + int error, flags; + + memmap.gpa = gpa; + memmap.segid = segid; + memmap.segoff = off; + memmap.len = len; + memmap.prot = prot; + memmap.flags = 0; + + if (ctx->memflags & VM_MEM_F_WIRED) + memmap.flags |= VM_MEMMAP_F_WIRED; + + /* + * If this mapping already exists then don't create it again. This + * is the common case for SYSMEM mappings created by bhyveload(8). + */ + error = vm_mmap_getnext(ctx, &gpa, &segid, &off, &len, &prot, &flags); + if (error == 0 && gpa == memmap.gpa) { + if (segid != memmap.segid || off != memmap.segoff || + prot != memmap.prot || flags != memmap.flags) { + errno = EEXIST; + return (-1); + } else { + return (0); + } + } + + error = ioctl(ctx->fd, VM_MMAP_MEMSEG, &memmap); + return (error); +} + +int +vm_mmap_getnext(struct vmctx *ctx, vm_paddr_t *gpa, int *segid, + vm_ooffset_t *segoff, size_t *len, int *prot, int *flags) +{ + struct vm_memmap memmap; + int error; + + bzero(&memmap, sizeof(struct vm_memmap)); + memmap.gpa = *gpa; + error = ioctl(ctx->fd, VM_MMAP_GETNEXT, &memmap); + if (error == 0) { + *gpa = memmap.gpa; + *segid = memmap.segid; + *segoff = memmap.segoff; + *len = memmap.len; + *prot = memmap.prot; + *flags = memmap.flags; + } + return (error); +} + +/* + * Return 0 if the segments are identical and non-zero otherwise. + * + * This is slightly complicated by the fact that only device memory segments + * are named. + */ +static int +cmpseg(size_t len, const char *str, size_t len2, const char *str2) +{ + + if (len == len2) { + if ((!str && !str2) || (str && str2 && !strcmp(str, str2))) + return (0); + } + return (-1); +} + +static int +vm_alloc_memseg(struct vmctx *ctx, int segid, size_t len, const char *name) +{ + struct vm_memseg memseg; + size_t n; + int error; + + /* + * If the memory segment has already been created then just return. + * This is the usual case for the SYSMEM segment created by userspace + * loaders like bhyveload(8). + */ + error = vm_get_memseg(ctx, segid, &memseg.len, memseg.name, + sizeof(memseg.name)); + if (error) + return (error); + + if (memseg.len != 0) { + if (cmpseg(len, name, memseg.len, VM_MEMSEG_NAME(&memseg))) { + errno = EINVAL; + return (-1); + } else { + return (0); + } + } + + bzero(&memseg, sizeof(struct vm_memseg)); + memseg.segid = segid; + memseg.len = len; + if (name != NULL) { + n = strlcpy(memseg.name, name, sizeof(memseg.name)); + if (n >= sizeof(memseg.name)) { + errno = ENAMETOOLONG; + return (-1); + } + } + + error = ioctl(ctx->fd, VM_ALLOC_MEMSEG, &memseg); + return (error); +} + +int +vm_get_memseg(struct vmctx *ctx, int segid, size_t *lenp, char *namebuf, + size_t bufsize) +{ + struct vm_memseg memseg; + size_t n; + int error; + + memseg.segid = segid; + error = ioctl(ctx->fd, VM_GET_MEMSEG, &memseg); + if (error == 0) { + *lenp = memseg.len; + n = strlcpy(namebuf, memseg.name, bufsize); + if (n >= bufsize) { + errno = ENAMETOOLONG; + error = -1; + } + } + return (error); +} + +static int +setup_memory_segment(struct vmctx *ctx, vm_paddr_t gpa, size_t len, char *base) +{ + char *ptr; + int error, flags; + + /* Map 'len' bytes starting at 'gpa' in the guest address space */ + error = vm_mmap_memseg(ctx, gpa, VM_SYSMEM, gpa, len, PROT_ALL); + if (error) + return (error); + + flags = MAP_SHARED | MAP_FIXED; + if ((ctx->memflags & VM_MEM_F_INCORE) == 0) + flags |= MAP_NOCORE; + + /* mmap into the process address space on the host */ + ptr = mmap(base + gpa, len, PROT_RW, flags, ctx->fd, gpa); + if (ptr == MAP_FAILED) + return (-1); + + return (0); +} + +int +vm_setup_memory(struct vmctx *ctx, size_t memsize, enum vm_mmap_style vms) +{ + size_t objsize, len; + vm_paddr_t gpa; + char *baseaddr, *ptr; + int error; + + assert(vms == VM_MMAP_ALL); + + /* + * If 'memsize' cannot fit entirely in the 'lowmem' segment then + * create another 'highmem' segment above 4GB for the remainder. + */ + if (memsize > ctx->lowmem_limit) { + ctx->lowmem = ctx->lowmem_limit; + ctx->highmem = memsize - ctx->lowmem_limit; + objsize = 4*GB + ctx->highmem; + } else { + ctx->lowmem = memsize; + ctx->highmem = 0; + objsize = ctx->lowmem; + } + + error = vm_alloc_memseg(ctx, VM_SYSMEM, objsize, NULL); + if (error) + return (error); + + /* + * Stake out a contiguous region covering the guest physical memory + * and the adjoining guard regions. + */ + len = VM_MMAP_GUARD_SIZE + objsize + VM_MMAP_GUARD_SIZE; + ptr = mmap(NULL, len, PROT_NONE, MAP_GUARD | MAP_ALIGNED_SUPER, -1, 0); + if (ptr == MAP_FAILED) + return (-1); + + baseaddr = ptr + VM_MMAP_GUARD_SIZE; + if (ctx->highmem > 0) { + gpa = 4*GB; + len = ctx->highmem; + error = setup_memory_segment(ctx, gpa, len, baseaddr); + if (error) + return (error); + } + + if (ctx->lowmem > 0) { + gpa = 0; + len = ctx->lowmem; + error = setup_memory_segment(ctx, gpa, len, baseaddr); + if (error) + return (error); + } + + ctx->baseaddr = baseaddr; + + return (0); +} + +/* + * Returns a non-NULL pointer if [gaddr, gaddr+len) is entirely contained in + * the lowmem or highmem regions. + * + * In particular return NULL if [gaddr, gaddr+len) falls in guest MMIO region. + * The instruction emulation code depends on this behavior. + */ +void * +vm_map_gpa(struct vmctx *ctx, vm_paddr_t gaddr, size_t len) +{ + + if (ctx->lowmem > 0) { + if (gaddr < ctx->lowmem && len <= ctx->lowmem && + gaddr + len <= ctx->lowmem) + return (ctx->baseaddr + gaddr); + } + + if (ctx->highmem > 0) { + if (gaddr >= 4*GB) { + if (gaddr < 4*GB + ctx->highmem && + len <= ctx->highmem && + gaddr + len <= 4*GB + ctx->highmem) + return (ctx->baseaddr + gaddr); + } + } + + return (NULL); +} + +size_t +vm_get_lowmem_size(struct vmctx *ctx) +{ + + return (ctx->lowmem); +} + +size_t +vm_get_highmem_size(struct vmctx *ctx) +{ + + return (ctx->highmem); +} + +void * +vm_create_devmem(struct vmctx *ctx, int segid, const char *name, size_t len) +{ + char pathname[MAXPATHLEN]; + size_t len2; + char *base, *ptr; + int fd, error, flags; + + fd = -1; + ptr = MAP_FAILED; + if (name == NULL || strlen(name) == 0) { + errno = EINVAL; + goto done; + } + + error = vm_alloc_memseg(ctx, segid, len, name); + if (error) + goto done; + + strlcpy(pathname, "/dev/vmm.io/", sizeof(pathname)); + strlcat(pathname, ctx->name, sizeof(pathname)); + strlcat(pathname, ".", sizeof(pathname)); + strlcat(pathname, name, sizeof(pathname)); + + fd = open(pathname, O_RDWR); + if (fd < 0) + goto done; + + /* + * Stake out a contiguous region covering the device memory and the + * adjoining guard regions. + */ + len2 = VM_MMAP_GUARD_SIZE + len + VM_MMAP_GUARD_SIZE; + base = mmap(NULL, len2, PROT_NONE, MAP_GUARD | MAP_ALIGNED_SUPER, -1, + 0); + if (base == MAP_FAILED) + goto done; + + flags = MAP_SHARED | MAP_FIXED; + if ((ctx->memflags & VM_MEM_F_INCORE) == 0) + flags |= MAP_NOCORE; + + /* mmap the devmem region in the host address space */ + ptr = mmap(base + VM_MMAP_GUARD_SIZE, len, PROT_RW, flags, fd, 0); +done: + if (fd >= 0) + close(fd); + return (ptr); +} + +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_get_seg_desc(struct vmctx *ctx, int vcpu, int reg, struct seg_desc *seg_desc) +{ + int error; + + error = vm_get_desc(ctx, vcpu, reg, &seg_desc->base, &seg_desc->limit, + &seg_desc->access); + return (error); +} + +int +vm_inject_exception(struct vmctx *ctx, int vcpu, int vector, int errcode_valid, + uint32_t errcode, int restart_instruction) +{ + struct vm_exception exc; + + exc.cpuid = vcpu; + exc.vector = vector; + exc.error_code = errcode; + exc.error_code_valid = errcode_valid; + exc.restart_instruction = restart_instruction; + + return (ioctl(ctx->fd, VM_INJECT_EXCEPTION, &exc)); +} + +int +vm_apicid2vcpu(struct vmctx *ctx, int apicid) +{ + /* + * The apic id associated with the 'vcpu' has the same numerical value + * as the 'vcpu' itself. + */ + return (apicid); +} + +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_lapic_local_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_LOCAL_IRQ, &vmirq)); +} + +int +vm_lapic_msi(struct vmctx *ctx, uint64_t addr, uint64_t msg) +{ + struct vm_lapic_msi vmmsi; + + bzero(&vmmsi, sizeof(vmmsi)); + vmmsi.addr = addr; + vmmsi.msg = msg; + + return (ioctl(ctx->fd, VM_LAPIC_MSI, &vmmsi)); +} + +int +vm_ioapic_assert_irq(struct vmctx *ctx, int irq) +{ + struct vm_ioapic_irq ioapic_irq; + + bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq)); + ioapic_irq.irq = irq; + + return (ioctl(ctx->fd, VM_IOAPIC_ASSERT_IRQ, &ioapic_irq)); +} + +int +vm_ioapic_deassert_irq(struct vmctx *ctx, int irq) +{ + struct vm_ioapic_irq ioapic_irq; + + bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq)); + ioapic_irq.irq = irq; + + return (ioctl(ctx->fd, VM_IOAPIC_DEASSERT_IRQ, &ioapic_irq)); +} + +int +vm_ioapic_pulse_irq(struct vmctx *ctx, int irq) +{ + struct vm_ioapic_irq ioapic_irq; + + bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq)); + ioapic_irq.irq = irq; + + return (ioctl(ctx->fd, VM_IOAPIC_PULSE_IRQ, &ioapic_irq)); +} + +int +vm_ioapic_pincount(struct vmctx *ctx, int *pincount) +{ + + return (ioctl(ctx->fd, VM_IOAPIC_PINCOUNT, pincount)); +} + +int +vm_isa_assert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq) +{ + struct vm_isa_irq isa_irq; + + bzero(&isa_irq, sizeof(struct vm_isa_irq)); + isa_irq.atpic_irq = atpic_irq; + isa_irq.ioapic_irq = ioapic_irq; + + return (ioctl(ctx->fd, VM_ISA_ASSERT_IRQ, &isa_irq)); +} + +int +vm_isa_deassert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq) +{ + struct vm_isa_irq isa_irq; + + bzero(&isa_irq, sizeof(struct vm_isa_irq)); + isa_irq.atpic_irq = atpic_irq; + isa_irq.ioapic_irq = ioapic_irq; + + return (ioctl(ctx->fd, VM_ISA_DEASSERT_IRQ, &isa_irq)); +} + +int +vm_isa_pulse_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq) +{ + struct vm_isa_irq isa_irq; + + bzero(&isa_irq, sizeof(struct vm_isa_irq)); + isa_irq.atpic_irq = atpic_irq; + isa_irq.ioapic_irq = ioapic_irq; + + return (ioctl(ctx->fd, VM_ISA_PULSE_IRQ, &isa_irq)); +} + +int +vm_isa_set_irq_trigger(struct vmctx *ctx, int atpic_irq, + enum vm_intr_trigger trigger) +{ + struct vm_isa_irq_trigger isa_irq_trigger; + + bzero(&isa_irq_trigger, sizeof(struct vm_isa_irq_trigger)); + isa_irq_trigger.atpic_irq = atpic_irq; + isa_irq_trigger.trigger = trigger; + + return (ioctl(ctx->fd, VM_ISA_SET_IRQ_TRIGGER, &isa_irq_trigger)); +} + +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)); +} + +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 }, + { "enable_invpcid", VM_CAP_ENABLE_INVPCID }, + { 0 } +}; + +int +vm_capability_name2type(const char *capname) +{ + int i; + + for (i = 0; capstrmap[i].name != NULL && capname != NULL; i++) { + if (strcmp(capstrmap[i].name, capname) == 0) + return (capstrmap[i].type); + } + + return (-1); +} + +const char * +vm_capability_type2name(int type) +{ + int i; + + for (i = 0; capstrmap[i].name != NULL; i++) { + if (capstrmap[i].type == type) + return (capstrmap[i].name); + } + + return (NULL); +} + +int +vm_get_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state *state) +{ + int error; + struct vm_x2apic x2apic; + + bzero(&x2apic, sizeof(x2apic)); + x2apic.cpuid = vcpu; + + error = ioctl(ctx->fd, VM_GET_X2APIC_STATE, &x2apic); + *state = x2apic.state; + return (error); +} + +int +vm_set_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state state) +{ + int error; + struct vm_x2apic x2apic; + + bzero(&x2apic, sizeof(x2apic)); + x2apic.cpuid = vcpu; + x2apic.state = state; + + error = ioctl(ctx->fd, VM_SET_X2APIC_STATE, &x2apic); + + return (error); +} + +/* + * 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 = 0; + 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); +} + +int +vm_get_hpet_capabilities(struct vmctx *ctx, uint32_t *capabilities) +{ + int error; + struct vm_hpet_cap cap; + + bzero(&cap, sizeof(struct vm_hpet_cap)); + error = ioctl(ctx->fd, VM_GET_HPET_CAPABILITIES, &cap); + if (capabilities != NULL) + *capabilities = cap.capabilities; + return (error); +} + +int +vm_get_intinfo(struct vmctx *ctx, int vcpu, uint64_t *info1, uint64_t *info2) +{ + struct vm_intinfo vmii; + int error; + + bzero(&vmii, sizeof(struct vm_intinfo)); + vmii.vcpuid = vcpu; + error = ioctl(ctx->fd, VM_GET_INTINFO, &vmii); + if (error == 0) { + *info1 = vmii.info1; + *info2 = vmii.info2; + } + return (error); +} + +int +vm_set_intinfo(struct vmctx *ctx, int vcpu, uint64_t info1) +{ + struct vm_intinfo vmii; + int error; + + bzero(&vmii, sizeof(struct vm_intinfo)); + vmii.vcpuid = vcpu; + vmii.info1 = info1; + error = ioctl(ctx->fd, VM_SET_INTINFO, &vmii); + return (error); +} + +int +vm_rtc_write(struct vmctx *ctx, int offset, uint8_t value) +{ + struct vm_rtc_data rtcdata; + int error; + + bzero(&rtcdata, sizeof(struct vm_rtc_data)); + rtcdata.offset = offset; + rtcdata.value = value; + error = ioctl(ctx->fd, VM_RTC_WRITE, &rtcdata); + return (error); +} + +int +vm_rtc_read(struct vmctx *ctx, int offset, uint8_t *retval) +{ + struct vm_rtc_data rtcdata; + int error; + + bzero(&rtcdata, sizeof(struct vm_rtc_data)); + rtcdata.offset = offset; + error = ioctl(ctx->fd, VM_RTC_READ, &rtcdata); + if (error == 0) + *retval = rtcdata.value; + return (error); +} + +int +vm_rtc_settime(struct vmctx *ctx, time_t secs) +{ + struct vm_rtc_time rtctime; + int error; + + bzero(&rtctime, sizeof(struct vm_rtc_time)); + rtctime.secs = secs; + error = ioctl(ctx->fd, VM_RTC_SETTIME, &rtctime); + return (error); +} + +int +vm_rtc_gettime(struct vmctx *ctx, time_t *secs) +{ + struct vm_rtc_time rtctime; + int error; + + bzero(&rtctime, sizeof(struct vm_rtc_time)); + error = ioctl(ctx->fd, VM_RTC_GETTIME, &rtctime); + if (error == 0) + *secs = rtctime.secs; + return (error); +} + +int +vm_get_device_fd(struct vmctx *ctx) +{ + + return (ctx->fd); +} + +const char * +vm_get_name(struct vmctx *ctx) +{ + return (ctx->name); +} + +const cap_ioctl_t * +vm_get_ioctls(size_t *len) +{ + cap_ioctl_t *cmds; + /* keep in sync with machine/vmm_dev.h */ + static const cap_ioctl_t vm_ioctl_cmds[] = { VM_RUN, VM_SUSPEND, VM_REINIT, + VM_ALLOC_MEMSEG, VM_GET_MEMSEG, VM_MMAP_MEMSEG, VM_MMAP_MEMSEG, + VM_MMAP_GETNEXT, VM_SET_REGISTER, VM_GET_REGISTER, + VM_SET_SEGMENT_DESCRIPTOR, VM_GET_SEGMENT_DESCRIPTOR, + VM_SET_REGISTER_SET, VM_GET_REGISTER_SET, + VM_INJECT_EXCEPTION, VM_LAPIC_IRQ, VM_LAPIC_LOCAL_IRQ, + VM_LAPIC_MSI, VM_IOAPIC_ASSERT_IRQ, VM_IOAPIC_DEASSERT_IRQ, + VM_IOAPIC_PULSE_IRQ, VM_IOAPIC_PINCOUNT, VM_ISA_ASSERT_IRQ, + VM_ISA_DEASSERT_IRQ, VM_ISA_PULSE_IRQ, VM_ISA_SET_IRQ_TRIGGER, + VM_SET_CAPABILITY, VM_GET_CAPABILITY, VM_BIND_PPTDEV, + VM_UNBIND_PPTDEV, VM_MAP_PPTDEV_MMIO, VM_PPTDEV_MSI, + VM_PPTDEV_MSIX, VM_INJECT_NMI, VM_STATS, VM_STAT_DESC, + VM_SET_X2APIC_STATE, VM_GET_X2APIC_STATE, + VM_GET_HPET_CAPABILITIES, VM_GET_GPA_PMAP, VM_GLA2GPA, + VM_GLA2GPA_NOFAULT, + VM_ACTIVATE_CPU, VM_GET_CPUS, VM_SUSPEND_CPU, VM_RESUME_CPU, + VM_SET_INTINFO, VM_GET_INTINFO, + VM_RTC_WRITE, VM_RTC_READ, VM_RTC_SETTIME, VM_RTC_GETTIME, + VM_RESTART_INSTRUCTION, VM_SET_TOPOLOGY, VM_GET_TOPOLOGY }; + + if (len == NULL) { + cmds = malloc(sizeof(vm_ioctl_cmds)); + if (cmds == NULL) + return (NULL); + bcopy(vm_ioctl_cmds, cmds, sizeof(vm_ioctl_cmds)); + return (cmds); + } + + *len = nitems(vm_ioctl_cmds); + return (NULL); +} Index: lib/libvmmapi/vmmapi.h =================================================================== --- lib/libvmmapi/vmmapi.h +++ lib/libvmmapi/vmmapi.h @@ -34,6 +34,12 @@ #include #include +struct vmctx; + +#if defined(__amd64__) +#include +#endif + /* * API version for out-of-tree consumers like grub-bhyve for making compile * time decisions. @@ -41,8 +47,6 @@ #define VMMAPI_VERSION 0103 /* 2 digit major followed by 2 digit minor */ struct iovec; -struct vmctx; -enum x2apic_state; /* * Different styles of mapping the memory assigned to a VM into the address @@ -60,49 +64,6 @@ #define VM_MEM_F_INCORE 0x01 /* include guest memory in core file */ #define VM_MEM_F_WIRED 0x02 /* guest memory is wired */ -/* - * Identifiers for memory segments: - * - vm_setup_memory() uses VM_SYSMEM for the system memory segment. - * - the remaining identifiers can be used to create devmem segments. - */ -enum { - VM_SYSMEM, - VM_BOOTROM, - VM_FRAMEBUFFER, -}; - -/* - * Get the length and name of the memory segment identified by 'segid'. - * Note that system memory segments are identified with a nul name. - * - * Returns 0 on success and non-zero otherwise. - */ -int vm_get_memseg(struct vmctx *ctx, int ident, size_t *lenp, char *name, - size_t namesiz); - -/* - * Iterate over the guest address space. This function finds an address range - * that starts at an address >= *gpa. - * - * Returns 0 if the next address range was found and non-zero otherwise. - */ -int vm_mmap_getnext(struct vmctx *ctx, vm_paddr_t *gpa, int *segid, - vm_ooffset_t *segoff, size_t *len, int *prot, int *flags); -/* - * Create a device memory segment identified by 'segid'. - * - * Returns a pointer to the memory segment on success and MAP_FAILED otherwise. - */ -void *vm_create_devmem(struct vmctx *ctx, int segid, const char *name, - size_t len); - -/* - * Map the memory segment identified by 'segid' into the guest address space - * at [gpa,gpa+len) with protection 'prot'. - */ -int vm_mmap_memseg(struct vmctx *ctx, vm_paddr_t gpa, int segid, - vm_ooffset_t segoff, size_t len, int prot); - int vm_create(const char *name); int vm_get_device_fd(struct vmctx *ctx); struct vmctx *vm_open(const char *name); @@ -116,18 +77,8 @@ int vm_gla2gpa_nofault(struct vmctx *, int vcpuid, struct vm_guest_paging *paging, uint64_t gla, int prot, uint64_t *gpa, int *fault); -uint32_t vm_get_lowmem_limit(struct vmctx *ctx); -void vm_set_lowmem_limit(struct vmctx *ctx, uint32_t limit); void vm_set_memflags(struct vmctx *ctx, int flags); int vm_get_memflags(struct vmctx *ctx); -size_t vm_get_lowmem_size(struct vmctx *ctx); -size_t vm_get_highmem_size(struct vmctx *ctx); -int vm_set_desc(struct vmctx *ctx, int vcpu, int reg, - uint64_t base, uint32_t limit, uint32_t access); -int vm_get_desc(struct vmctx *ctx, int vcpu, int reg, - uint64_t *base, uint32_t *limit, uint32_t *access); -int vm_get_seg_desc(struct vmctx *ctx, int vcpu, int reg, - struct seg_desc *seg_desc); int vm_set_register(struct vmctx *ctx, int vcpu, int reg, uint64_t val); int vm_get_register(struct vmctx *ctx, int vcpu, int reg, uint64_t *retval); int vm_set_register_set(struct vmctx *ctx, int vcpu, unsigned int count, @@ -137,22 +88,6 @@ int vm_run(struct vmctx *ctx, int vcpu, struct vm_exit *ret_vmexit); int vm_suspend(struct vmctx *ctx, enum vm_suspend_how how); int vm_reinit(struct vmctx *ctx); -int vm_apicid2vcpu(struct vmctx *ctx, int apicid); -int vm_inject_exception(struct vmctx *ctx, int vcpu, int vector, - int errcode_valid, uint32_t errcode, int restart_instruction); -int vm_lapic_irq(struct vmctx *ctx, int vcpu, int vector); -int vm_lapic_local_irq(struct vmctx *ctx, int vcpu, int vector); -int vm_lapic_msi(struct vmctx *ctx, uint64_t addr, uint64_t msg); -int vm_ioapic_assert_irq(struct vmctx *ctx, int irq); -int vm_ioapic_deassert_irq(struct vmctx *ctx, int irq); -int vm_ioapic_pulse_irq(struct vmctx *ctx, int irq); -int vm_ioapic_pincount(struct vmctx *ctx, int *pincount); -int vm_isa_assert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq); -int vm_isa_deassert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq); -int vm_isa_pulse_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq); -int vm_isa_set_irq_trigger(struct vmctx *ctx, int atpic_irq, - enum vm_intr_trigger trigger); -int vm_inject_nmi(struct vmctx *ctx, int vcpu); int vm_capability_name2type(const char *capname); const char *vm_capability_type2name(int type); int vm_get_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap, @@ -169,9 +104,6 @@ int func, int idx, uint64_t addr, uint64_t msg, uint32_t vector_control); -int vm_get_intinfo(struct vmctx *ctx, int vcpu, uint64_t *i1, uint64_t *i2); -int vm_set_intinfo(struct vmctx *ctx, int vcpu, uint64_t exit_intinfo); - const cap_ioctl_t *vm_get_ioctls(size_t *len); /* @@ -181,11 +113,6 @@ int *ret_entries); const char *vm_get_stat_desc(struct vmctx *ctx, int index); -int vm_get_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state *s); -int vm_set_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state s); - -int vm_get_hpet_capabilities(struct vmctx *ctx, uint32_t *capabilities); - /* * Translate the GLA range [gla,gla+len) into GPA segments in 'iov'. * The 'iovcnt' should be big enough to accommodate all GPA segments. @@ -205,15 +132,6 @@ void vm_copy_teardown(struct vmctx *ctx, int vcpu, struct iovec *iov, int iovcnt); -/* RTC */ -int vm_rtc_write(struct vmctx *ctx, int offset, uint8_t value); -int vm_rtc_read(struct vmctx *ctx, int offset, uint8_t *retval); -int vm_rtc_settime(struct vmctx *ctx, time_t secs); -int vm_rtc_gettime(struct vmctx *ctx, time_t *secs); - -/* Reset vcpu register state */ -int vcpu_reset(struct vmctx *ctx, int vcpu); - int vm_active_cpus(struct vmctx *ctx, cpuset_t *cpus); int vm_suspended_cpus(struct vmctx *ctx, cpuset_t *cpus); int vm_debug_cpus(struct vmctx *ctx, cpuset_t *cpus); @@ -226,15 +144,4 @@ uint16_t threads, uint16_t maxcpus); int vm_get_topology(struct vmctx *ctx, uint16_t *sockets, uint16_t *cores, uint16_t *threads, uint16_t *maxcpus); - -/* - * FreeBSD specific APIs - */ -int vm_setup_freebsd_registers(struct vmctx *ctx, int vcpu, - uint64_t rip, uint64_t cr3, uint64_t gdtbase, - uint64_t rsp); -int vm_setup_freebsd_registers_i386(struct vmctx *vmctx, int vcpu, - uint32_t eip, uint32_t gdtbase, - uint32_t esp); -void vm_setup_freebsd_gdt(uint64_t *gdtr); #endif /* _VMMAPI_H_ */ Index: lib/libvmmapi/vmmapi.c =================================================================== --- lib/libvmmapi/vmmapi.c +++ lib/libvmmapi/vmmapi.c @@ -38,9 +38,6 @@ #include #include -#include -#include - #include #include #include @@ -56,50 +53,15 @@ #include "vmmapi.h" -#define MB (1024 * 1024UL) -#define GB (1024 * 1024 * 1024UL) - -/* - * Size of the guard region before and after the virtual address space - * mapping the guest physical memory. This must be a multiple of the - * superpage size for performance reasons. - */ -#define VM_MMAP_GUARD_SIZE (4 * MB) - -#define PROT_RW (PROT_READ | PROT_WRITE) -#define PROT_ALL (PROT_READ | PROT_WRITE | PROT_EXEC) +#if defined(__amd64__) +#include "amd64/vmctx_amd64.h" +#endif -struct vmctx { - int fd; - uint32_t lowmem_limit; - int memflags; - size_t lowmem; - size_t highmem; - char *baseaddr; - char *name; -}; +#define MB (1024 * 1024UL) #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) { @@ -107,29 +69,6 @@ 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->memflags = 0; - vm->lowmem_limit = 3 * GB; - 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) { @@ -166,396 +105,6 @@ return (error); } -uint32_t -vm_get_lowmem_limit(struct vmctx *ctx) -{ - - return (ctx->lowmem_limit); -} - -void -vm_set_lowmem_limit(struct vmctx *ctx, uint32_t limit) -{ - - ctx->lowmem_limit = limit; -} - -void -vm_set_memflags(struct vmctx *ctx, int flags) -{ - - ctx->memflags = flags; -} - -int -vm_get_memflags(struct vmctx *ctx) -{ - - return (ctx->memflags); -} - -/* - * Map segment 'segid' starting at 'off' into guest address range [gpa,gpa+len). - */ -int -vm_mmap_memseg(struct vmctx *ctx, vm_paddr_t gpa, int segid, vm_ooffset_t off, - size_t len, int prot) -{ - struct vm_memmap memmap; - int error, flags; - - memmap.gpa = gpa; - memmap.segid = segid; - memmap.segoff = off; - memmap.len = len; - memmap.prot = prot; - memmap.flags = 0; - - if (ctx->memflags & VM_MEM_F_WIRED) - memmap.flags |= VM_MEMMAP_F_WIRED; - - /* - * If this mapping already exists then don't create it again. This - * is the common case for SYSMEM mappings created by bhyveload(8). - */ - error = vm_mmap_getnext(ctx, &gpa, &segid, &off, &len, &prot, &flags); - if (error == 0 && gpa == memmap.gpa) { - if (segid != memmap.segid || off != memmap.segoff || - prot != memmap.prot || flags != memmap.flags) { - errno = EEXIST; - return (-1); - } else { - return (0); - } - } - - error = ioctl(ctx->fd, VM_MMAP_MEMSEG, &memmap); - return (error); -} - -int -vm_mmap_getnext(struct vmctx *ctx, vm_paddr_t *gpa, int *segid, - vm_ooffset_t *segoff, size_t *len, int *prot, int *flags) -{ - struct vm_memmap memmap; - int error; - - bzero(&memmap, sizeof(struct vm_memmap)); - memmap.gpa = *gpa; - error = ioctl(ctx->fd, VM_MMAP_GETNEXT, &memmap); - if (error == 0) { - *gpa = memmap.gpa; - *segid = memmap.segid; - *segoff = memmap.segoff; - *len = memmap.len; - *prot = memmap.prot; - *flags = memmap.flags; - } - return (error); -} - -/* - * Return 0 if the segments are identical and non-zero otherwise. - * - * This is slightly complicated by the fact that only device memory segments - * are named. - */ -static int -cmpseg(size_t len, const char *str, size_t len2, const char *str2) -{ - - if (len == len2) { - if ((!str && !str2) || (str && str2 && !strcmp(str, str2))) - return (0); - } - return (-1); -} - -static int -vm_alloc_memseg(struct vmctx *ctx, int segid, size_t len, const char *name) -{ - struct vm_memseg memseg; - size_t n; - int error; - - /* - * If the memory segment has already been created then just return. - * This is the usual case for the SYSMEM segment created by userspace - * loaders like bhyveload(8). - */ - error = vm_get_memseg(ctx, segid, &memseg.len, memseg.name, - sizeof(memseg.name)); - if (error) - return (error); - - if (memseg.len != 0) { - if (cmpseg(len, name, memseg.len, VM_MEMSEG_NAME(&memseg))) { - errno = EINVAL; - return (-1); - } else { - return (0); - } - } - - bzero(&memseg, sizeof(struct vm_memseg)); - memseg.segid = segid; - memseg.len = len; - if (name != NULL) { - n = strlcpy(memseg.name, name, sizeof(memseg.name)); - if (n >= sizeof(memseg.name)) { - errno = ENAMETOOLONG; - return (-1); - } - } - - error = ioctl(ctx->fd, VM_ALLOC_MEMSEG, &memseg); - return (error); -} - -int -vm_get_memseg(struct vmctx *ctx, int segid, size_t *lenp, char *namebuf, - size_t bufsize) -{ - struct vm_memseg memseg; - size_t n; - int error; - - memseg.segid = segid; - error = ioctl(ctx->fd, VM_GET_MEMSEG, &memseg); - if (error == 0) { - *lenp = memseg.len; - n = strlcpy(namebuf, memseg.name, bufsize); - if (n >= bufsize) { - errno = ENAMETOOLONG; - error = -1; - } - } - return (error); -} - -static int -setup_memory_segment(struct vmctx *ctx, vm_paddr_t gpa, size_t len, char *base) -{ - char *ptr; - int error, flags; - - /* Map 'len' bytes starting at 'gpa' in the guest address space */ - error = vm_mmap_memseg(ctx, gpa, VM_SYSMEM, gpa, len, PROT_ALL); - if (error) - return (error); - - flags = MAP_SHARED | MAP_FIXED; - if ((ctx->memflags & VM_MEM_F_INCORE) == 0) - flags |= MAP_NOCORE; - - /* mmap into the process address space on the host */ - ptr = mmap(base + gpa, len, PROT_RW, flags, ctx->fd, gpa); - if (ptr == MAP_FAILED) - return (-1); - - return (0); -} - -int -vm_setup_memory(struct vmctx *ctx, size_t memsize, enum vm_mmap_style vms) -{ - size_t objsize, len; - vm_paddr_t gpa; - char *baseaddr, *ptr; - int error; - - assert(vms == VM_MMAP_ALL); - - /* - * If 'memsize' cannot fit entirely in the 'lowmem' segment then - * create another 'highmem' segment above 4GB for the remainder. - */ - if (memsize > ctx->lowmem_limit) { - ctx->lowmem = ctx->lowmem_limit; - ctx->highmem = memsize - ctx->lowmem_limit; - objsize = 4*GB + ctx->highmem; - } else { - ctx->lowmem = memsize; - ctx->highmem = 0; - objsize = ctx->lowmem; - } - - error = vm_alloc_memseg(ctx, VM_SYSMEM, objsize, NULL); - if (error) - return (error); - - /* - * Stake out a contiguous region covering the guest physical memory - * and the adjoining guard regions. - */ - len = VM_MMAP_GUARD_SIZE + objsize + VM_MMAP_GUARD_SIZE; - ptr = mmap(NULL, len, PROT_NONE, MAP_GUARD | MAP_ALIGNED_SUPER, -1, 0); - if (ptr == MAP_FAILED) - return (-1); - - baseaddr = ptr + VM_MMAP_GUARD_SIZE; - if (ctx->highmem > 0) { - gpa = 4*GB; - len = ctx->highmem; - error = setup_memory_segment(ctx, gpa, len, baseaddr); - if (error) - return (error); - } - - if (ctx->lowmem > 0) { - gpa = 0; - len = ctx->lowmem; - error = setup_memory_segment(ctx, gpa, len, baseaddr); - if (error) - return (error); - } - - ctx->baseaddr = baseaddr; - - return (0); -} - -/* - * Returns a non-NULL pointer if [gaddr, gaddr+len) is entirely contained in - * the lowmem or highmem regions. - * - * In particular return NULL if [gaddr, gaddr+len) falls in guest MMIO region. - * The instruction emulation code depends on this behavior. - */ -void * -vm_map_gpa(struct vmctx *ctx, vm_paddr_t gaddr, size_t len) -{ - - if (ctx->lowmem > 0) { - if (gaddr < ctx->lowmem && len <= ctx->lowmem && - gaddr + len <= ctx->lowmem) - return (ctx->baseaddr + gaddr); - } - - if (ctx->highmem > 0) { - if (gaddr >= 4*GB) { - if (gaddr < 4*GB + ctx->highmem && - len <= ctx->highmem && - gaddr + len <= 4*GB + ctx->highmem) - return (ctx->baseaddr + gaddr); - } - } - - return (NULL); -} - -size_t -vm_get_lowmem_size(struct vmctx *ctx) -{ - - return (ctx->lowmem); -} - -size_t -vm_get_highmem_size(struct vmctx *ctx) -{ - - return (ctx->highmem); -} - -void * -vm_create_devmem(struct vmctx *ctx, int segid, const char *name, size_t len) -{ - char pathname[MAXPATHLEN]; - size_t len2; - char *base, *ptr; - int fd, error, flags; - - fd = -1; - ptr = MAP_FAILED; - if (name == NULL || strlen(name) == 0) { - errno = EINVAL; - goto done; - } - - error = vm_alloc_memseg(ctx, segid, len, name); - if (error) - goto done; - - strlcpy(pathname, "/dev/vmm.io/", sizeof(pathname)); - strlcat(pathname, ctx->name, sizeof(pathname)); - strlcat(pathname, ".", sizeof(pathname)); - strlcat(pathname, name, sizeof(pathname)); - - fd = open(pathname, O_RDWR); - if (fd < 0) - goto done; - - /* - * Stake out a contiguous region covering the device memory and the - * adjoining guard regions. - */ - len2 = VM_MMAP_GUARD_SIZE + len + VM_MMAP_GUARD_SIZE; - base = mmap(NULL, len2, PROT_NONE, MAP_GUARD | MAP_ALIGNED_SUPER, -1, - 0); - if (base == MAP_FAILED) - goto done; - - flags = MAP_SHARED | MAP_FIXED; - if ((ctx->memflags & VM_MEM_F_INCORE) == 0) - flags |= MAP_NOCORE; - - /* mmap the devmem region in the host address space */ - ptr = mmap(base + VM_MMAP_GUARD_SIZE, len, PROT_RW, flags, fd, 0); -done: - if (fd >= 0) - close(fd); - return (ptr); -} - -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_get_seg_desc(struct vmctx *ctx, int vcpu, int reg, struct seg_desc *seg_desc) -{ - int error; - - error = vm_get_desc(ctx, vcpu, reg, &seg_desc->base, &seg_desc->limit, - &seg_desc->access); - return (error); -} - int vm_set_register(struct vmctx *ctx, int vcpu, int reg, uint64_t val) { @@ -651,205 +200,6 @@ return (ioctl(ctx->fd, VM_REINIT, 0)); } -int -vm_inject_exception(struct vmctx *ctx, int vcpu, int vector, int errcode_valid, - uint32_t errcode, int restart_instruction) -{ - struct vm_exception exc; - - exc.cpuid = vcpu; - exc.vector = vector; - exc.error_code = errcode; - exc.error_code_valid = errcode_valid; - exc.restart_instruction = restart_instruction; - - return (ioctl(ctx->fd, VM_INJECT_EXCEPTION, &exc)); -} - -int -vm_apicid2vcpu(struct vmctx *ctx, int apicid) -{ - /* - * The apic id associated with the 'vcpu' has the same numerical value - * as the 'vcpu' itself. - */ - return (apicid); -} - -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_lapic_local_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_LOCAL_IRQ, &vmirq)); -} - -int -vm_lapic_msi(struct vmctx *ctx, uint64_t addr, uint64_t msg) -{ - struct vm_lapic_msi vmmsi; - - bzero(&vmmsi, sizeof(vmmsi)); - vmmsi.addr = addr; - vmmsi.msg = msg; - - return (ioctl(ctx->fd, VM_LAPIC_MSI, &vmmsi)); -} - -int -vm_ioapic_assert_irq(struct vmctx *ctx, int irq) -{ - struct vm_ioapic_irq ioapic_irq; - - bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq)); - ioapic_irq.irq = irq; - - return (ioctl(ctx->fd, VM_IOAPIC_ASSERT_IRQ, &ioapic_irq)); -} - -int -vm_ioapic_deassert_irq(struct vmctx *ctx, int irq) -{ - struct vm_ioapic_irq ioapic_irq; - - bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq)); - ioapic_irq.irq = irq; - - return (ioctl(ctx->fd, VM_IOAPIC_DEASSERT_IRQ, &ioapic_irq)); -} - -int -vm_ioapic_pulse_irq(struct vmctx *ctx, int irq) -{ - struct vm_ioapic_irq ioapic_irq; - - bzero(&ioapic_irq, sizeof(struct vm_ioapic_irq)); - ioapic_irq.irq = irq; - - return (ioctl(ctx->fd, VM_IOAPIC_PULSE_IRQ, &ioapic_irq)); -} - -int -vm_ioapic_pincount(struct vmctx *ctx, int *pincount) -{ - - return (ioctl(ctx->fd, VM_IOAPIC_PINCOUNT, pincount)); -} - -int -vm_isa_assert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq) -{ - struct vm_isa_irq isa_irq; - - bzero(&isa_irq, sizeof(struct vm_isa_irq)); - isa_irq.atpic_irq = atpic_irq; - isa_irq.ioapic_irq = ioapic_irq; - - return (ioctl(ctx->fd, VM_ISA_ASSERT_IRQ, &isa_irq)); -} - -int -vm_isa_deassert_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq) -{ - struct vm_isa_irq isa_irq; - - bzero(&isa_irq, sizeof(struct vm_isa_irq)); - isa_irq.atpic_irq = atpic_irq; - isa_irq.ioapic_irq = ioapic_irq; - - return (ioctl(ctx->fd, VM_ISA_DEASSERT_IRQ, &isa_irq)); -} - -int -vm_isa_pulse_irq(struct vmctx *ctx, int atpic_irq, int ioapic_irq) -{ - struct vm_isa_irq isa_irq; - - bzero(&isa_irq, sizeof(struct vm_isa_irq)); - isa_irq.atpic_irq = atpic_irq; - isa_irq.ioapic_irq = ioapic_irq; - - return (ioctl(ctx->fd, VM_ISA_PULSE_IRQ, &isa_irq)); -} - -int -vm_isa_set_irq_trigger(struct vmctx *ctx, int atpic_irq, - enum vm_intr_trigger trigger) -{ - struct vm_isa_irq_trigger isa_irq_trigger; - - bzero(&isa_irq_trigger, sizeof(struct vm_isa_irq_trigger)); - isa_irq_trigger.atpic_irq = atpic_irq; - isa_irq_trigger.trigger = trigger; - - return (ioctl(ctx->fd, VM_ISA_SET_IRQ_TRIGGER, &isa_irq_trigger)); -} - -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)); -} - -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 }, - { "enable_invpcid", VM_CAP_ENABLE_INVPCID }, - { 0 } -}; - -int -vm_capability_name2type(const char *capname) -{ - int i; - - for (i = 0; capstrmap[i].name != NULL && capname != NULL; i++) { - if (strcmp(capstrmap[i].name, capname) == 0) - return (capstrmap[i].type); - } - - return (-1); -} - -const char * -vm_capability_type2name(int type) -{ - int i; - - for (i = 0; capstrmap[i].name != NULL; i++) { - if (capstrmap[i].type == type) - return (capstrmap[i].name); - } - - return (NULL); -} - int vm_get_capability(struct vmctx *ctx, int vcpu, enum vm_cap_type cap, int *retval) @@ -992,192 +342,6 @@ return (NULL); } -int -vm_get_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state *state) -{ - int error; - struct vm_x2apic x2apic; - - bzero(&x2apic, sizeof(x2apic)); - x2apic.cpuid = vcpu; - - error = ioctl(ctx->fd, VM_GET_X2APIC_STATE, &x2apic); - *state = x2apic.state; - return (error); -} - -int -vm_set_x2apic_state(struct vmctx *ctx, int vcpu, enum x2apic_state state) -{ - int error; - struct vm_x2apic x2apic; - - bzero(&x2apic, sizeof(x2apic)); - x2apic.cpuid = vcpu; - x2apic.state = state; - - error = ioctl(ctx->fd, VM_SET_X2APIC_STATE, &x2apic); - - return (error); -} - -/* - * 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 = 0; - 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); -} - int vm_get_gpa_pmap(struct vmctx *ctx, uint64_t gpa, uint64_t *pte, int *num) { @@ -1198,19 +362,6 @@ return (error); } -int -vm_get_hpet_capabilities(struct vmctx *ctx, uint32_t *capabilities) -{ - int error; - struct vm_hpet_cap cap; - - bzero(&cap, sizeof(struct vm_hpet_cap)); - error = ioctl(ctx->fd, VM_GET_HPET_CAPABILITIES, &cap); - if (capabilities != NULL) - *capabilities = cap.capabilities; - return (error); -} - int vm_gla2gpa(struct vmctx *ctx, int vcpu, struct vm_guest_paging *paging, uint64_t gla, int prot, uint64_t *gpa, int *fault) @@ -1415,87 +566,6 @@ return (error); } -int -vm_get_intinfo(struct vmctx *ctx, int vcpu, uint64_t *info1, uint64_t *info2) -{ - struct vm_intinfo vmii; - int error; - - bzero(&vmii, sizeof(struct vm_intinfo)); - vmii.vcpuid = vcpu; - error = ioctl(ctx->fd, VM_GET_INTINFO, &vmii); - if (error == 0) { - *info1 = vmii.info1; - *info2 = vmii.info2; - } - return (error); -} - -int -vm_set_intinfo(struct vmctx *ctx, int vcpu, uint64_t info1) -{ - struct vm_intinfo vmii; - int error; - - bzero(&vmii, sizeof(struct vm_intinfo)); - vmii.vcpuid = vcpu; - vmii.info1 = info1; - error = ioctl(ctx->fd, VM_SET_INTINFO, &vmii); - return (error); -} - -int -vm_rtc_write(struct vmctx *ctx, int offset, uint8_t value) -{ - struct vm_rtc_data rtcdata; - int error; - - bzero(&rtcdata, sizeof(struct vm_rtc_data)); - rtcdata.offset = offset; - rtcdata.value = value; - error = ioctl(ctx->fd, VM_RTC_WRITE, &rtcdata); - return (error); -} - -int -vm_rtc_read(struct vmctx *ctx, int offset, uint8_t *retval) -{ - struct vm_rtc_data rtcdata; - int error; - - bzero(&rtcdata, sizeof(struct vm_rtc_data)); - rtcdata.offset = offset; - error = ioctl(ctx->fd, VM_RTC_READ, &rtcdata); - if (error == 0) - *retval = rtcdata.value; - return (error); -} - -int -vm_rtc_settime(struct vmctx *ctx, time_t secs) -{ - struct vm_rtc_time rtctime; - int error; - - bzero(&rtctime, sizeof(struct vm_rtc_time)); - rtctime.secs = secs; - error = ioctl(ctx->fd, VM_RTC_SETTIME, &rtctime); - return (error); -} - -int -vm_rtc_gettime(struct vmctx *ctx, time_t *secs) -{ - struct vm_rtc_time rtctime; - int error; - - bzero(&rtctime, sizeof(struct vm_rtc_time)); - error = ioctl(ctx->fd, VM_RTC_GETTIME, &rtctime); - if (error == 0) - *secs = rtctime.secs; - return (error); -} - int vm_restart_instruction(void *arg, int vcpu) { @@ -1535,47 +605,3 @@ } return (error); } - -int -vm_get_device_fd(struct vmctx *ctx) -{ - - return (ctx->fd); -} - -const cap_ioctl_t * -vm_get_ioctls(size_t *len) -{ - cap_ioctl_t *cmds; - /* keep in sync with machine/vmm_dev.h */ - static const cap_ioctl_t vm_ioctl_cmds[] = { VM_RUN, VM_SUSPEND, VM_REINIT, - VM_ALLOC_MEMSEG, VM_GET_MEMSEG, VM_MMAP_MEMSEG, VM_MMAP_MEMSEG, - VM_MMAP_GETNEXT, VM_SET_REGISTER, VM_GET_REGISTER, - VM_SET_SEGMENT_DESCRIPTOR, VM_GET_SEGMENT_DESCRIPTOR, - VM_SET_REGISTER_SET, VM_GET_REGISTER_SET, - VM_INJECT_EXCEPTION, VM_LAPIC_IRQ, VM_LAPIC_LOCAL_IRQ, - VM_LAPIC_MSI, VM_IOAPIC_ASSERT_IRQ, VM_IOAPIC_DEASSERT_IRQ, - VM_IOAPIC_PULSE_IRQ, VM_IOAPIC_PINCOUNT, VM_ISA_ASSERT_IRQ, - VM_ISA_DEASSERT_IRQ, VM_ISA_PULSE_IRQ, VM_ISA_SET_IRQ_TRIGGER, - VM_SET_CAPABILITY, VM_GET_CAPABILITY, VM_BIND_PPTDEV, - VM_UNBIND_PPTDEV, VM_MAP_PPTDEV_MMIO, VM_PPTDEV_MSI, - VM_PPTDEV_MSIX, VM_INJECT_NMI, VM_STATS, VM_STAT_DESC, - VM_SET_X2APIC_STATE, VM_GET_X2APIC_STATE, - VM_GET_HPET_CAPABILITIES, VM_GET_GPA_PMAP, VM_GLA2GPA, - VM_GLA2GPA_NOFAULT, - VM_ACTIVATE_CPU, VM_GET_CPUS, VM_SUSPEND_CPU, VM_RESUME_CPU, - VM_SET_INTINFO, VM_GET_INTINFO, - VM_RTC_WRITE, VM_RTC_READ, VM_RTC_SETTIME, VM_RTC_GETTIME, - VM_RESTART_INSTRUCTION, VM_SET_TOPOLOGY, VM_GET_TOPOLOGY }; - - if (len == NULL) { - cmds = malloc(sizeof(vm_ioctl_cmds)); - if (cmds == NULL) - return (NULL); - bcopy(vm_ioctl_cmds, cmds, sizeof(vm_ioctl_cmds)); - return (cmds); - } - - *len = nitems(vm_ioctl_cmds); - return (NULL); -} Index: lib/libvmmapi/vmmapi_internal.h =================================================================== --- /dev/null +++ lib/libvmmapi/vmmapi_internal.h @@ -0,0 +1,35 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * 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. + * + * $FreeBSD$ + */ + +#ifndef _VMMAPI_INTERNAL_H_ +#define _VMMAPI_INTERNAL_H_ + +int vm_device_open(const char *name); +#endif Index: lib/libvmmapi/vmmapi_internal.c =================================================================== --- /dev/null +++ lib/libvmmapi/vmmapi_internal.c @@ -0,0 +1,57 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * 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 +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include +#include + +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); +} Index: usr.sbin/bhyve/Makefile =================================================================== --- usr.sbin/bhyve/Makefile +++ usr.sbin/bhyve/Makefile @@ -3,7 +3,7 @@ # .include -CFLAGS+=-I${SRCTOP}/sys +CFLAGS+=-I${SRCTOP}/sys -I${SRCTOP}/lib/libvmmapi .PATH: ${SRCTOP}/sys/cam/ctl PROG= bhyve Index: usr.sbin/bhyvectl/Makefile =================================================================== --- usr.sbin/bhyvectl/Makefile +++ usr.sbin/bhyvectl/Makefile @@ -12,6 +12,6 @@ WARNS?= 3 -CFLAGS+= -I${SRCTOP}/sys/amd64/vmm +CFLAGS+= -I${SRCTOP}/sys/amd64/vmm -I${SRCTOP}/lib/libvmmapi .include Index: usr.sbin/bhyveload/Makefile =================================================================== --- usr.sbin/bhyveload/Makefile +++ usr.sbin/bhyveload/Makefile @@ -9,6 +9,6 @@ WARNS?= 3 -CFLAGS+=-I${SRCTOP}/stand/userboot +CFLAGS+=-I${SRCTOP}/stand/userboot -I${SRCTOP}/lib/libvmmapi .include