[qemu.git] / kvm-all.c

/*
 * QEMU KVM support
 *
 * Copyright IBM, Corp. 2008
 *           Red Hat, Inc. 2008
 *
 * Authors:
 *  Anthony Liguori   <aliguori@us.ibm.com>
 *  Glauber Costa     <gcosta@redhat.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 *
 */

#include <sys/types.h>
#include <sys/ioctl.h>
#include <sys/mman.h>
#include <stdarg.h>

#include <linux/kvm.h>

#include "qemu-common.h"
#include "sysemu.h"
#include "kvm.h"

//#define DEBUG_KVM

#ifdef DEBUG_KVM
#define dprintf(fmt, ...) \
    do { fprintf(stderr, fmt, ## __VA_ARGS__); } while (0)
#else
#define dprintf(fmt, ...) \
    do { } while (0)
#endif

typedef struct KVMSlot
{
    target_phys_addr_t start_addr;
    ram_addr_t memory_size;
    ram_addr_t phys_offset;
    int slot;
    int flags;
} KVMSlot;

typedef struct kvm_dirty_log KVMDirtyLog;

int kvm_allowed = 0;

struct KVMState
{
    KVMSlot slots[32];
    int fd;
    int vmfd;
};

static KVMState *kvm_state;

static KVMSlot *kvm_alloc_slot(KVMState *s)
{
    int i;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
        /* KVM private memory slots */
        if (i >= 8 && i < 12)
            continue;
        if (s->slots[i].memory_size == 0)
            return &s->slots[i];
    }

    return NULL;
}

static KVMSlot *kvm_lookup_slot(KVMState *s, target_phys_addr_t start_addr)
{
    int i;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
        KVMSlot *mem = &s->slots[i];

        if (start_addr >= mem->start_addr &&
            start_addr < (mem->start_addr + mem->memory_size))
            return mem;
    }

    return NULL;
}

static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
{
    struct kvm_userspace_memory_region mem;

    mem.slot = slot->slot;
    mem.guest_phys_addr = slot->start_addr;
    mem.memory_size = slot->memory_size;
    mem.userspace_addr = (unsigned long)phys_ram_base + slot->phys_offset;
    mem.flags = slot->flags;

    return kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
}


int kvm_init_vcpu(CPUState *env)
{
    KVMState *s = kvm_state;
    long mmap_size;
    int ret;

    dprintf("kvm_init_vcpu\n");

    ret = kvm_vm_ioctl(s, KVM_CREATE_VCPU, env->cpu_index);
    if (ret < 0) {
        dprintf("kvm_create_vcpu failed\n");
        goto err;
    }

    env->kvm_fd = ret;
    env->kvm_state = s;

    mmap_size = kvm_ioctl(s, KVM_GET_VCPU_MMAP_SIZE, 0);
    if (mmap_size < 0) {
        dprintf("KVM_GET_VCPU_MMAP_SIZE failed\n");
        goto err;
    }

    env->kvm_run = mmap(NULL, mmap_size, PROT_READ | PROT_WRITE, MAP_SHARED,
                        env->kvm_fd, 0);
    if (env->kvm_run == MAP_FAILED) {
        ret = -errno;
        dprintf("mmap'ing vcpu state failed\n");
        goto err;
    }

    ret = kvm_arch_init_vcpu(env);

err:
    return ret;
}

/*
 * dirty pages logging control
 */
static int kvm_dirty_pages_log_change(target_phys_addr_t phys_addr, target_phys_addr_t end_addr,
                                      unsigned flags,
                                      unsigned mask)
{
    KVMState *s = kvm_state;
    KVMSlot *mem = kvm_lookup_slot(s, phys_addr);
    if (mem == NULL)  {
            dprintf("invalid parameters %llx-%llx\n", phys_addr, end_addr);
            return -EINVAL;
    }

    flags = (mem->flags & ~mask) | flags;
    /* Nothing changed, no need to issue ioctl */
    if (flags == mem->flags)
            return 0;

    mem->flags = flags;

    return kvm_set_user_memory_region(s, mem);
}

int kvm_log_start(target_phys_addr_t phys_addr, target_phys_addr_t end_addr)
{
        return kvm_dirty_pages_log_change(phys_addr, end_addr,
                                          KVM_MEM_LOG_DIRTY_PAGES,
                                          KVM_MEM_LOG_DIRTY_PAGES);
}

int kvm_log_stop(target_phys_addr_t phys_addr, target_phys_addr_t end_addr)
{
        return kvm_dirty_pages_log_change(phys_addr, end_addr,
                                          0,
                                          KVM_MEM_LOG_DIRTY_PAGES);
}

/**
 * kvm_physical_sync_dirty_bitmap - Grab dirty bitmap from kernel space
 * This function updates qemu's dirty bitmap using cpu_physical_memory_set_dirty().
 * This means all bits are set to dirty.
 *
 * @start_add: start of logged region. This is what we use to search the memslot
 * @end_addr: end of logged region.
 */
void kvm_physical_sync_dirty_bitmap(target_phys_addr_t start_addr, target_phys_addr_t end_addr)
{
    KVMState *s = kvm_state;
    KVMDirtyLog d;
    KVMSlot *mem = kvm_lookup_slot(s, start_addr);
    unsigned long alloc_size;
    ram_addr_t addr;
    target_phys_addr_t phys_addr = start_addr;

    dprintf("sync addr: %llx into %lx\n", start_addr, mem->phys_offset);
    if (mem == NULL) {
            fprintf(stderr, "BUG: %s: invalid parameters\n", __func__);
            return;
    }

    alloc_size = mem->memory_size >> TARGET_PAGE_BITS / sizeof(d.dirty_bitmap);
    d.dirty_bitmap = qemu_mallocz(alloc_size);

    if (d.dirty_bitmap == NULL) {
        dprintf("Could not allocate dirty bitmap\n");
        return;
    }

    d.slot = mem->slot;
    dprintf("slot %d, phys_addr %llx, uaddr: %llx\n",
            d.slot, mem->start_addr, mem->phys_offset);

    if (kvm_vm_ioctl(s, KVM_GET_DIRTY_LOG, &d) == -1) {
        dprintf("ioctl failed %d\n", errno);
        goto out;
    }

    phys_addr = start_addr;
    for (addr = mem->phys_offset; phys_addr < end_addr; phys_addr+= TARGET_PAGE_SIZE, addr += TARGET_PAGE_SIZE) {
        unsigned long *bitmap = (unsigned long *)d.dirty_bitmap;
        unsigned nr = (phys_addr - start_addr) >> TARGET_PAGE_BITS;
        unsigned word = nr / (sizeof(*bitmap) * 8);
        unsigned bit = nr % (sizeof(*bitmap) * 8);
        if ((bitmap[word] >> bit) & 1)
            cpu_physical_memory_set_dirty(addr);
    }
out:
    qemu_free(d.dirty_bitmap);
}

int kvm_init(int smp_cpus)
{
    KVMState *s;
    int ret;
    int i;

    if (smp_cpus > 1)
        return -EINVAL;

    s = qemu_mallocz(sizeof(KVMState));
    if (s == NULL)
        return -ENOMEM;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++)
        s->slots[i].slot = i;

    s->vmfd = -1;
    s->fd = open("/dev/kvm", O_RDWR);
    if (s->fd == -1) {
        fprintf(stderr, "Could not access KVM kernel module: %m\n");
        ret = -errno;
        goto err;
    }

    ret = kvm_ioctl(s, KVM_GET_API_VERSION, 0);
    if (ret < KVM_API_VERSION) {
        if (ret > 0)
            ret = -EINVAL;
        fprintf(stderr, "kvm version too old\n");
        goto err;
    }

    if (ret > KVM_API_VERSION) {
        ret = -EINVAL;
        fprintf(stderr, "kvm version not supported\n");
        goto err;
    }

    s->vmfd = kvm_ioctl(s, KVM_CREATE_VM, 0);
    if (s->vmfd < 0)
        goto err;

    /* initially, KVM allocated its own memory and we had to jump through
     * hooks to make phys_ram_base point to this.  Modern versions of KVM
     * just use a user allocated buffer so we can use phys_ram_base
     * unmodified.  Make sure we have a sufficiently modern version of KVM.
     */
    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_USER_MEMORY);
    if (ret <= 0) {
        if (ret == 0)
            ret = -EINVAL;
        fprintf(stderr, "kvm does not support KVM_CAP_USER_MEMORY\n");
        goto err;
    }

    ret = kvm_arch_init(s, smp_cpus);
    if (ret < 0)
        goto err;

    kvm_state = s;

    return 0;

err:
    if (s) {
        if (s->vmfd != -1)
            close(s->vmfd);
        if (s->fd != -1)
            close(s->fd);
    }
    qemu_free(s);

    return ret;
}

static int kvm_handle_io(CPUState *env, uint16_t port, void *data,
                         int direction, int size, uint32_t count)
{
    int i;
    uint8_t *ptr = data;

    for (i = 0; i < count; i++) {
        if (direction == KVM_EXIT_IO_IN) {
            switch (size) {
            case 1:
                stb_p(ptr, cpu_inb(env, port));
                break;
            case 2:
                stw_p(ptr, cpu_inw(env, port));
                break;
            case 4:
                stl_p(ptr, cpu_inl(env, port));
                break;
            }
        } else {
            switch (size) {
            case 1:
                cpu_outb(env, port, ldub_p(ptr));
                break;
            case 2:
                cpu_outw(env, port, lduw_p(ptr));
                break;
            case 4:
                cpu_outl(env, port, ldl_p(ptr));
                break;
            }
        }

        ptr += size;
    }

    return 1;
}

int kvm_cpu_exec(CPUState *env)
{
    struct kvm_run *run = env->kvm_run;
    int ret;

    dprintf("kvm_cpu_exec()\n");

    do {
        kvm_arch_pre_run(env, run);

        if ((env->interrupt_request & CPU_INTERRUPT_EXIT)) {
            dprintf("interrupt exit requested\n");
            ret = 0;
            break;
        }

        ret = kvm_vcpu_ioctl(env, KVM_RUN, 0);
        kvm_arch_post_run(env, run);

        if (ret == -EINTR || ret == -EAGAIN) {
            dprintf("io window exit\n");
            ret = 0;
            break;
        }

        if (ret < 0) {
            dprintf("kvm run failed %s\n", strerror(-ret));
            abort();
        }

        ret = 0; /* exit loop */
        switch (run->exit_reason) {
        case KVM_EXIT_IO:
            dprintf("handle_io\n");
            ret = kvm_handle_io(env, run->io.port,
                                (uint8_t *)run + run->io.data_offset,
                                run->io.direction,
                                run->io.size,
                                run->io.count);
            break;
        case KVM_EXIT_MMIO:
            dprintf("handle_mmio\n");
            cpu_physical_memory_rw(run->mmio.phys_addr,
                                   run->mmio.data,
                                   run->mmio.len,
                                   run->mmio.is_write);
            ret = 1;
            break;
        case KVM_EXIT_IRQ_WINDOW_OPEN:
            dprintf("irq_window_open\n");
            break;
        case KVM_EXIT_SHUTDOWN:
            dprintf("shutdown\n");
            qemu_system_reset_request();
            ret = 1;
            break;
        case KVM_EXIT_UNKNOWN:
            dprintf("kvm_exit_unknown\n");
            break;
        case KVM_EXIT_FAIL_ENTRY:
            dprintf("kvm_exit_fail_entry\n");
            break;
        case KVM_EXIT_EXCEPTION:
            dprintf("kvm_exit_exception\n");
            break;
        case KVM_EXIT_DEBUG:
            dprintf("kvm_exit_debug\n");
            break;
        default:
            dprintf("kvm_arch_handle_exit\n");
            ret = kvm_arch_handle_exit(env, run);
            break;
        }
    } while (ret > 0);

    if ((env->interrupt_request & CPU_INTERRUPT_EXIT)) {
        env->interrupt_request &= ~CPU_INTERRUPT_EXIT;
        env->exception_index = EXCP_INTERRUPT;
    }

    return ret;
}

void kvm_set_phys_mem(target_phys_addr_t start_addr,
                      ram_addr_t size,
                      ram_addr_t phys_offset)
{
    KVMState *s = kvm_state;
    ram_addr_t flags = phys_offset & ~TARGET_PAGE_MASK;
    KVMSlot *mem;

    /* KVM does not support read-only slots */
    phys_offset &= ~IO_MEM_ROM;

    mem = kvm_lookup_slot(s, start_addr);
    if (mem) {
        if ((flags == IO_MEM_UNASSIGNED) || (flags >= TLB_MMIO)) {
            mem->memory_size = 0;
            mem->start_addr = start_addr;
            mem->phys_offset = 0;
            mem->flags = 0;

            kvm_set_user_memory_region(s, mem);
        } else if (start_addr >= mem->start_addr &&
                   (start_addr + size) <= (mem->start_addr +
                                           mem->memory_size)) {
            KVMSlot slot;
            target_phys_addr_t mem_start;
            ram_addr_t mem_size, mem_offset;

            /* Not splitting */
            if ((phys_offset - (start_addr - mem->start_addr)) == 
                mem->phys_offset)
                return;

            /* unregister whole slot */
            memcpy(&slot, mem, sizeof(slot));
            mem->memory_size = 0;
            kvm_set_user_memory_region(s, mem);

            /* register prefix slot */
            mem_start = slot.start_addr;
            mem_size = start_addr - slot.start_addr;
            mem_offset = slot.phys_offset;
            if (mem_size)
                kvm_set_phys_mem(mem_start, mem_size, mem_offset);

            /* register new slot */
            kvm_set_phys_mem(start_addr, size, phys_offset);

            /* register suffix slot */
            mem_start = start_addr + size;
            mem_offset += mem_size + size;
            mem_size = slot.memory_size - mem_size - size;
            if (mem_size)
                kvm_set_phys_mem(mem_start, mem_size, mem_offset);

            return;
        } else {
            printf("Registering overlapping slot\n");
            abort();
        }
    }
    /* KVM does not need to know about this memory */
    if (flags >= IO_MEM_UNASSIGNED)
        return;

    mem = kvm_alloc_slot(s);
    mem->memory_size = size;
    mem->start_addr = start_addr;
    mem->phys_offset = phys_offset;
    mem->flags = 0;

    kvm_set_user_memory_region(s, mem);
    /* FIXME deal with errors */
}

int kvm_ioctl(KVMState *s, int type, ...)
{
    int ret;
    void *arg;
    va_list ap;

    va_start(ap, type);
    arg = va_arg(ap, void *);
    va_end(ap);

    ret = ioctl(s->fd, type, arg);
    if (ret == -1)
        ret = -errno;

    return ret;
}

int kvm_vm_ioctl(KVMState *s, int type, ...)
{
    int ret;
    void *arg;
    va_list ap;

    va_start(ap, type);
    arg = va_arg(ap, void *);
    va_end(ap);

    ret = ioctl(s->vmfd, type, arg);
    if (ret == -1)
        ret = -errno;

    return ret;
}

int kvm_vcpu_ioctl(CPUState *env, int type, ...)
{
    int ret;
    void *arg;
    va_list ap;

    va_start(ap, type);
    arg = va_arg(ap, void *);
    va_end(ap);

    ret = ioctl(env->kvm_fd, type, arg);
    if (ret == -1)
        ret = -errno;

    return ret;
}
Commit	Line	Data
05330448 AL	1	/*
	2	* QEMU KVM support
	3	*
	4	* Copyright IBM, Corp. 2008
5832d1f2	5	* Red Hat, Inc. 2008
05330448 AL	6	*
	7	* Authors:
	8	* Anthony Liguori <aliguori@us.ibm.com>
5832d1f2	9	* Glauber Costa <gcosta@redhat.com>
05330448 AL	10	*
	11	* This work is licensed under the terms of the GNU GPL, version 2 or later.
	12	* See the COPYING file in the top-level directory.
	13	*
	14	*/
	15
	16	#include <sys/types.h>
	17	#include <sys/ioctl.h>
	18	#include <sys/mman.h>
984b5181	19	#include <stdarg.h>
05330448 AL	20
	21	#include <linux/kvm.h>
	22
	23	#include "qemu-common.h"
	24	#include "sysemu.h"
	25	#include "kvm.h"
	26
	27	//#define DEBUG_KVM
	28
	29	#ifdef DEBUG_KVM
	30	#define dprintf(fmt, ...) \
	31	do { fprintf(stderr, fmt, ## __VA_ARGS__); } while (0)
	32	#else
	33	#define dprintf(fmt, ...) \
	34	do { } while (0)
	35	#endif
	36
34fc643f AL	37	typedef struct KVMSlot
	38	{
	39	target_phys_addr_t start_addr;
	40	ram_addr_t memory_size;
	41	ram_addr_t phys_offset;
	42	int slot;
	43	int flags;
	44	} KVMSlot;
05330448	45
5832d1f2 AL	46	typedef struct kvm_dirty_log KVMDirtyLog;
5832d1f2 AL	47
05330448 AL	48	int kvm_allowed = 0;
	49
	50	struct KVMState
	51	{
	52	KVMSlot slots[32];
	53	int fd;
	54	int vmfd;
	55	};
	56
	57	static KVMState *kvm_state;
	58
	59	static KVMSlot kvm_alloc_slot(KVMState s)
	60	{
	61	int i;
	62
	63	for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
62d60e8c AL	64	/* KVM private memory slots */
	65	if (i >= 8 && i < 12)
	66	continue;
05330448 AL	67	if (s->slots[i].memory_size == 0)
	68	return &s->slots[i];
	69	}
	70
	71	return NULL;
	72	}
	73
	74	static KVMSlot kvm_lookup_slot(KVMState s, target_phys_addr_t start_addr)
	75	{
	76	int i;
	77
	78	for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
	79	KVMSlot *mem = &s->slots[i];
	80
34fc643f AL	81	if (start_addr >= mem->start_addr &&
34fc643f AL	82	start_addr < (mem->start_addr + mem->memory_size))
05330448 AL	83	return mem;
	84	}
	85
	86	return NULL;
	87	}
	88
5832d1f2 AL	89	static int kvm_set_user_memory_region(KVMState s, KVMSlot slot)
	90	{
	91	struct kvm_userspace_memory_region mem;
	92
	93	mem.slot = slot->slot;
	94	mem.guest_phys_addr = slot->start_addr;
	95	mem.memory_size = slot->memory_size;
	96	mem.userspace_addr = (unsigned long)phys_ram_base + slot->phys_offset;
	97	mem.flags = slot->flags;
	98
	99	return kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
	100	}
	101
	102
05330448 AL	103	int kvm_init_vcpu(CPUState *env)
	104	{
	105	KVMState *s = kvm_state;
	106	long mmap_size;
	107	int ret;
	108
	109	dprintf("kvm_init_vcpu\n");
	110
984b5181	111	ret = kvm_vm_ioctl(s, KVM_CREATE_VCPU, env->cpu_index);
05330448 AL	112	if (ret < 0) {
	113	dprintf("kvm_create_vcpu failed\n");
	114	goto err;
	115	}
	116
	117	env->kvm_fd = ret;
	118	env->kvm_state = s;
	119
	120	mmap_size = kvm_ioctl(s, KVM_GET_VCPU_MMAP_SIZE, 0);
	121	if (mmap_size < 0) {
	122	dprintf("KVM_GET_VCPU_MMAP_SIZE failed\n");
	123	goto err;
	124	}
	125
	126	env->kvm_run = mmap(NULL, mmap_size, PROT_READ \| PROT_WRITE, MAP_SHARED,
	127	env->kvm_fd, 0);
	128	if (env->kvm_run == MAP_FAILED) {
	129	ret = -errno;
	130	dprintf("mmap'ing vcpu state failed\n");
	131	goto err;
	132	}
	133
	134	ret = kvm_arch_init_vcpu(env);
	135
	136	err:
	137	return ret;
	138	}
	139
5832d1f2 AL	140	/*
	141	* dirty pages logging control
	142	*/
	143	static int kvm_dirty_pages_log_change(target_phys_addr_t phys_addr, target_phys_addr_t end_addr,
	144	unsigned flags,
	145	unsigned mask)
	146	{
	147	KVMState *s = kvm_state;
	148	KVMSlot *mem = kvm_lookup_slot(s, phys_addr);
	149	if (mem == NULL) {
	150	dprintf("invalid parameters %llx-%llx\n", phys_addr, end_addr);
	151	return -EINVAL;
	152	}
	153
	154	flags = (mem->flags & ~mask) \| flags;
	155	/* Nothing changed, no need to issue ioctl */
	156	if (flags == mem->flags)
	157	return 0;
	158
	159	mem->flags = flags;
	160
	161	return kvm_set_user_memory_region(s, mem);
	162	}
	163
	164	int kvm_log_start(target_phys_addr_t phys_addr, target_phys_addr_t end_addr)
	165	{
	166	return kvm_dirty_pages_log_change(phys_addr, end_addr,
	167	KVM_MEM_LOG_DIRTY_PAGES,
	168	KVM_MEM_LOG_DIRTY_PAGES);
	169	}
	170
	171	int kvm_log_stop(target_phys_addr_t phys_addr, target_phys_addr_t end_addr)
	172	{
	173	return kvm_dirty_pages_log_change(phys_addr, end_addr,
	174	0,
	175	KVM_MEM_LOG_DIRTY_PAGES);
	176	}
	177
	178	/**
	179	* kvm_physical_sync_dirty_bitmap - Grab dirty bitmap from kernel space
	180	* This function updates qemu's dirty bitmap using cpu_physical_memory_set_dirty().
	181	* This means all bits are set to dirty.
	182	*
	183	* @start_add: start of logged region. This is what we use to search the memslot
	184	* @end_addr: end of logged region.
	185	*/
	186	void kvm_physical_sync_dirty_bitmap(target_phys_addr_t start_addr, target_phys_addr_t end_addr)
	187	{
	188	KVMState *s = kvm_state;
	189	KVMDirtyLog d;
	190	KVMSlot *mem = kvm_lookup_slot(s, start_addr);
	191	unsigned long alloc_size;
	192	ram_addr_t addr;
	193	target_phys_addr_t phys_addr = start_addr;
	194
	195	dprintf("sync addr: %llx into %lx\n", start_addr, mem->phys_offset);
	196	if (mem == NULL) {
	197	fprintf(stderr, "BUG: %s: invalid parameters\n", __func__);
	198	return;
	199	}
	200
	201	alloc_size = mem->memory_size >> TARGET_PAGE_BITS / sizeof(d.dirty_bitmap);
	202	d.dirty_bitmap = qemu_mallocz(alloc_size);
	203
204	if (d.dirty_bitmap == NULL) {
205	dprintf("Could not allocate dirty bitmap\n");
206	return;
207	}
208
209	d.slot = mem->slot;
210	dprintf("slot %d, phys_addr %llx, uaddr: %llx\n",
211	d.slot, mem->start_addr, mem->phys_offset);
212
213	if (kvm_vm_ioctl(s, KVM_GET_DIRTY_LOG, &d) == -1) {
214	dprintf("ioctl failed %d\n", errno);
215	goto out;
216	}
217
218	phys_addr = start_addr;
219	for (addr = mem->phys_offset; phys_addr < end_addr; phys_addr+= TARGET_PAGE_SIZE, addr += TARGET_PAGE_SIZE) {
220	unsigned long bitmap = (unsigned long )d.dirty_bitmap;
221	unsigned nr = (phys_addr - start_addr) >> TARGET_PAGE_BITS;
222	unsigned word = nr / (sizeof(bitmap) 8);
223	unsigned bit = nr % (sizeof(bitmap) 8);
224	if ((bitmap[word] >> bit) & 1)
225	cpu_physical_memory_set_dirty(addr);
226	}
227	out:
228	qemu_free(d.dirty_bitmap);
229	}
230
05330448 AL	231	int kvm_init(int smp_cpus)
	232	{
	233	KVMState *s;
	234	int ret;
	235	int i;
	236
	237	if (smp_cpus > 1)
	238	return -EINVAL;
	239
	240	s = qemu_mallocz(sizeof(KVMState));
	241	if (s == NULL)
	242	return -ENOMEM;
	243
	244	for (i = 0; i < ARRAY_SIZE(s->slots); i++)
	245	s->slots[i].slot = i;
	246
	247	s->vmfd = -1;
	248	s->fd = open("/dev/kvm", O_RDWR);
	249	if (s->fd == -1) {
	250	fprintf(stderr, "Could not access KVM kernel module: %m\n");
	251	ret = -errno;
	252	goto err;
	253	}
	254
	255	ret = kvm_ioctl(s, KVM_GET_API_VERSION, 0);
	256	if (ret < KVM_API_VERSION) {
	257	if (ret > 0)
	258	ret = -EINVAL;
	259	fprintf(stderr, "kvm version too old\n");
	260	goto err;
	261	}
	262
	263	if (ret > KVM_API_VERSION) {
	264	ret = -EINVAL;
	265	fprintf(stderr, "kvm version not supported\n");
	266	goto err;
	267	}
	268
	269	s->vmfd = kvm_ioctl(s, KVM_CREATE_VM, 0);
	270	if (s->vmfd < 0)
	271	goto err;
	272
	273	/* initially, KVM allocated its own memory and we had to jump through
	274	* hooks to make phys_ram_base point to this. Modern versions of KVM
	275	* just use a user allocated buffer so we can use phys_ram_base
	276	* unmodified. Make sure we have a sufficiently modern version of KVM.
	277	*/
984b5181	278	ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_USER_MEMORY);
05330448 AL	279	if (ret <= 0) {
	280	if (ret == 0)
	281	ret = -EINVAL;
	282	fprintf(stderr, "kvm does not support KVM_CAP_USER_MEMORY\n");
	283	goto err;
	284	}
	285
	286	ret = kvm_arch_init(s, smp_cpus);
	287	if (ret < 0)
	288	goto err;
	289
	290	kvm_state = s;
	291
	292	return 0;
	293
	294	err:
	295	if (s) {
	296	if (s->vmfd != -1)
	297	close(s->vmfd);
	298	if (s->fd != -1)
	299	close(s->fd);
	300	}
	301	qemu_free(s);
	302
	303	return ret;
	304	}
	305
	306	static int kvm_handle_io(CPUState env, uint16_t port, void data,
	307	int direction, int size, uint32_t count)
	308	{
	309	int i;
	310	uint8_t *ptr = data;
	311
	312	for (i = 0; i < count; i++) {
	313	if (direction == KVM_EXIT_IO_IN) {
	314	switch (size) {
	315	case 1:
	316	stb_p(ptr, cpu_inb(env, port));
	317	break;
	318	case 2:
	319	stw_p(ptr, cpu_inw(env, port));
	320	break;
	321	case 4:
	322	stl_p(ptr, cpu_inl(env, port));
	323	break;
	324	}
	325	} else {
	326	switch (size) {
	327	case 1:
	328	cpu_outb(env, port, ldub_p(ptr));
	329	break;
	330	case 2:
	331	cpu_outw(env, port, lduw_p(ptr));
	332	break;
	333	case 4:
	334	cpu_outl(env, port, ldl_p(ptr));
	335	break;
	336	}
	337	}
	338
	339	ptr += size;
	340	}
	341
	342	return 1;
343	}
344
345	int kvm_cpu_exec(CPUState *env)
346	{
347	struct kvm_run *run = env->kvm_run;
348	int ret;
349
350	dprintf("kvm_cpu_exec()\n");
351
352	do {
353	kvm_arch_pre_run(env, run);
354
355	if ((env->interrupt_request & CPU_INTERRUPT_EXIT)) {
356	dprintf("interrupt exit requested\n");
357	ret = 0;
358	break;
359	}
360
361	ret = kvm_vcpu_ioctl(env, KVM_RUN, 0);
362	kvm_arch_post_run(env, run);
363
364	if (ret == -EINTR \|\| ret == -EAGAIN) {
365	dprintf("io window exit\n");
366	ret = 0;
367	break;
368	}
369
370	if (ret < 0) {
371	dprintf("kvm run failed %s\n", strerror(-ret));
372	abort();
373	}
374
375	ret = 0; /* exit loop */
376	switch (run->exit_reason) {
377	case KVM_EXIT_IO:
378	dprintf("handle_io\n");
379	ret = kvm_handle_io(env, run->io.port,
380	(uint8_t *)run + run->io.data_offset,
381	run->io.direction,
382	run->io.size,
383	run->io.count);
384	break;
385	case KVM_EXIT_MMIO:
386	dprintf("handle_mmio\n");
387	cpu_physical_memory_rw(run->mmio.phys_addr,
388	run->mmio.data,
389	run->mmio.len,
390	run->mmio.is_write);
391	ret = 1;
392	break;
393	case KVM_EXIT_IRQ_WINDOW_OPEN:
394	dprintf("irq_window_open\n");
395	break;
396	case KVM_EXIT_SHUTDOWN:
397	dprintf("shutdown\n");
398	qemu_system_reset_request();
399	ret = 1;
400	break;
401	case KVM_EXIT_UNKNOWN:
402	dprintf("kvm_exit_unknown\n");
403	break;
404	case KVM_EXIT_FAIL_ENTRY:
405	dprintf("kvm_exit_fail_entry\n");
406	break;
407	case KVM_EXIT_EXCEPTION:
408	dprintf("kvm_exit_exception\n");
409	break;
410	case KVM_EXIT_DEBUG:
411	dprintf("kvm_exit_debug\n");
412	break;
413	default:
414	dprintf("kvm_arch_handle_exit\n");
415	ret = kvm_arch_handle_exit(env, run);
416	break;
417	}
418	} while (ret > 0);
419
becfc390 AL	420	if ((env->interrupt_request & CPU_INTERRUPT_EXIT)) {
	421	env->interrupt_request &= ~CPU_INTERRUPT_EXIT;
	422	env->exception_index = EXCP_INTERRUPT;
	423	}
	424
05330448 AL	425	return ret;
	426	}
	427
	428	void kvm_set_phys_mem(target_phys_addr_t start_addr,
	429	ram_addr_t size,
	430	ram_addr_t phys_offset)
	431	{
	432	KVMState *s = kvm_state;
	433	ram_addr_t flags = phys_offset & ~TARGET_PAGE_MASK;
	434	KVMSlot *mem;
	435
	436	/* KVM does not support read-only slots */
	437	phys_offset &= ~IO_MEM_ROM;
	438
	439	mem = kvm_lookup_slot(s, start_addr);
	440	if (mem) {
a3d6841f	441	if ((flags == IO_MEM_UNASSIGNED) \|\| (flags >= TLB_MMIO)) {
05330448	442	mem->memory_size = 0;
34fc643f AL	443	mem->start_addr = start_addr;
34fc643f AL	444	mem->phys_offset = 0;
05330448 AL	445	mem->flags = 0;
05330448 AL	446
34fc643f AL	447	kvm_set_user_memory_region(s, mem);
	448	} else if (start_addr >= mem->start_addr &&
	449	(start_addr + size) <= (mem->start_addr +
62d60e8c AL	450	mem->memory_size)) {
	451	KVMSlot slot;
	452	target_phys_addr_t mem_start;
	453	ram_addr_t mem_size, mem_offset;
	454
	455	/* Not splitting */
34fc643f AL	456	if ((phys_offset - (start_addr - mem->start_addr)) ==
34fc643f AL	457	mem->phys_offset)
62d60e8c AL	458	return;
	459
	460	/* unregister whole slot */
	461	memcpy(&slot, mem, sizeof(slot));
	462	mem->memory_size = 0;
34fc643f	463	kvm_set_user_memory_region(s, mem);
62d60e8c AL	464
62d60e8c AL	465	/* register prefix slot */
34fc643f AL	466	mem_start = slot.start_addr;
	467	mem_size = start_addr - slot.start_addr;
	468	mem_offset = slot.phys_offset;
62d60e8c AL	469	if (mem_size)
	470	kvm_set_phys_mem(mem_start, mem_size, mem_offset);
	471
	472	/* register new slot */
	473	kvm_set_phys_mem(start_addr, size, phys_offset);
	474
	475	/* register suffix slot */
	476	mem_start = start_addr + size;
	477	mem_offset += mem_size + size;
	478	mem_size = slot.memory_size - mem_size - size;
	479	if (mem_size)
	480	kvm_set_phys_mem(mem_start, mem_size, mem_offset);
	481
05330448	482	return;
62d60e8c AL	483	} else {
	484	printf("Registering overlapping slot\n");
	485	abort();
	486	}
05330448	487	}
05330448 AL	488	/* KVM does not need to know about this memory */
	489	if (flags >= IO_MEM_UNASSIGNED)
	490	return;
	491
	492	mem = kvm_alloc_slot(s);
	493	mem->memory_size = size;
34fc643f AL	494	mem->start_addr = start_addr;
34fc643f AL	495	mem->phys_offset = phys_offset;
05330448 AL	496	mem->flags = 0;
05330448 AL	497
34fc643f	498	kvm_set_user_memory_region(s, mem);
05330448 AL	499	/* FIXME deal with errors */
	500	}
	501
984b5181	502	int kvm_ioctl(KVMState *s, int type, ...)
05330448 AL	503	{
05330448 AL	504	int ret;
984b5181 AL	505	void *arg;
984b5181 AL	506	va_list ap;
05330448	507
984b5181 AL	508	va_start(ap, type);
	509	arg = va_arg(ap, void *);
	510	va_end(ap);
	511
	512	ret = ioctl(s->fd, type, arg);
05330448 AL	513	if (ret == -1)
	514	ret = -errno;
	515
	516	return ret;
	517	}
	518
984b5181	519	int kvm_vm_ioctl(KVMState *s, int type, ...)
05330448 AL	520	{
05330448 AL	521	int ret;
984b5181 AL	522	void *arg;
	523	va_list ap;
	524
	525	va_start(ap, type);
	526	arg = va_arg(ap, void *);
	527	va_end(ap);
05330448	528
984b5181	529	ret = ioctl(s->vmfd, type, arg);
05330448 AL	530	if (ret == -1)
	531	ret = -errno;
	532
	533	return ret;
	534	}
	535
984b5181	536	int kvm_vcpu_ioctl(CPUState *env, int type, ...)
05330448 AL	537	{
05330448 AL	538	int ret;
984b5181 AL	539	void *arg;
	540	va_list ap;
	541
	542	va_start(ap, type);
	543	arg = va_arg(ap, void *);
	544	va_end(ap);
05330448	545
984b5181	546	ret = ioctl(env->kvm_fd, type, arg);
05330448 AL	547	if (ret == -1)
	548	ret = -errno;
	549
	550	return ret;
	551	}