/*
 * 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 "cpu.h"
#include "qemu-common.h"
#include "sysemu/sysemu.h"
#include "hw/hw.h"
#include "android/kvm.h"
#include "exec/gdbstub.h"
#include "exec/ram_addr.h"
#include "sysemu/kvm.h"

#include "android/crashreport/crash-handler.h"

/* KVM uses PAGE_SIZE in it's definition of COALESCED_MMIO_MAX */
#define PAGE_SIZE TARGET_PAGE_SIZE

//#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

static void __attribute__((noreturn)) die(const char* message, const char* function) {
    crashhandler_die_format("KVM operation failed: %s (%s)", message, function);
    abort(); // just in case
}

typedef struct KVMSlot
{
    hwaddr 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;
    int coalesced_mmio;
    int broken_set_mem_region;
    int migration_log;
#ifdef KVM_CAP_SET_GUEST_DEBUG
    struct kvm_sw_breakpoint_head kvm_sw_breakpoints;
#endif
};

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];
    }

    die("no free slot available", __func__);
}

static KVMSlot *kvm_lookup_matching_slot(KVMState *s,
                                         hwaddr start_addr,
                                         ram_addr_t size)
{
    int i;

    for (i = 0; i < ARRAY_SIZE(s->slots); i++) {
        KVMSlot *mem = &s->slots[i];
        if (start_addr == mem->start_addr && size == mem->memory_size) {
            return mem;
        }
    }

    return NULL;
}

/*
 * Find overlapping slot with lowest start address
 */
static KVMSlot *kvm_lookup_overlapping_slot(KVMState *s,
                                            hwaddr start_addr,
                                            ram_addr_t size)
{
    KVMSlot *found = NULL;
    int i;

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

        // Skip empty slots.
        if (!mem->memory_size)
            continue;

        // Skip non-overlapping slots, conditions are:
        //    start_addr + size <= mem->start_addr ||
        //    start_addr >= mem->start_addr + mem->memory_size
        //
        // However, we want to avoid wrapping errors, so avoid
        // additions and only compare positive values.
        if (start_addr <= mem->start_addr) {
            if (mem->start_addr - start_addr >= size) {
                continue;
            }
        } else if (start_addr - mem->start_addr >= mem->memory_size) {
            continue;
        }

        if (found && found->start_addr < mem->start_addr) {
            continue;
        }

        found = mem;
    }

    return found;
}

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)qemu_get_ram_ptr(slot->phys_offset);
    mem.flags = slot->flags;
    if (s->migration_log) {
        mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
    }
    return kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
}


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

    dprintf("kvm_init_vcpu\n");

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

    cpu->kvm_fd = ret;
    cpu->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;
    }

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

    ret = kvm_arch_init_vcpu(cpu);

err:
    return ret;
}

int kvm_put_mp_state(CPUState *cpu)
{
    CPUArchState *env = cpu->env_ptr;
    struct kvm_mp_state mp_state = { .mp_state = env->mp_state };

    return kvm_vcpu_ioctl(cpu, KVM_SET_MP_STATE, &mp_state);
}

int kvm_get_mp_state(CPUState *cpu)
{
    struct kvm_mp_state mp_state;
    int ret;

    ret = kvm_vcpu_ioctl(cpu, KVM_GET_MP_STATE, &mp_state);
    if (ret < 0) {
        return ret;
    }
    CPUArchState *env = cpu->env_ptr;
    env->mp_state = mp_state.mp_state;
    return 0;
}

int kvm_sync_vcpus(void)
{
    CPUState *cpu;

    CPU_FOREACH(cpu) {
        int ret = kvm_arch_put_registers(cpu);
        if (ret)
            return ret;
    }

    return 0;
}

/*
 * dirty pages logging control
 */
static int kvm_dirty_pages_log_change(hwaddr phys_addr,
                                      ram_addr_t size, int flags, int mask)
{
    KVMState *s = kvm_state;
    KVMSlot *mem = kvm_lookup_matching_slot(s, phys_addr, size);
    int old_flags;

    if (mem == NULL)  {
            fprintf(stderr, "BUG: %s: invalid parameters " TARGET_FMT_plx "-"
                    TARGET_FMT_plx "\n", __func__, (hwaddr)phys_addr,
                    (hwaddr)(phys_addr + size - 1));
            return -EINVAL;
    }

    old_flags = mem->flags;

    flags = (mem->flags & ~mask) | flags;
    mem->flags = flags;

    /* If nothing changed effectively, no need to issue ioctl */
    if (s->migration_log) {
        flags |= KVM_MEM_LOG_DIRTY_PAGES;
    }
    if (flags == old_flags) {
            return 0;
    }

    return kvm_set_user_memory_region(s, mem);
}

int kvm_log_start(hwaddr phys_addr, ram_addr_t size)
{
        return kvm_dirty_pages_log_change(phys_addr, size,
                                          KVM_MEM_LOG_DIRTY_PAGES,
                                          KVM_MEM_LOG_DIRTY_PAGES);
}

int kvm_log_stop(hwaddr phys_addr, ram_addr_t size)
{
        return kvm_dirty_pages_log_change(phys_addr, size,
                                          0,
                                          KVM_MEM_LOG_DIRTY_PAGES);
}

int kvm_set_migration_log(int enable)
{
    KVMState *s = kvm_state;
    KVMSlot *mem;
    int i, err;

    s->migration_log = enable;

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

        if (!!(mem->flags & KVM_MEM_LOG_DIRTY_PAGES) == enable) {
            continue;
        }
        err = kvm_set_user_memory_region(s, mem);
        if (err) {
            return err;
        }
    }
    return 0;
}

#define ALIGN(x, y)  (((x)+(y)-1) & ~((y)-1))

/**
 * 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.
 * @end_addr: end of logged region.
 */
int kvm_physical_sync_dirty_bitmap(hwaddr start_addr,
                                   hwaddr end_addr)
{
    KVMState *s = kvm_state;
    size_t size, allocated_size = 0;
    KVMDirtyLog d;
    KVMSlot *mem;
    int ret = 0;

    d.dirty_bitmap = NULL;
    while (start_addr < end_addr) {
        ram_addr_t start_size = (ram_addr_t)(end_addr - start_addr);

        mem = kvm_lookup_overlapping_slot(s, start_addr, start_size);
        if (mem == NULL) {
            break;
        }

        size = ALIGN(((mem->memory_size) >> TARGET_PAGE_BITS),
                     /*HOST_LONG_BITS*/ 64) / 8;
        if (size > allocated_size) {
            d.dirty_bitmap = g_realloc(d.dirty_bitmap, size);
            allocated_size = size;
        }
        memset(d.dirty_bitmap, 0, allocated_size);

        d.slot = mem->slot;

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

        cpu_physical_memory_set_dirty_lebitmap(d.dirty_bitmap,
                                               mem->phys_offset,
                                               mem->memory_size / getpagesize());
        start_addr += mem->memory_size;
        if (!start_addr) {
            // Handle wrap-around, which happens when a slot is mapped
            // at the end of the physical address space.
            break;
        }
    }
    g_free(d.dirty_bitmap);

    return ret;
}

int kvm_coalesce_mmio_region(hwaddr start, ram_addr_t size)
{
    int ret = -ENOSYS;
#ifdef KVM_CAP_COALESCED_MMIO
    KVMState *s = kvm_state;

    if (s->coalesced_mmio) {
        struct kvm_coalesced_mmio_zone zone;

        zone.addr = start;
        zone.size = size;

        ret = kvm_vm_ioctl(s, KVM_REGISTER_COALESCED_MMIO, &zone);
    }
#endif

    return ret;
}

int kvm_uncoalesce_mmio_region(hwaddr start, ram_addr_t size)
{
    int ret = -ENOSYS;
#ifdef KVM_CAP_COALESCED_MMIO
    KVMState *s = kvm_state;

    if (s->coalesced_mmio) {
        struct kvm_coalesced_mmio_zone zone;

        zone.addr = start;
        zone.size = size;

        ret = kvm_vm_ioctl(s, KVM_UNREGISTER_COALESCED_MMIO, &zone);
    }
#endif

    return ret;
}

int kvm_check_extension(KVMState *s, unsigned int extension)
{
    int ret;

    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, extension);
    if (ret < 0) {
        ret = 0;
    }

    return ret;
}

static void kvm_reset_vcpus(void *opaque)
{
    kvm_sync_vcpus();
}

int kvm_init(int smp_cpus)
{
    static const char upgrade_note[] =
        "Please upgrade to at least kernel 2.6.29 or recent kvm-kmod\n"
        "(see http://sourceforge.net/projects/kvm).\n";
    KVMState *s;
    int ret;
    int i;

    if (smp_cpus > 1) {
        fprintf(stderr, "No SMP KVM support, use '-smp 1'\n");
        return -EINVAL;
    }

    s = g_malloc0(sizeof(KVMState));

#ifdef KVM_CAP_SET_GUEST_DEBUG
    QTAILQ_INIT(&s->kvm_sw_breakpoints);
#endif
    for (i = 0; i < ARRAY_SIZE(s->slots); i++)
        s->slots[i].slot = i;

    char* kvm_device = getenv(KVM_DEVICE_NAME_ENV);
    if (NULL == kvm_device) {
      kvm_device = "/dev/kvm";
    }

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

    fcntl(s->fd, F_SETFD, FD_CLOEXEC);

    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;
    }

    do {
      s->vmfd = kvm_ioctl(s, KVM_CREATE_VM, 0);
    } while (s->vmfd < 0 && (EINTR == errno || EAGAIN == errno));

    if (s->vmfd < 0) {
        ret = -errno;
        fprintf(stderr, "ioctl(KVM_CREATE_VM) failed: %d %s\n", errno, strerror(errno));
        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 regular pages
     * unmodified.  Make sure we have a sufficiently modern version of KVM.
     */
    if (!kvm_check_extension(s, KVM_CAP_USER_MEMORY)) {
        ret = -EINVAL;
        fprintf(stderr, "kvm does not support KVM_CAP_USER_MEMORY\n%s",
                upgrade_note);
        goto err;
    }

    /* There was a nasty bug in < kvm-80 that prevents memory slots from being
     * destroyed properly.  Since we rely on this capability, refuse to work
     * with any kernel without this capability. */
    if (!kvm_check_extension(s, KVM_CAP_DESTROY_MEMORY_REGION_WORKS)) {
        ret = -EINVAL;

        fprintf(stderr,
                "KVM kernel module broken (DESTROY_MEMORY_REGION).\n%s",
                upgrade_note);
        goto err;
    }

#ifdef KVM_CAP_COALESCED_MMIO
    s->coalesced_mmio = kvm_check_extension(s, KVM_CAP_COALESCED_MMIO);
#else
    s->coalesced_mmio = 0;
#endif

    s->broken_set_mem_region = 1;
#ifdef KVM_CAP_JOIN_MEMORY_REGIONS_WORKS
    ret = kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_JOIN_MEMORY_REGIONS_WORKS);
    if (ret > 0) {
        s->broken_set_mem_region = 0;
    }
#endif

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

    qemu_register_reset(kvm_reset_vcpus, INT_MAX, NULL);

    kvm_state = s;

    return 0;

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

    return ret;
}

static int kvm_handle_io(CPUState *cpu, 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(port));
                break;
            case 2:
                stw_p(ptr, cpu_inw(port));
                break;
            case 4:
                stl_p(ptr, cpu_inl(port));
                break;
            }
        } else {
            switch (size) {
            case 1:
                cpu_outb(port, ldub_p(ptr));
                break;
            case 2:
                cpu_outw(port, lduw_p(ptr));
                break;
            case 4:
                cpu_outl(port, ldl_p(ptr));
                break;
            }
        }

        ptr += size;
    }

    return 1;
}

static void kvm_run_coalesced_mmio(CPUState *cpu, struct kvm_run *run)
{
#ifdef KVM_CAP_COALESCED_MMIO
    KVMState *s = kvm_state;
    if (s->coalesced_mmio) {
        struct kvm_coalesced_mmio_ring *ring;

        ring = (void *)run + (s->coalesced_mmio * TARGET_PAGE_SIZE);
        while (ring->first != ring->last) {
            struct kvm_coalesced_mmio *ent;

            ent = &ring->coalesced_mmio[ring->first];

            cpu_physical_memory_write(ent->phys_addr, ent->data, ent->len);
            /* FIXME smp_wmb() */
            ring->first = (ring->first + 1) % KVM_COALESCED_MMIO_MAX;
        }
    }
#endif
}

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

    dprintf("kvm_cpu_exec()\n");

    do {
        if (cpu->exit_request) {
            dprintf("interrupt exit requested\n");
            ret = 0;
            break;
        }

        kvm_arch_pre_run(cpu, run);
        ret = kvm_arch_vcpu_run(cpu);
        kvm_arch_post_run(cpu, 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));
            die("kvm vcpu run failed", __func__);
        }

        kvm_run_coalesced_mmio(cpu, run);

        ret = 0; /* exit loop */
        switch (run->exit_reason) {
        case KVM_EXIT_IO:
            dprintf("handle_io\n");
            ret = kvm_handle_io(cpu, 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");
#ifdef KVM_CAP_SET_GUEST_DEBUG
            if (kvm_arch_debug(&run->debug.arch)) {
                gdb_set_stop_cpu(cpu);
                vm_stop(EXCP_DEBUG);\
                env->exception_index = EXCP_DEBUG;
                return 0;
            }
            /* re-enter, this exception was guest-internal */
            ret = 1;
#endif /* KVM_CAP_SET_GUEST_DEBUG */
            break;
        default:
            dprintf("kvm_arch_handle_exit\n");
            ret = kvm_arch_handle_exit(cpu, run);
            break;
        }
    } while (ret > 0);

    if (cpu->exit_request) {
        cpu->exit_request = 0;
        env->exception_index = EXCP_INTERRUPT;
    }

    return ret;
}

void kvm_set_phys_mem(hwaddr 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, old;
    int err;

    if (start_addr & ~TARGET_PAGE_MASK) {
        if (flags >= IO_MEM_UNASSIGNED) {
            if (!kvm_lookup_overlapping_slot(s, start_addr, size)) {
                return;
            }
            fprintf(stderr, "Unaligned split of a KVM memory slot\n");
        } else {
            fprintf(stderr, "Only page-aligned memory slots supported\n");
        }
        die("bad start address", __func__);
    }

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

    while (1) {
        mem = kvm_lookup_overlapping_slot(s, start_addr, size);
        if (!mem) {
            break;
        }

        if (flags < IO_MEM_UNASSIGNED && start_addr >= mem->start_addr &&
            (start_addr + size <= mem->start_addr + mem->memory_size) &&
            (phys_offset - start_addr == mem->phys_offset - mem->start_addr)) {
            /* The new slot fits into the existing one and comes with
             * identical parameters - nothing to be done. */
            return;
        }

        old = *mem;

        /* unregister the overlapping slot */
        mem->memory_size = 0;
        err = kvm_set_user_memory_region(s, mem);
        if (err) {
            fprintf(stderr, "%s: error unregistering overlapping slot: %s\n",
                    __func__, strerror(-err));
            die("error unregistering overlapping slot", __func__);
        }

        /* Workaround for older KVM versions: we can't join slots, even not by
         * unregistering the previous ones and then registering the larger
         * slot. We have to maintain the existing fragmentation. Sigh.
         *
         * This workaround assumes that the new slot starts at the same
         * address as the first existing one. If not or if some overlapping
         * slot comes around later, we will fail (not seen in practice so far)
         * - and actually require a recent KVM version. */
        if (s->broken_set_mem_region &&
            old.start_addr == start_addr && old.memory_size < size &&
            flags < IO_MEM_UNASSIGNED) {
            mem = kvm_alloc_slot(s);
            mem->memory_size = old.memory_size;
            mem->start_addr = old.start_addr;
            mem->phys_offset = old.phys_offset;
            mem->flags = 0;

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error updating slot: %s\n", __func__,
                        strerror(-err));
                die("failed to set user memory region "
                    "(most probably out of memory)", __func__);
            }

            start_addr += old.memory_size;
            phys_offset += old.memory_size;
            size -= old.memory_size;
            continue;
        }

        /* register prefix slot */
        if (old.start_addr < start_addr) {
            mem = kvm_alloc_slot(s);
            mem->memory_size = start_addr - old.start_addr;
            mem->start_addr = old.start_addr;
            mem->phys_offset = old.phys_offset;
            mem->flags = 0;

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error registering prefix slot: %s\n",
                        __func__, strerror(-err));
                die("failed to set user memory region "
                    "(most probably out of memory)", __func__);
            }
        }

        /* register suffix slot */
        if (old.start_addr + old.memory_size > start_addr + size) {
            ram_addr_t size_delta;

            mem = kvm_alloc_slot(s);
            mem->start_addr = start_addr + size;
            size_delta = mem->start_addr - old.start_addr;
            mem->memory_size = old.memory_size - size_delta;
            mem->phys_offset = old.phys_offset + size_delta;
            mem->flags = 0;

            err = kvm_set_user_memory_region(s, mem);
            if (err) {
                fprintf(stderr, "%s: error registering suffix slot: %s\n",
                        __func__, strerror(-err));
                die("failed to set user memory region "
                    "(most probably out of memory)", __func__);
            }
        }
    }

    /* in case the KVM bug workaround already "consumed" the new slot */
    if (!size)
        return;

    /* 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;

    err = kvm_set_user_memory_region(s, mem);
    if (err) {
        fprintf(stderr, "%s: error registering slot: %s\n", __func__,
                strerror(-err));
        die("Failed to set user memory region (most probably out of memory)",
            __func__);
    }
}

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 *cpu, int type, ...)
{
    int ret;
    void *arg;
    va_list ap;

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

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

    return ret;
}

int kvm_has_sync_mmu(void)
{
#ifdef KVM_CAP_SYNC_MMU
    KVMState *s = kvm_state;

    return kvm_check_extension(s, KVM_CAP_SYNC_MMU);
#else
    return 0;
#endif
}

void kvm_setup_guest_memory(void *start, size_t size)
{
    if (!kvm_has_sync_mmu()) {
#ifdef MADV_DONTFORK
        int ret = madvise(start, size, MADV_DONTFORK);

        if (ret) {
            perror("madvice");
            exit(1);
        }
#else
        fprintf(stderr,
                "Need MADV_DONTFORK in absence of synchronous KVM MMU\n");
        exit(1);
#endif
    }
}

#ifdef KVM_CAP_SET_GUEST_DEBUG
struct kvm_sw_breakpoint *kvm_find_sw_breakpoint(CPUState *cpu,
                                                 target_ulong pc)
{
    struct kvm_sw_breakpoint *bp;

    QTAILQ_FOREACH(bp, &cpu->kvm_state->kvm_sw_breakpoints, entry) {
        if (bp->pc == pc)
            return bp;
    }
    return NULL;
}

int kvm_sw_breakpoints_active(CPUState *cpu)
{
    return !QTAILQ_EMPTY(&cpu->kvm_state->kvm_sw_breakpoints);
}

int kvm_update_guest_debug(CPUState *cpu, unsigned long reinject_trap)
{
    struct kvm_guest_debug dbg;

    dbg.control = 0;
    if (cpu->singlestep_enabled)
        dbg.control = KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_SINGLESTEP;

    kvm_arch_update_guest_debug(cpu, &dbg);
    dbg.control |= reinject_trap;

    return kvm_vcpu_ioctl(cpu, KVM_SET_GUEST_DEBUG, &dbg);
}

int kvm_insert_breakpoint(CPUState *cpu, target_ulong addr,
                          target_ulong len, int type)
{
    struct kvm_sw_breakpoint *bp;
    int err;

    if (type == GDB_BREAKPOINT_SW) {
        bp = kvm_find_sw_breakpoint(cpu, addr);
        if (bp) {
            bp->use_count++;
            return 0;
        }

        bp = g_malloc(sizeof(struct kvm_sw_breakpoint));
        if (!bp)
            return -ENOMEM;

        bp->pc = addr;
        bp->use_count = 1;
        err = kvm_arch_insert_sw_breakpoint(cpu, bp);
        if (err) {
            free(bp);
            return err;
        }

        QTAILQ_INSERT_HEAD(&cpu->kvm_state->kvm_sw_breakpoints,
                          bp, entry);
    } else {
        err = kvm_arch_insert_hw_breakpoint(addr, len, type);
        if (err)
            return err;
    }

    CPU_FOREACH(cpu) {
        err = kvm_update_guest_debug(cpu, 0);
        if (err)
            return err;
    }
    return 0;
}

int kvm_remove_breakpoint(CPUState *cpu, target_ulong addr,
                          target_ulong len, int type)
{
    struct kvm_sw_breakpoint *bp;
    int err;

    if (type == GDB_BREAKPOINT_SW) {
        bp = kvm_find_sw_breakpoint(cpu, addr);
        if (!bp)
            return -ENOENT;

        if (bp->use_count > 1) {
            bp->use_count--;
            return 0;
        }

        err = kvm_arch_remove_sw_breakpoint(cpu, bp);
        if (err)
            return err;

        QTAILQ_REMOVE(&cpu->kvm_state->kvm_sw_breakpoints, bp, entry);
        g_free(bp);
    } else {
        err = kvm_arch_remove_hw_breakpoint(addr, len, type);
        if (err)
            return err;
    }

    CPU_FOREACH(cpu) {
        err = kvm_update_guest_debug(cpu, 0);
        if (err)
            return err;
    }
    return 0;
}

void kvm_remove_all_breakpoints(CPUState *cpu)
{
    struct kvm_sw_breakpoint *bp, *next;
    KVMState *s = cpu->kvm_state;

    QTAILQ_FOREACH_SAFE(bp, &s->kvm_sw_breakpoints, entry, next) {
        if (kvm_arch_remove_sw_breakpoint(cpu, bp) != 0) {
            /* Try harder to find a CPU that currently sees the breakpoint. */
            CPU_FOREACH(cpu) {
                if (kvm_arch_remove_sw_breakpoint(cpu, bp) == 0)
                    break;
            }
        }
    }
    kvm_arch_remove_all_hw_breakpoints();

    CPU_FOREACH(cpu) {
        kvm_update_guest_debug(cpu, 0);
    }
}

#else /* !KVM_CAP_SET_GUEST_DEBUG */

int kvm_update_guest_debug(CPUState *cpu, unsigned long reinject_trap)
{
    return -EINVAL;
}

int kvm_insert_breakpoint(CPUState *cpu, target_ulong addr,
                          target_ulong len, int type)
{
    return -EINVAL;
}

int kvm_remove_breakpoint(CPUState *cpu, target_ulong addr,
                          target_ulong len, int type)
{
    return -EINVAL;
}

void kvm_remove_all_breakpoints(CPUState *cpu)
{
}
#endif /* !KVM_CAP_SET_GUEST_DEBUG */