[qemu.git] / hw / vhost.c

/*
 * vhost support
 *
 * Copyright Red Hat, Inc. 2010
 *
 * Authors:
 *  Michael S. Tsirkin <[email protected]>
 *
 * This work is licensed under the terms of the GNU GPL, version 2.  See
 * the COPYING file in the top-level directory.
 */

#include <sys/ioctl.h>
#include "vhost.h"
#include "hw/hw.h"
#include "range.h"
#include <linux/vhost.h>

static void vhost_dev_sync_region(struct vhost_dev *dev,
                                  uint64_t mfirst, uint64_t mlast,
                                  uint64_t rfirst, uint64_t rlast)
{
    uint64_t start = MAX(mfirst, rfirst);
    uint64_t end = MIN(mlast, rlast);
    vhost_log_chunk_t *from = dev->log + start / VHOST_LOG_CHUNK;
    vhost_log_chunk_t *to = dev->log + end / VHOST_LOG_CHUNK + 1;
    uint64_t addr = (start / VHOST_LOG_CHUNK) * VHOST_LOG_CHUNK;

    assert(end / VHOST_LOG_CHUNK < dev->log_size);
    assert(start / VHOST_LOG_CHUNK < dev->log_size);
    if (end < start) {
        return;
    }
    for (;from < to; ++from) {
        vhost_log_chunk_t log;
        int bit;
        /* We first check with non-atomic: much cheaper,
         * and we expect non-dirty to be the common case. */
        if (!*from) {
            continue;
        }
        /* Data must be read atomically. We don't really
         * need the barrier semantics of __sync
         * builtins, but it's easier to use them than
         * roll our own. */
        log = __sync_fetch_and_and(from, 0);
        while ((bit = sizeof(log) > sizeof(int) ?
                ffsll(log) : ffs(log))) {
            bit -= 1;
            cpu_physical_memory_set_dirty(addr + bit * VHOST_LOG_PAGE);
            log &= ~(0x1ull << bit);
        }
        addr += VHOST_LOG_CHUNK;
    }
}

static int vhost_client_sync_dirty_bitmap(CPUPhysMemoryClient *client,
                                          target_phys_addr_t start_addr,
                                          target_phys_addr_t end_addr)
{
    struct vhost_dev *dev = container_of(client, struct vhost_dev, client);
    int i;
    if (!dev->log_enabled || !dev->started) {
        return 0;
    }
    for (i = 0; i < dev->mem->nregions; ++i) {
        struct vhost_memory_region *reg = dev->mem->regions + i;
        vhost_dev_sync_region(dev, start_addr, end_addr,
                              reg->guest_phys_addr,
                              range_get_last(reg->guest_phys_addr,
                                             reg->memory_size));
    }
    for (i = 0; i < dev->nvqs; ++i) {
        struct vhost_virtqueue *vq = dev->vqs + i;
        vhost_dev_sync_region(dev, start_addr, end_addr, vq->used_phys,
                              range_get_last(vq->used_phys, vq->used_size));
    }
    return 0;
}

/* Assign/unassign. Keep an unsorted array of non-overlapping
 * memory regions in dev->mem. */
static void vhost_dev_unassign_memory(struct vhost_dev *dev,
                                      uint64_t start_addr,
                                      uint64_t size)
{
    int from, to, n = dev->mem->nregions;
    /* Track overlapping/split regions for sanity checking. */
    int overlap_start = 0, overlap_end = 0, overlap_middle = 0, split = 0;

    for (from = 0, to = 0; from < n; ++from, ++to) {
        struct vhost_memory_region *reg = dev->mem->regions + to;
        uint64_t reglast;
        uint64_t memlast;
        uint64_t change;

        /* clone old region */
        if (to != from) {
            memcpy(reg, dev->mem->regions + from, sizeof *reg);
        }

        /* No overlap is simple */
        if (!ranges_overlap(reg->guest_phys_addr, reg->memory_size,
                            start_addr, size)) {
            continue;
        }

        /* Split only happens if supplied region
         * is in the middle of an existing one. Thus it can not
         * overlap with any other existing region. */
        assert(!split);

        reglast = range_get_last(reg->guest_phys_addr, reg->memory_size);
        memlast = range_get_last(start_addr, size);

        /* Remove whole region */
        if (start_addr <= reg->guest_phys_addr && memlast >= reglast) {
            --dev->mem->nregions;
            --to;
            assert(to >= 0);
            ++overlap_middle;
            continue;
        }

        /* Shrink region */
        if (memlast >= reglast) {
            reg->memory_size = start_addr - reg->guest_phys_addr;
            assert(reg->memory_size);
            assert(!overlap_end);
            ++overlap_end;
            continue;
        }

        /* Shift region */
        if (start_addr <= reg->guest_phys_addr) {
            change = memlast + 1 - reg->guest_phys_addr;
            reg->memory_size -= change;
            reg->guest_phys_addr += change;
            reg->userspace_addr += change;
            assert(reg->memory_size);
            assert(!overlap_start);
            ++overlap_start;
            continue;
        }

        /* This only happens if supplied region
         * is in the middle of an existing one. Thus it can not
         * overlap with any other existing region. */
        assert(!overlap_start);
        assert(!overlap_end);
        assert(!overlap_middle);
        /* Split region: shrink first part, shift second part. */
        memcpy(dev->mem->regions + n, reg, sizeof *reg);
        reg->memory_size = start_addr - reg->guest_phys_addr;
        assert(reg->memory_size);
        change = memlast + 1 - reg->guest_phys_addr;
        reg = dev->mem->regions + n;
        reg->memory_size -= change;
        assert(reg->memory_size);
        reg->guest_phys_addr += change;
        reg->userspace_addr += change;
        /* Never add more than 1 region */
        assert(dev->mem->nregions == n);
        ++dev->mem->nregions;
        ++split;
    }
}

/* Called after unassign, so no regions overlap the given range. */
static void vhost_dev_assign_memory(struct vhost_dev *dev,
                                    uint64_t start_addr,
                                    uint64_t size,
                                    uint64_t uaddr)
{
    int from, to;
    struct vhost_memory_region *merged = NULL;
    for (from = 0, to = 0; from < dev->mem->nregions; ++from, ++to) {
        struct vhost_memory_region *reg = dev->mem->regions + to;
        uint64_t prlast, urlast;
        uint64_t pmlast, umlast;
        uint64_t s, e, u;

        /* clone old region */
        if (to != from) {
            memcpy(reg, dev->mem->regions + from, sizeof *reg);
        }
        prlast = range_get_last(reg->guest_phys_addr, reg->memory_size);
        pmlast = range_get_last(start_addr, size);
        urlast = range_get_last(reg->userspace_addr, reg->memory_size);
        umlast = range_get_last(uaddr, size);

        /* check for overlapping regions: should never happen. */
        assert(prlast < start_addr || pmlast < reg->guest_phys_addr);
        /* Not an adjacent or overlapping region - do not merge. */
        if ((prlast + 1 != start_addr || urlast + 1 != uaddr) &&
            (pmlast + 1 != reg->guest_phys_addr ||
             umlast + 1 != reg->userspace_addr)) {
            continue;
        }

        if (merged) {
            --to;
            assert(to >= 0);
        } else {
            merged = reg;
        }
        u = MIN(uaddr, reg->userspace_addr);
        s = MIN(start_addr, reg->guest_phys_addr);
        e = MAX(pmlast, prlast);
        uaddr = merged->userspace_addr = u;
        start_addr = merged->guest_phys_addr = s;
        size = merged->memory_size = e - s + 1;
        assert(merged->memory_size);
    }

    if (!merged) {
        struct vhost_memory_region *reg = dev->mem->regions + to;
        memset(reg, 0, sizeof *reg);
        reg->memory_size = size;
        assert(reg->memory_size);
        reg->guest_phys_addr = start_addr;
        reg->userspace_addr = uaddr;
        ++to;
    }
    assert(to <= dev->mem->nregions + 1);
    dev->mem->nregions = to;
}

static uint64_t vhost_get_log_size(struct vhost_dev *dev)
{
    uint64_t log_size = 0;
    int i;
    for (i = 0; i < dev->mem->nregions; ++i) {
        struct vhost_memory_region *reg = dev->mem->regions + i;
        uint64_t last = range_get_last(reg->guest_phys_addr,
                                       reg->memory_size);
        log_size = MAX(log_size, last / VHOST_LOG_CHUNK + 1);
    }
    for (i = 0; i < dev->nvqs; ++i) {
        struct vhost_virtqueue *vq = dev->vqs + i;
        uint64_t last = vq->used_phys + vq->used_size - 1;
        log_size = MAX(log_size, last / VHOST_LOG_CHUNK + 1);
    }
    return log_size;
}

static inline void vhost_dev_log_resize(struct vhost_dev* dev, uint64_t size)
{
    vhost_log_chunk_t *log;
    uint64_t log_base;
    int r;
    if (size) {
        log = qemu_mallocz(size * sizeof *log);
    } else {
        log = NULL;
    }
    log_base = (uint64_t)(unsigned long)log;
    r = ioctl(dev->control, VHOST_SET_LOG_BASE, &log_base);
    assert(r >= 0);
    vhost_client_sync_dirty_bitmap(&dev->client, 0,
                                   (target_phys_addr_t)~0x0ull);
    if (dev->log) {
        qemu_free(dev->log);
    }
    dev->log = log;
    dev->log_size = size;
}

static int vhost_verify_ring_mappings(struct vhost_dev *dev,
                                      uint64_t start_addr,
                                      uint64_t size)
{
    int i;
    for (i = 0; i < dev->nvqs; ++i) {
        struct vhost_virtqueue *vq = dev->vqs + i;
        target_phys_addr_t l;
        void *p;

        if (!ranges_overlap(start_addr, size, vq->ring_phys, vq->ring_size)) {
            continue;
        }
        l = vq->ring_size;
        p = cpu_physical_memory_map(vq->ring_phys, &l, 1);
        if (!p || l != vq->ring_size) {
            fprintf(stderr, "Unable to map ring buffer for ring %d\n", i);
            return -ENOMEM;
        }
        if (p != vq->ring) {
            fprintf(stderr, "Ring buffer relocated for ring %d\n", i);
            return -EBUSY;
        }
        cpu_physical_memory_unmap(p, l, 0, 0);
    }
    return 0;
}

static void vhost_client_set_memory(CPUPhysMemoryClient *client,
                                    target_phys_addr_t start_addr,
                                    ram_addr_t size,
                                    ram_addr_t phys_offset)
{
    struct vhost_dev *dev = container_of(client, struct vhost_dev, client);
    ram_addr_t flags = phys_offset & ~TARGET_PAGE_MASK;
    int s = offsetof(struct vhost_memory, regions) +
        (dev->mem->nregions + 1) * sizeof dev->mem->regions[0];
    uint64_t log_size;
    int r;
    dev->mem = qemu_realloc(dev->mem, s);

    assert(size);

    vhost_dev_unassign_memory(dev, start_addr, size);
    if (flags == IO_MEM_RAM) {
        /* Add given mapping, merging adjacent regions if any */
        vhost_dev_assign_memory(dev, start_addr, size,
                                (uintptr_t)qemu_get_ram_ptr(phys_offset));
    } else {
        /* Remove old mapping for this memory, if any. */
        vhost_dev_unassign_memory(dev, start_addr, size);
    }

    if (!dev->started) {
        return;
    }

    if (dev->started) {
        r = vhost_verify_ring_mappings(dev, start_addr, size);
        assert(r >= 0);
    }

    if (!dev->log_enabled) {
        r = ioctl(dev->control, VHOST_SET_MEM_TABLE, dev->mem);
        assert(r >= 0);
        return;
    }
    log_size = vhost_get_log_size(dev);
    /* We allocate an extra 4K bytes to log,
     * to reduce the * number of reallocations. */
#define VHOST_LOG_BUFFER (0x1000 / sizeof *dev->log)
    /* To log more, must increase log size before table update. */
    if (dev->log_size < log_size) {
        vhost_dev_log_resize(dev, log_size + VHOST_LOG_BUFFER);
    }
    r = ioctl(dev->control, VHOST_SET_MEM_TABLE, dev->mem);
    assert(r >= 0);
    /* To log less, can only decrease log size after table update. */
    if (dev->log_size > log_size + VHOST_LOG_BUFFER) {
        vhost_dev_log_resize(dev, log_size);
    }
}

static int vhost_virtqueue_set_addr(struct vhost_dev *dev,
                                    struct vhost_virtqueue *vq,
                                    unsigned idx, bool enable_log)
{
    struct vhost_vring_addr addr = {
        .index = idx,
        .desc_user_addr = (uint64_t)(unsigned long)vq->desc,
        .avail_user_addr = (uint64_t)(unsigned long)vq->avail,
        .used_user_addr = (uint64_t)(unsigned long)vq->used,
        .log_guest_addr = vq->used_phys,
        .flags = enable_log ? (1 << VHOST_VRING_F_LOG) : 0,
    };
    int r = ioctl(dev->control, VHOST_SET_VRING_ADDR, &addr);
    if (r < 0) {
        return -errno;
    }
    return 0;
}

static int vhost_dev_set_features(struct vhost_dev *dev, bool enable_log)
{
    uint64_t features = dev->acked_features;
    int r;
    if (enable_log) {
        features |= 0x1 << VHOST_F_LOG_ALL;
    }
    r = ioctl(dev->control, VHOST_SET_FEATURES, &features);
    return r < 0 ? -errno : 0;
}

static int vhost_dev_set_log(struct vhost_dev *dev, bool enable_log)
{
    int r, t, i;
    r = vhost_dev_set_features(dev, enable_log);
    if (r < 0) {
        goto err_features;
    }
    for (i = 0; i < dev->nvqs; ++i) {
        r = vhost_virtqueue_set_addr(dev, dev->vqs + i, i,
                                     enable_log);
        if (r < 0) {
            goto err_vq;
        }
    }
    return 0;
err_vq:
    for (; i >= 0; --i) {
        t = vhost_virtqueue_set_addr(dev, dev->vqs + i, i,
                                     dev->log_enabled);
        assert(t >= 0);
    }
    t = vhost_dev_set_features(dev, dev->log_enabled);
    assert(t >= 0);
err_features:
    return r;
}

static int vhost_client_migration_log(CPUPhysMemoryClient *client,
                                      int enable)
{
    struct vhost_dev *dev = container_of(client, struct vhost_dev, client);
    int r;
    if (!!enable == dev->log_enabled) {
        return 0;
    }
    if (!dev->started) {
        dev->log_enabled = enable;
        return 0;
    }
    if (!enable) {
        r = vhost_dev_set_log(dev, false);
        if (r < 0) {
            return r;
        }
        if (dev->log) {
            qemu_free(dev->log);
        }
        dev->log = NULL;
        dev->log_size = 0;
    } else {
        vhost_dev_log_resize(dev, vhost_get_log_size(dev));
        r = vhost_dev_set_log(dev, true);
        if (r < 0) {
            return r;
        }
    }
    dev->log_enabled = enable;
    return 0;
}

static int vhost_virtqueue_init(struct vhost_dev *dev,
                                struct VirtIODevice *vdev,
                                struct vhost_virtqueue *vq,
                                unsigned idx)
{
    target_phys_addr_t s, l, a;
    int r;
    struct vhost_vring_file file = {
        .index = idx,
    };
    struct vhost_vring_state state = {
        .index = idx,
    };
    struct VirtQueue *vvq = virtio_get_queue(vdev, idx);

    if (!vdev->binding->set_host_notifier) {
        fprintf(stderr, "binding does not support host notifiers\n");
        return -ENOSYS;
    }

    vq->num = state.num = virtio_queue_get_num(vdev, idx);
    r = ioctl(dev->control, VHOST_SET_VRING_NUM, &state);
    if (r) {
        return -errno;
    }

    state.num = virtio_queue_get_last_avail_idx(vdev, idx);
    r = ioctl(dev->control, VHOST_SET_VRING_BASE, &state);
    if (r) {
        return -errno;
    }

    s = l = virtio_queue_get_desc_size(vdev, idx);
    a = virtio_queue_get_desc_addr(vdev, idx);
    vq->desc = cpu_physical_memory_map(a, &l, 0);
    if (!vq->desc || l != s) {
        r = -ENOMEM;
        goto fail_alloc_desc;
    }
    s = l = virtio_queue_get_avail_size(vdev, idx);
    a = virtio_queue_get_avail_addr(vdev, idx);
    vq->avail = cpu_physical_memory_map(a, &l, 0);
    if (!vq->avail || l != s) {
        r = -ENOMEM;
        goto fail_alloc_avail;
    }
    vq->used_size = s = l = virtio_queue_get_used_size(vdev, idx);
    vq->used_phys = a = virtio_queue_get_used_addr(vdev, idx);
    vq->used = cpu_physical_memory_map(a, &l, 1);
    if (!vq->used || l != s) {
        r = -ENOMEM;
        goto fail_alloc_used;
    }

    vq->ring_size = s = l = virtio_queue_get_ring_size(vdev, idx);
    vq->ring_phys = a = virtio_queue_get_ring_addr(vdev, idx);
    vq->ring = cpu_physical_memory_map(a, &l, 1);
    if (!vq->ring || l != s) {
        r = -ENOMEM;
        goto fail_alloc_ring;
    }

    r = vhost_virtqueue_set_addr(dev, vq, idx, dev->log_enabled);
    if (r < 0) {
        r = -errno;
        goto fail_alloc;
    }
    r = vdev->binding->set_host_notifier(vdev->binding_opaque, idx, true);
    if (r < 0) {
        fprintf(stderr, "Error binding host notifier: %d\n", -r);
        goto fail_host_notifier;
    }

    file.fd = event_notifier_get_fd(virtio_queue_get_host_notifier(vvq));
    r = ioctl(dev->control, VHOST_SET_VRING_KICK, &file);
    if (r) {
        r = -errno;
        goto fail_kick;
    }

    file.fd = event_notifier_get_fd(virtio_queue_get_guest_notifier(vvq));
    r = ioctl(dev->control, VHOST_SET_VRING_CALL, &file);
    if (r) {
        r = -errno;
        goto fail_call;
    }

    return 0;

fail_call:
fail_kick:
    vdev->binding->set_host_notifier(vdev->binding_opaque, idx, false);
fail_host_notifier:
fail_alloc:
    cpu_physical_memory_unmap(vq->ring, virtio_queue_get_ring_size(vdev, idx),
                              0, 0);
fail_alloc_ring:
    cpu_physical_memory_unmap(vq->used, virtio_queue_get_used_size(vdev, idx),
                              0, 0);
fail_alloc_used:
    cpu_physical_memory_unmap(vq->avail, virtio_queue_get_avail_size(vdev, idx),
                              0, 0);
fail_alloc_avail:
    cpu_physical_memory_unmap(vq->desc, virtio_queue_get_desc_size(vdev, idx),
                              0, 0);
fail_alloc_desc:
    return r;
}

static void vhost_virtqueue_cleanup(struct vhost_dev *dev,
                                    struct VirtIODevice *vdev,
                                    struct vhost_virtqueue *vq,
                                    unsigned idx)
{
    struct vhost_vring_state state = {
        .index = idx,
    };
    int r;
    r = vdev->binding->set_host_notifier(vdev->binding_opaque, idx, false);
    if (r < 0) {
        fprintf(stderr, "vhost VQ %d host cleanup failed: %d\n", idx, r);
        fflush(stderr);
    }
    assert (r >= 0);
    r = ioctl(dev->control, VHOST_GET_VRING_BASE, &state);
    if (r < 0) {
        fprintf(stderr, "vhost VQ %d ring restore failed: %d\n", idx, r);
        fflush(stderr);
    }
    virtio_queue_set_last_avail_idx(vdev, idx, state.num);
    assert (r >= 0);
    cpu_physical_memory_unmap(vq->ring, virtio_queue_get_ring_size(vdev, idx),
                              0, virtio_queue_get_ring_size(vdev, idx));
    cpu_physical_memory_unmap(vq->used, virtio_queue_get_used_size(vdev, idx),
                              1, virtio_queue_get_used_size(vdev, idx));
    cpu_physical_memory_unmap(vq->avail, virtio_queue_get_avail_size(vdev, idx),
                              0, virtio_queue_get_avail_size(vdev, idx));
    cpu_physical_memory_unmap(vq->desc, virtio_queue_get_desc_size(vdev, idx),
                              0, virtio_queue_get_desc_size(vdev, idx));
}

int vhost_dev_init(struct vhost_dev *hdev, int devfd)
{
    uint64_t features;
    int r;
    if (devfd >= 0) {
        hdev->control = devfd;
    } else {
        hdev->control = open("/dev/vhost-net", O_RDWR);
        if (hdev->control < 0) {
            return -errno;
        }
    }
    r = ioctl(hdev->control, VHOST_SET_OWNER, NULL);
    if (r < 0) {
        goto fail;
    }

    r = ioctl(hdev->control, VHOST_GET_FEATURES, &features);
    if (r < 0) {
        goto fail;
    }
    hdev->features = features;

    hdev->client.set_memory = vhost_client_set_memory;
    hdev->client.sync_dirty_bitmap = vhost_client_sync_dirty_bitmap;
    hdev->client.migration_log = vhost_client_migration_log;
    hdev->mem = qemu_mallocz(offsetof(struct vhost_memory, regions));
    hdev->log = NULL;
    hdev->log_size = 0;
    hdev->log_enabled = false;
    hdev->started = false;
    cpu_register_phys_memory_client(&hdev->client);
    return 0;
fail:
    r = -errno;
    close(hdev->control);
    return r;
}

void vhost_dev_cleanup(struct vhost_dev *hdev)
{
    cpu_unregister_phys_memory_client(&hdev->client);
    qemu_free(hdev->mem);
    close(hdev->control);
}

int vhost_dev_start(struct vhost_dev *hdev, VirtIODevice *vdev)
{
    int i, r;
    if (!vdev->binding->set_guest_notifiers) {
        fprintf(stderr, "binding does not support guest notifiers\n");
        r = -ENOSYS;
        goto fail;
    }

    r = vdev->binding->set_guest_notifiers(vdev->binding_opaque, true);
    if (r < 0) {
        fprintf(stderr, "Error binding guest notifier: %d\n", -r);
        goto fail_notifiers;
    }

    r = vhost_dev_set_features(hdev, hdev->log_enabled);
    if (r < 0) {
        goto fail_features;
    }
    r = ioctl(hdev->control, VHOST_SET_MEM_TABLE, hdev->mem);
    if (r < 0) {
        r = -errno;
        goto fail_mem;
    }
    for (i = 0; i < hdev->nvqs; ++i) {
        r = vhost_virtqueue_init(hdev,
                                 vdev,
                                 hdev->vqs + i,
                                 i);
        if (r < 0) {
            goto fail_vq;
        }
    }

    if (hdev->log_enabled) {
        hdev->log_size = vhost_get_log_size(hdev);
        hdev->log = hdev->log_size ?
            qemu_mallocz(hdev->log_size * sizeof *hdev->log) : NULL;
        r = ioctl(hdev->control, VHOST_SET_LOG_BASE,
                  (uint64_t)(unsigned long)hdev->log);
        if (r < 0) {
            r = -errno;
            goto fail_log;
        }
    }

    hdev->started = true;

    return 0;
fail_log:
fail_vq:
    while (--i >= 0) {
        vhost_virtqueue_cleanup(hdev,
                                vdev,
                                hdev->vqs + i,
                                i);
    }
fail_mem:
fail_features:
    vdev->binding->set_guest_notifiers(vdev->binding_opaque, false);
fail_notifiers:
fail:
    return r;
}

void vhost_dev_stop(struct vhost_dev *hdev, VirtIODevice *vdev)
{
    int i, r;

    for (i = 0; i < hdev->nvqs; ++i) {
        vhost_virtqueue_cleanup(hdev,
                                vdev,
                                hdev->vqs + i,
                                i);
    }
    vhost_client_sync_dirty_bitmap(&hdev->client, 0,
                                   (target_phys_addr_t)~0x0ull);
    r = vdev->binding->set_guest_notifiers(vdev->binding_opaque, false);
    if (r < 0) {
        fprintf(stderr, "vhost guest notifier cleanup failed: %d\n", r);
        fflush(stderr);
    }
    assert (r >= 0);

    hdev->started = false;
    qemu_free(hdev->log);
    hdev->log_size = 0;
}
Commit	Line	Data
d5970055 MT	1	/*
	2	* vhost support
	3	*
	4	* Copyright Red Hat, Inc. 2010
	5	*
	6	* Authors:
	7	* Michael S. Tsirkin <[email protected]>
	8	*
	9	* This work is licensed under the terms of the GNU GPL, version 2. See
	10	* the COPYING file in the top-level directory.
	11	*/
	12
d5970055	13	#include <sys/ioctl.h>
d5970055 MT	14	#include "vhost.h"
d5970055 MT	15	#include "hw/hw.h"
bf1b0071	16	#include "range.h"
11078ae3	17	#include <linux/vhost.h>
d5970055 MT	18
	19	static void vhost_dev_sync_region(struct vhost_dev *dev,
	20	uint64_t mfirst, uint64_t mlast,
	21	uint64_t rfirst, uint64_t rlast)
	22	{
	23	uint64_t start = MAX(mfirst, rfirst);
	24	uint64_t end = MIN(mlast, rlast);
	25	vhost_log_chunk_t *from = dev->log + start / VHOST_LOG_CHUNK;
	26	vhost_log_chunk_t *to = dev->log + end / VHOST_LOG_CHUNK + 1;
	27	uint64_t addr = (start / VHOST_LOG_CHUNK) * VHOST_LOG_CHUNK;
	28
	29	assert(end / VHOST_LOG_CHUNK < dev->log_size);
	30	assert(start / VHOST_LOG_CHUNK < dev->log_size);
	31	if (end < start) {
	32	return;
	33	}
	34	for (;from < to; ++from) {
	35	vhost_log_chunk_t log;
	36	int bit;
	37	/* We first check with non-atomic: much cheaper,
	38	* and we expect non-dirty to be the common case. */
	39	if (!*from) {
	40	continue;
	41	}
	42	/* Data must be read atomically. We don't really
	43	* need the barrier semantics of __sync
	44	* builtins, but it's easier to use them than
	45	* roll our own. */
	46	log = __sync_fetch_and_and(from, 0);
	47	while ((bit = sizeof(log) > sizeof(int) ?
	48	ffsll(log) : ffs(log))) {
	49	bit -= 1;
	50	cpu_physical_memory_set_dirty(addr + bit * VHOST_LOG_PAGE);
	51	log &= ~(0x1ull << bit);
	52	}
	53	addr += VHOST_LOG_CHUNK;
	54	}
	55	}
	56
	57	static int vhost_client_sync_dirty_bitmap(CPUPhysMemoryClient *client,
	58	target_phys_addr_t start_addr,
	59	target_phys_addr_t end_addr)
	60	{
	61	struct vhost_dev *dev = container_of(client, struct vhost_dev, client);
	62	int i;
	63	if (!dev->log_enabled \|\| !dev->started) {
	64	return 0;
	65	}
	66	for (i = 0; i < dev->mem->nregions; ++i) {
	67	struct vhost_memory_region *reg = dev->mem->regions + i;
	68	vhost_dev_sync_region(dev, start_addr, end_addr,
	69	reg->guest_phys_addr,
	70	range_get_last(reg->guest_phys_addr,
	71	reg->memory_size));
	72	}
	73	for (i = 0; i < dev->nvqs; ++i) {
	74	struct vhost_virtqueue *vq = dev->vqs + i;
	75	vhost_dev_sync_region(dev, start_addr, end_addr, vq->used_phys,
	76	range_get_last(vq->used_phys, vq->used_size));
	77	}
	78	return 0;
	79	}
	80
	81	/* Assign/unassign. Keep an unsorted array of non-overlapping
82	* memory regions in dev->mem. */
83	static void vhost_dev_unassign_memory(struct vhost_dev *dev,
84	uint64_t start_addr,
85	uint64_t size)
86	{
87	int from, to, n = dev->mem->nregions;
88	/* Track overlapping/split regions for sanity checking. */
89	int overlap_start = 0, overlap_end = 0, overlap_middle = 0, split = 0;
90
91	for (from = 0, to = 0; from < n; ++from, ++to) {
92	struct vhost_memory_region *reg = dev->mem->regions + to;
93	uint64_t reglast;
94	uint64_t memlast;
95	uint64_t change;
96
97	/* clone old region */
98	if (to != from) {
99	memcpy(reg, dev->mem->regions + from, sizeof *reg);
100	}
101
102	/* No overlap is simple */
103	if (!ranges_overlap(reg->guest_phys_addr, reg->memory_size,
104	start_addr, size)) {
105	continue;
106	}
107
108	/* Split only happens if supplied region
109	* is in the middle of an existing one. Thus it can not
110	* overlap with any other existing region. */
111	assert(!split);
112
113	reglast = range_get_last(reg->guest_phys_addr, reg->memory_size);
114	memlast = range_get_last(start_addr, size);
115
116	/* Remove whole region */
117	if (start_addr <= reg->guest_phys_addr && memlast >= reglast) {
118	--dev->mem->nregions;
119	--to;
120	assert(to >= 0);
121	++overlap_middle;
122	continue;
123	}
124
125	/* Shrink region */
126	if (memlast >= reglast) {
127	reg->memory_size = start_addr - reg->guest_phys_addr;
128	assert(reg->memory_size);
129	assert(!overlap_end);
130	++overlap_end;
131	continue;
132	}
133
134	/* Shift region */
135	if (start_addr <= reg->guest_phys_addr) {
136	change = memlast + 1 - reg->guest_phys_addr;
137	reg->memory_size -= change;
138	reg->guest_phys_addr += change;
139	reg->userspace_addr += change;
140	assert(reg->memory_size);
141	assert(!overlap_start);
142	++overlap_start;
143	continue;
144	}
145
146	/* This only happens if supplied region
147	* is in the middle of an existing one. Thus it can not
148	* overlap with any other existing region. */
149	assert(!overlap_start);
150	assert(!overlap_end);
151	assert(!overlap_middle);
152	/* Split region: shrink first part, shift second part. */
153	memcpy(dev->mem->regions + n, reg, sizeof *reg);
154	reg->memory_size = start_addr - reg->guest_phys_addr;
155	assert(reg->memory_size);
156	change = memlast + 1 - reg->guest_phys_addr;
157	reg = dev->mem->regions + n;
158	reg->memory_size -= change;
159	assert(reg->memory_size);
160	reg->guest_phys_addr += change;
161	reg->userspace_addr += change;
162	/* Never add more than 1 region */
163	assert(dev->mem->nregions == n);
164	++dev->mem->nregions;
165	++split;
166	}
167	}
168
169	/* Called after unassign, so no regions overlap the given range. */
170	static void vhost_dev_assign_memory(struct vhost_dev *dev,
171	uint64_t start_addr,
172	uint64_t size,
173	uint64_t uaddr)
174	{
175	int from, to;
176	struct vhost_memory_region *merged = NULL;
177	for (from = 0, to = 0; from < dev->mem->nregions; ++from, ++to) {
178	struct vhost_memory_region *reg = dev->mem->regions + to;
179	uint64_t prlast, urlast;
180	uint64_t pmlast, umlast;
181	uint64_t s, e, u;
182
183	/* clone old region */
184	if (to != from) {
185	memcpy(reg, dev->mem->regions + from, sizeof *reg);
186	}
187	prlast = range_get_last(reg->guest_phys_addr, reg->memory_size);
188	pmlast = range_get_last(start_addr, size);
189	urlast = range_get_last(reg->userspace_addr, reg->memory_size);
190	umlast = range_get_last(uaddr, size);
191
192	/* check for overlapping regions: should never happen. */
193	assert(prlast < start_addr \|\| pmlast < reg->guest_phys_addr);
194	/* Not an adjacent or overlapping region - do not merge. */
195	if ((prlast + 1 != start_addr \|\| urlast + 1 != uaddr) &&
196	(pmlast + 1 != reg->guest_phys_addr \|\|
197	umlast + 1 != reg->userspace_addr)) {
198	continue;
199	}
200
201	if (merged) {
202	--to;
203	assert(to >= 0);
204	} else {
205	merged = reg;
206	}
207	u = MIN(uaddr, reg->userspace_addr);
208	s = MIN(start_addr, reg->guest_phys_addr);
209	e = MAX(pmlast, prlast);
210	uaddr = merged->userspace_addr = u;
211	start_addr = merged->guest_phys_addr = s;
212	size = merged->memory_size = e - s + 1;
213	assert(merged->memory_size);
214	}
215
216	if (!merged) {
217	struct vhost_memory_region *reg = dev->mem->regions + to;
218	memset(reg, 0, sizeof *reg);
219	reg->memory_size = size;
220	assert(reg->memory_size);
221	reg->guest_phys_addr = start_addr;
222	reg->userspace_addr = uaddr;
223	++to;
224	}
225	assert(to <= dev->mem->nregions + 1);
226	dev->mem->nregions = to;
227	}
228
229	static uint64_t vhost_get_log_size(struct vhost_dev *dev)
230	{
231	uint64_t log_size = 0;
232	int i;
233	for (i = 0; i < dev->mem->nregions; ++i) {
234	struct vhost_memory_region *reg = dev->mem->regions + i;
235	uint64_t last = range_get_last(reg->guest_phys_addr,
236	reg->memory_size);
237	log_size = MAX(log_size, last / VHOST_LOG_CHUNK + 1);
238	}
239	for (i = 0; i < dev->nvqs; ++i) {
240	struct vhost_virtqueue *vq = dev->vqs + i;
241	uint64_t last = vq->used_phys + vq->used_size - 1;
242	log_size = MAX(log_size, last / VHOST_LOG_CHUNK + 1);
243	}
244	return log_size;
245	}
246
247	static inline void vhost_dev_log_resize(struct vhost_dev* dev, uint64_t size)
248	{
249	vhost_log_chunk_t *log;
250	uint64_t log_base;
251	int r;
252	if (size) {
253	log = qemu_mallocz(size * sizeof *log);
254	} else {
255	log = NULL;
256	}
257	log_base = (uint64_t)(unsigned long)log;
258	r = ioctl(dev->control, VHOST_SET_LOG_BASE, &log_base);
259	assert(r >= 0);
260	vhost_client_sync_dirty_bitmap(&dev->client, 0,
261	(target_phys_addr_t)~0x0ull);
262	if (dev->log) {
263	qemu_free(dev->log);
264	}
265	dev->log = log;
266	dev->log_size = size;
267	}
268
269	static int vhost_verify_ring_mappings(struct vhost_dev *dev,
270	uint64_t start_addr,
271	uint64_t size)
272	{
273	int i;
274	for (i = 0; i < dev->nvqs; ++i) {
275	struct vhost_virtqueue *vq = dev->vqs + i;
276	target_phys_addr_t l;
277	void *p;
278
279	if (!ranges_overlap(start_addr, size, vq->ring_phys, vq->ring_size)) {
280	continue;
281	}
282	l = vq->ring_size;
283	p = cpu_physical_memory_map(vq->ring_phys, &l, 1);
284	if (!p \|\| l != vq->ring_size) {
285	fprintf(stderr, "Unable to map ring buffer for ring %d\n", i);
286	return -ENOMEM;
287	}
288	if (p != vq->ring) {
289	fprintf(stderr, "Ring buffer relocated for ring %d\n", i);
290	return -EBUSY;
291	}
292	cpu_physical_memory_unmap(p, l, 0, 0);
293	}
294	return 0;
295	}
296
297	static void vhost_client_set_memory(CPUPhysMemoryClient *client,
298	target_phys_addr_t start_addr,
299	ram_addr_t size,
300	ram_addr_t phys_offset)
301	{
302	struct vhost_dev *dev = container_of(client, struct vhost_dev, client);
303	ram_addr_t flags = phys_offset & ~TARGET_PAGE_MASK;
304	int s = offsetof(struct vhost_memory, regions) +
305	(dev->mem->nregions + 1) * sizeof dev->mem->regions[0];
306	uint64_t log_size;
307	int r;
308	dev->mem = qemu_realloc(dev->mem, s);
309
310	assert(size);
311
312	vhost_dev_unassign_memory(dev, start_addr, size);
313	if (flags == IO_MEM_RAM) {
314	/* Add given mapping, merging adjacent regions if any */
315	vhost_dev_assign_memory(dev, start_addr, size,
316	(uintptr_t)qemu_get_ram_ptr(phys_offset));
317	} else {
318	/* Remove old mapping for this memory, if any. */
319	vhost_dev_unassign_memory(dev, start_addr, size);
320	}
321
322	if (!dev->started) {
323	return;
324	}
325
326	if (dev->started) {
327	r = vhost_verify_ring_mappings(dev, start_addr, size);
328	assert(r >= 0);
329	}
330
331	if (!dev->log_enabled) {
332	r = ioctl(dev->control, VHOST_SET_MEM_TABLE, dev->mem);
333	assert(r >= 0);
334	return;
335	}
336	log_size = vhost_get_log_size(dev);
337	/* We allocate an extra 4K bytes to log,
338	* to reduce the * number of reallocations. */
339	#define VHOST_LOG_BUFFER (0x1000 / sizeof *dev->log)
340	/* To log more, must increase log size before table update. */
341	if (dev->log_size < log_size) {
342	vhost_dev_log_resize(dev, log_size + VHOST_LOG_BUFFER);
343	}
344	r = ioctl(dev->control, VHOST_SET_MEM_TABLE, dev->mem);
345	assert(r >= 0);
346	/* To log less, can only decrease log size after table update. */
347	if (dev->log_size > log_size + VHOST_LOG_BUFFER) {
348	vhost_dev_log_resize(dev, log_size);
349	}
350	}
351
352	static int vhost_virtqueue_set_addr(struct vhost_dev *dev,
353	struct vhost_virtqueue *vq,
354	unsigned idx, bool enable_log)
355	{
356	struct vhost_vring_addr addr = {
357	.index = idx,
2b3af999 SW	358	.desc_user_addr = (uint64_t)(unsigned long)vq->desc,
	359	.avail_user_addr = (uint64_t)(unsigned long)vq->avail,
	360	.used_user_addr = (uint64_t)(unsigned long)vq->used,
d5970055 MT	361	.log_guest_addr = vq->used_phys,
	362	.flags = enable_log ? (1 << VHOST_VRING_F_LOG) : 0,
	363	};
	364	int r = ioctl(dev->control, VHOST_SET_VRING_ADDR, &addr);
	365	if (r < 0) {
	366	return -errno;
	367	}
	368	return 0;
	369	}
	370
	371	static int vhost_dev_set_features(struct vhost_dev *dev, bool enable_log)
	372	{
	373	uint64_t features = dev->acked_features;
	374	int r;
	375	if (enable_log) {
	376	features \|= 0x1 << VHOST_F_LOG_ALL;
	377	}
	378	r = ioctl(dev->control, VHOST_SET_FEATURES, &features);
	379	return r < 0 ? -errno : 0;
	380	}
	381
	382	static int vhost_dev_set_log(struct vhost_dev *dev, bool enable_log)
	383	{
	384	int r, t, i;
	385	r = vhost_dev_set_features(dev, enable_log);
	386	if (r < 0) {
	387	goto err_features;
	388	}
	389	for (i = 0; i < dev->nvqs; ++i) {
	390	r = vhost_virtqueue_set_addr(dev, dev->vqs + i, i,
	391	enable_log);
	392	if (r < 0) {
	393	goto err_vq;
	394	}
	395	}
	396	return 0;
	397	err_vq:
	398	for (; i >= 0; --i) {
	399	t = vhost_virtqueue_set_addr(dev, dev->vqs + i, i,
	400	dev->log_enabled);
	401	assert(t >= 0);
	402	}
	403	t = vhost_dev_set_features(dev, dev->log_enabled);
	404	assert(t >= 0);
	405	err_features:
	406	return r;
	407	}
	408
	409	static int vhost_client_migration_log(CPUPhysMemoryClient *client,
	410	int enable)
	411	{
	412	struct vhost_dev *dev = container_of(client, struct vhost_dev, client);
	413	int r;
	414	if (!!enable == dev->log_enabled) {
	415	return 0;
	416	}
	417	if (!dev->started) {
	418	dev->log_enabled = enable;
	419	return 0;
	420	}
	421	if (!enable) {
	422	r = vhost_dev_set_log(dev, false);
	423	if (r < 0) {
	424	return r;
425	}
426	if (dev->log) {
427	qemu_free(dev->log);
428	}
429	dev->log = NULL;
430	dev->log_size = 0;
431	} else {
432	vhost_dev_log_resize(dev, vhost_get_log_size(dev));
433	r = vhost_dev_set_log(dev, true);
434	if (r < 0) {
435	return r;
436	}
437	}
438	dev->log_enabled = enable;
439	return 0;
440	}
441
442	static int vhost_virtqueue_init(struct vhost_dev *dev,
443	struct VirtIODevice *vdev,
444	struct vhost_virtqueue *vq,
445	unsigned idx)
446	{
447	target_phys_addr_t s, l, a;
448	int r;
449	struct vhost_vring_file file = {
450	.index = idx,
451	};
452	struct vhost_vring_state state = {
453	.index = idx,
454	};
455	struct VirtQueue *vvq = virtio_get_queue(vdev, idx);
456
d5970055 MT	457	if (!vdev->binding->set_host_notifier) {
	458	fprintf(stderr, "binding does not support host notifiers\n");
	459	return -ENOSYS;
	460	}
	461
	462	vq->num = state.num = virtio_queue_get_num(vdev, idx);
	463	r = ioctl(dev->control, VHOST_SET_VRING_NUM, &state);
	464	if (r) {
	465	return -errno;
	466	}
	467
	468	state.num = virtio_queue_get_last_avail_idx(vdev, idx);
	469	r = ioctl(dev->control, VHOST_SET_VRING_BASE, &state);
	470	if (r) {
	471	return -errno;
	472	}
	473
	474	s = l = virtio_queue_get_desc_size(vdev, idx);
	475	a = virtio_queue_get_desc_addr(vdev, idx);
	476	vq->desc = cpu_physical_memory_map(a, &l, 0);
	477	if (!vq->desc \|\| l != s) {
	478	r = -ENOMEM;
	479	goto fail_alloc_desc;
	480	}
	481	s = l = virtio_queue_get_avail_size(vdev, idx);
	482	a = virtio_queue_get_avail_addr(vdev, idx);
	483	vq->avail = cpu_physical_memory_map(a, &l, 0);
	484	if (!vq->avail \|\| l != s) {
	485	r = -ENOMEM;
	486	goto fail_alloc_avail;
	487	}
	488	vq->used_size = s = l = virtio_queue_get_used_size(vdev, idx);
	489	vq->used_phys = a = virtio_queue_get_used_addr(vdev, idx);
	490	vq->used = cpu_physical_memory_map(a, &l, 1);
	491	if (!vq->used \|\| l != s) {
	492	r = -ENOMEM;
	493	goto fail_alloc_used;
	494	}
	495
	496	vq->ring_size = s = l = virtio_queue_get_ring_size(vdev, idx);
	497	vq->ring_phys = a = virtio_queue_get_ring_addr(vdev, idx);
	498	vq->ring = cpu_physical_memory_map(a, &l, 1);
	499	if (!vq->ring \|\| l != s) {
	500	r = -ENOMEM;
	501	goto fail_alloc_ring;
	502	}
	503
	504	r = vhost_virtqueue_set_addr(dev, vq, idx, dev->log_enabled);
	505	if (r < 0) {
	506	r = -errno;
	507	goto fail_alloc;
	508	}
d5970055 MT	509	r = vdev->binding->set_host_notifier(vdev->binding_opaque, idx, true);
	510	if (r < 0) {
	511	fprintf(stderr, "Error binding host notifier: %d\n", -r);
	512	goto fail_host_notifier;
	513	}
	514
	515	file.fd = event_notifier_get_fd(virtio_queue_get_host_notifier(vvq));
	516	r = ioctl(dev->control, VHOST_SET_VRING_KICK, &file);
	517	if (r) {
c8852121	518	r = -errno;
d5970055 MT	519	goto fail_kick;
	520	}
	521
	522	file.fd = event_notifier_get_fd(virtio_queue_get_guest_notifier(vvq));
	523	r = ioctl(dev->control, VHOST_SET_VRING_CALL, &file);
	524	if (r) {
c8852121	525	r = -errno;
d5970055 MT	526	goto fail_call;
	527	}
	528
	529	return 0;
	530
	531	fail_call:
	532	fail_kick:
	533	vdev->binding->set_host_notifier(vdev->binding_opaque, idx, false);
	534	fail_host_notifier:
d5970055 MT	535	fail_alloc:
	536	cpu_physical_memory_unmap(vq->ring, virtio_queue_get_ring_size(vdev, idx),
	537	0, 0);
	538	fail_alloc_ring:
	539	cpu_physical_memory_unmap(vq->used, virtio_queue_get_used_size(vdev, idx),
	540	0, 0);
	541	fail_alloc_used:
	542	cpu_physical_memory_unmap(vq->avail, virtio_queue_get_avail_size(vdev, idx),
	543	0, 0);
	544	fail_alloc_avail:
	545	cpu_physical_memory_unmap(vq->desc, virtio_queue_get_desc_size(vdev, idx),
	546	0, 0);
	547	fail_alloc_desc:
	548	return r;
	549	}
	550
	551	static void vhost_virtqueue_cleanup(struct vhost_dev *dev,
	552	struct VirtIODevice *vdev,
	553	struct vhost_virtqueue *vq,
	554	unsigned idx)
	555	{
	556	struct vhost_vring_state state = {
	557	.index = idx,
	558	};
	559	int r;
d5970055 MT	560	r = vdev->binding->set_host_notifier(vdev->binding_opaque, idx, false);
	561	if (r < 0) {
	562	fprintf(stderr, "vhost VQ %d host cleanup failed: %d\n", idx, r);
	563	fflush(stderr);
	564	}
	565	assert (r >= 0);
	566	r = ioctl(dev->control, VHOST_GET_VRING_BASE, &state);
	567	if (r < 0) {
	568	fprintf(stderr, "vhost VQ %d ring restore failed: %d\n", idx, r);
	569	fflush(stderr);
	570	}
	571	virtio_queue_set_last_avail_idx(vdev, idx, state.num);
	572	assert (r >= 0);
	573	cpu_physical_memory_unmap(vq->ring, virtio_queue_get_ring_size(vdev, idx),
	574	0, virtio_queue_get_ring_size(vdev, idx));
	575	cpu_physical_memory_unmap(vq->used, virtio_queue_get_used_size(vdev, idx),
	576	1, virtio_queue_get_used_size(vdev, idx));
	577	cpu_physical_memory_unmap(vq->avail, virtio_queue_get_avail_size(vdev, idx),
	578	0, virtio_queue_get_avail_size(vdev, idx));
	579	cpu_physical_memory_unmap(vq->desc, virtio_queue_get_desc_size(vdev, idx),
	580	0, virtio_queue_get_desc_size(vdev, idx));
	581	}
	582
	583	int vhost_dev_init(struct vhost_dev *hdev, int devfd)
	584	{
	585	uint64_t features;
	586	int r;
	587	if (devfd >= 0) {
	588	hdev->control = devfd;
	589	} else {
	590	hdev->control = open("/dev/vhost-net", O_RDWR);
	591	if (hdev->control < 0) {
	592	return -errno;
	593	}
	594	}
	595	r = ioctl(hdev->control, VHOST_SET_OWNER, NULL);
	596	if (r < 0) {
	597	goto fail;
	598	}
	599
	600	r = ioctl(hdev->control, VHOST_GET_FEATURES, &features);
	601	if (r < 0) {
	602	goto fail;
	603	}
	604	hdev->features = features;
	605
	606	hdev->client.set_memory = vhost_client_set_memory;
	607	hdev->client.sync_dirty_bitmap = vhost_client_sync_dirty_bitmap;
	608	hdev->client.migration_log = vhost_client_migration_log;
	609	hdev->mem = qemu_mallocz(offsetof(struct vhost_memory, regions));
	610	hdev->log = NULL;
	611	hdev->log_size = 0;
	612	hdev->log_enabled = false;
	613	hdev->started = false;
	614	cpu_register_phys_memory_client(&hdev->client);
	615	return 0;
	616	fail:
	617	r = -errno;
	618	close(hdev->control);
	619	return r;
	620	}
	621
	622	void vhost_dev_cleanup(struct vhost_dev *hdev)
	623	{
624	cpu_unregister_phys_memory_client(&hdev->client);
625	qemu_free(hdev->mem);
626	close(hdev->control);
627	}
628
629	int vhost_dev_start(struct vhost_dev hdev, VirtIODevice vdev)
630	{
631	int i, r;
54dd9321 MT	632	if (!vdev->binding->set_guest_notifiers) {
	633	fprintf(stderr, "binding does not support guest notifiers\n");
	634	r = -ENOSYS;
	635	goto fail;
	636	}
	637
	638	r = vdev->binding->set_guest_notifiers(vdev->binding_opaque, true);
	639	if (r < 0) {
	640	fprintf(stderr, "Error binding guest notifier: %d\n", -r);
	641	goto fail_notifiers;
	642	}
d5970055 MT	643
	644	r = vhost_dev_set_features(hdev, hdev->log_enabled);
	645	if (r < 0) {
54dd9321	646	goto fail_features;
d5970055 MT	647	}
	648	r = ioctl(hdev->control, VHOST_SET_MEM_TABLE, hdev->mem);
	649	if (r < 0) {
	650	r = -errno;
54dd9321	651	goto fail_mem;
d5970055	652	}
d154e0ba MT	653	for (i = 0; i < hdev->nvqs; ++i) {
	654	r = vhost_virtqueue_init(hdev,
	655	vdev,
	656	hdev->vqs + i,
	657	i);
	658	if (r < 0) {
	659	goto fail_vq;
	660	}
	661	}
	662
d5970055 MT	663	if (hdev->log_enabled) {
	664	hdev->log_size = vhost_get_log_size(hdev);
	665	hdev->log = hdev->log_size ?
	666	qemu_mallocz(hdev->log_size * sizeof *hdev->log) : NULL;
	667	r = ioctl(hdev->control, VHOST_SET_LOG_BASE,
	668	(uint64_t)(unsigned long)hdev->log);
	669	if (r < 0) {
	670	r = -errno;
54dd9321	671	goto fail_log;
d5970055 MT	672	}
d5970055 MT	673	}
d154e0ba	674
d5970055 MT	675	hdev->started = true;
	676
	677	return 0;
54dd9321	678	fail_log:
d5970055 MT	679	fail_vq:
	680	while (--i >= 0) {
	681	vhost_virtqueue_cleanup(hdev,
	682	vdev,
	683	hdev->vqs + i,
	684	i);
	685	}
54dd9321 MT	686	fail_mem:
	687	fail_features:
	688	vdev->binding->set_guest_notifiers(vdev->binding_opaque, false);
	689	fail_notifiers:
d5970055 MT	690	fail:
	691	return r;
	692	}
	693
	694	void vhost_dev_stop(struct vhost_dev hdev, VirtIODevice vdev)
	695	{
54dd9321 MT	696	int i, r;
54dd9321 MT	697
d5970055 MT	698	for (i = 0; i < hdev->nvqs; ++i) {
	699	vhost_virtqueue_cleanup(hdev,
	700	vdev,
	701	hdev->vqs + i,
	702	i);
	703	}
	704	vhost_client_sync_dirty_bitmap(&hdev->client, 0,
	705	(target_phys_addr_t)~0x0ull);
54dd9321 MT	706	r = vdev->binding->set_guest_notifiers(vdev->binding_opaque, false);
	707	if (r < 0) {
	708	fprintf(stderr, "vhost guest notifier cleanup failed: %d\n", r);
	709	fflush(stderr);
	710	}
	711	assert (r >= 0);
	712
d5970055 MT	713	hdev->started = false;
	714	qemu_free(hdev->log);
	715	hdev->log_size = 0;
	716	}