contrib/qemu: Import qemu block source code

This qemu block format source code and its minimal
dependency files will be used in the next patch to implement
a qemu-block format translator.

Change-Id: Ic87638972f7ea9b3df84d7a0539512a250c11c1c
BUG: 986775
Signed-off-by: Anand Avati <avati@redhat.com>
Reviewed-on: http://review.gluster.org/5366
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>

107 files changed, 39119 insertions, 0 deletions

diff --git a/contrib/qemu/block.c b/contrib/qemu/block.c
new file mode 100644
index 000000000..b56024113
--- /dev/null
+++ b/contrib/qemu/block.c
@@ -0,0 +1,4604 @@
+/*
+ * QEMU System Emulator block driver
+ *
+ * Copyright (c) 2003 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "config-host.h"
+#include "qemu-common.h"
+#include "trace.h"
+#include "monitor/monitor.h"
+#include "block/block_int.h"
+#include "block/blockjob.h"
+#include "qemu/module.h"
+#include "qapi/qmp/qjson.h"
+#include "sysemu/sysemu.h"
+#include "qemu/notify.h"
+#include "block/coroutine.h"
+#include "qmp-commands.h"
+#include "qemu/timer.h"
+
+#ifdef CONFIG_BSD
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/ioctl.h>
+#include <sys/queue.h>
+#ifndef __DragonFly__
+#include <sys/disk.h>
+#endif
+#endif
+
+#ifdef _WIN32
+#include <windows.h>
+#endif
+
+#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */
+
+typedef enum {
+    BDRV_REQ_COPY_ON_READ = 0x1,
+    BDRV_REQ_ZERO_WRITE   = 0x2,
+} BdrvRequestFlags;
+
+static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load);
+static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque);
+static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque);
+static int coroutine_fn bdrv_co_readv_em(BlockDriverState *bs,
+                                         int64_t sector_num, int nb_sectors,
+                                         QEMUIOVector *iov);
+static int coroutine_fn bdrv_co_writev_em(BlockDriverState *bs,
+                                         int64_t sector_num, int nb_sectors,
+                                         QEMUIOVector *iov);
+static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+    BdrvRequestFlags flags);
+static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+    BdrvRequestFlags flags);
+static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
+                                               int64_t sector_num,
+                                               QEMUIOVector *qiov,
+                                               int nb_sectors,
+                                               BlockDriverCompletionFunc *cb,
+                                               void *opaque,
+                                               bool is_write);
+static void coroutine_fn bdrv_co_do_rw(void *opaque);
+static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors);
+
+static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
+        bool is_write, double elapsed_time, uint64_t *wait);
+static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
+        double elapsed_time, uint64_t *wait);
+static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
+        bool is_write, int64_t *wait);
+
+static QTAILQ_HEAD(, BlockDriverState) bdrv_states =
+    QTAILQ_HEAD_INITIALIZER(bdrv_states);
+
+static QLIST_HEAD(, BlockDriver) bdrv_drivers =
+    QLIST_HEAD_INITIALIZER(bdrv_drivers);
+
+/* If non-zero, use only whitelisted block drivers */
+static int use_bdrv_whitelist;
+
+#ifdef _WIN32
+static int is_windows_drive_prefix(const char *filename)
+{
+    return (((filename[0] >= 'a' && filename[0] <= 'z') ||
+             (filename[0] >= 'A' && filename[0] <= 'Z')) &&
+            filename[1] == ':');
+}
+
+int is_windows_drive(const char *filename)
+{
+    if (is_windows_drive_prefix(filename) &&
+        filename[2] == '\0')
+        return 1;
+    if (strstart(filename, "\\\\.\\", NULL) ||
+        strstart(filename, "//./", NULL))
+        return 1;
+    return 0;
+}
+#endif
+
+/* throttling disk I/O limits */
+void bdrv_io_limits_disable(BlockDriverState *bs)
+{
+    bs->io_limits_enabled = false;
+
+    while (qemu_co_queue_next(&bs->throttled_reqs));
+
+    if (bs->block_timer) {
+        qemu_del_timer(bs->block_timer);
+        qemu_free_timer(bs->block_timer);
+        bs->block_timer = NULL;
+    }
+
+    bs->slice_start = 0;
+    bs->slice_end   = 0;
+}
+
+static void bdrv_block_timer(void *opaque)
+{
+    BlockDriverState *bs = opaque;
+
+    qemu_co_queue_next(&bs->throttled_reqs);
+}
+
+void bdrv_io_limits_enable(BlockDriverState *bs)
+{
+    qemu_co_queue_init(&bs->throttled_reqs);
+    bs->block_timer = qemu_new_timer_ns(vm_clock, bdrv_block_timer, bs);
+    bs->io_limits_enabled = true;
+}
+
+bool bdrv_io_limits_enabled(BlockDriverState *bs)
+{
+    BlockIOLimit *io_limits = &bs->io_limits;
+    return io_limits->bps[BLOCK_IO_LIMIT_READ]
+         || io_limits->bps[BLOCK_IO_LIMIT_WRITE]
+         || io_limits->bps[BLOCK_IO_LIMIT_TOTAL]
+         || io_limits->iops[BLOCK_IO_LIMIT_READ]
+         || io_limits->iops[BLOCK_IO_LIMIT_WRITE]
+         || io_limits->iops[BLOCK_IO_LIMIT_TOTAL];
+}
+
+static void bdrv_io_limits_intercept(BlockDriverState *bs,
+                                     bool is_write, int nb_sectors)
+{
+    int64_t wait_time = -1;
+
+    if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
+        qemu_co_queue_wait(&bs->throttled_reqs);
+    }
+
+    /* In fact, we hope to keep each request's timing, in FIFO mode. The next
+     * throttled requests will not be dequeued until the current request is
+     * allowed to be serviced. So if the current request still exceeds the
+     * limits, it will be inserted to the head. All requests followed it will
+     * be still in throttled_reqs queue.
+     */
+
+    while (bdrv_exceed_io_limits(bs, nb_sectors, is_write, &wait_time)) {
+        qemu_mod_timer(bs->block_timer,
+                       wait_time + qemu_get_clock_ns(vm_clock));
+        qemu_co_queue_wait_insert_head(&bs->throttled_reqs);
+    }
+
+    qemu_co_queue_next(&bs->throttled_reqs);
+}
+
+/* check if the path starts with "<protocol>:" */
+static int path_has_protocol(const char *path)
+{
+    const char *p;
+
+#ifdef _WIN32
+    if (is_windows_drive(path) ||
+        is_windows_drive_prefix(path)) {
+        return 0;
+    }
+    p = path + strcspn(path, ":/\\");
+#else
+    p = path + strcspn(path, ":/");
+#endif
+
+    return *p == ':';
+}
+
+int path_is_absolute(const char *path)
+{
+#ifdef _WIN32
+    /* specific case for names like: "\\.\d:" */
+    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
+        return 1;
+    }
+    return (*path == '/' || *path == '\\');
+#else
+    return (*path == '/');
+#endif
+}
+
+/* if filename is absolute, just copy it to dest. Otherwise, build a
+   path to it by considering it is relative to base_path. URL are
+   supported. */
+void path_combine(char *dest, int dest_size,
+                  const char *base_path,
+                  const char *filename)
+{
+    const char *p, *p1;
+    int len;
+
+    if (dest_size <= 0)
+        return;
+    if (path_is_absolute(filename)) {
+        pstrcpy(dest, dest_size, filename);
+    } else {
+        p = strchr(base_path, ':');
+        if (p)
+            p++;
+        else
+            p = base_path;
+        p1 = strrchr(base_path, '/');
+#ifdef _WIN32
+        {
+            const char *p2;
+            p2 = strrchr(base_path, '\\');
+            if (!p1 || p2 > p1)
+                p1 = p2;
+        }
+#endif
+        if (p1)
+            p1++;
+        else
+            p1 = base_path;
+        if (p1 > p)
+            p = p1;
+        len = p - base_path;
+        if (len > dest_size - 1)
+            len = dest_size - 1;
+        memcpy(dest, base_path, len);
+        dest[len] = '\0';
+        pstrcat(dest, dest_size, filename);
+    }
+}
+
+void bdrv_get_full_backing_filename(BlockDriverState *bs, char *dest, size_t sz)
+{
+    if (bs->backing_file[0] == '\0' || path_has_protocol(bs->backing_file)) {
+        pstrcpy(dest, sz, bs->backing_file);
+    } else {
+        path_combine(dest, sz, bs->filename, bs->backing_file);
+    }
+}
+
+void bdrv_register(BlockDriver *bdrv)
+{
+    /* Block drivers without coroutine functions need emulation */
+    if (!bdrv->bdrv_co_readv) {
+        bdrv->bdrv_co_readv = bdrv_co_readv_em;
+        bdrv->bdrv_co_writev = bdrv_co_writev_em;
+
+        /* bdrv_co_readv_em()/brdv_co_writev_em() work in terms of aio, so if
+         * the block driver lacks aio we need to emulate that too.
+         */
+        if (!bdrv->bdrv_aio_readv) {
+            /* add AIO emulation layer */
+            bdrv->bdrv_aio_readv = bdrv_aio_readv_em;
+            bdrv->bdrv_aio_writev = bdrv_aio_writev_em;
+        }
+    }
+
+    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
+}
+
+/* create a new block device (by default it is empty) */
+BlockDriverState *bdrv_new(const char *device_name)
+{
+    BlockDriverState *bs;
+
+    bs = g_malloc0(sizeof(BlockDriverState));
+    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
+    if (device_name[0] != '\0') {
+        QTAILQ_INSERT_TAIL(&bdrv_states, bs, list);
+    }
+    bdrv_iostatus_disable(bs);
+    notifier_list_init(&bs->close_notifiers);
+    notifier_with_return_list_init(&bs->before_write_notifiers);
+
+    return bs;
+}
+
+void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify)
+{
+    notifier_list_add(&bs->close_notifiers, notify);
+}
+
+BlockDriver *bdrv_find_format(const char *format_name)
+{
+    BlockDriver *drv1;
+    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
+        if (!strcmp(drv1->format_name, format_name)) {
+            return drv1;
+        }
+    }
+    return NULL;
+}
+
+static int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
+{
+    static const char *whitelist_rw[] = {
+        CONFIG_BDRV_RW_WHITELIST
+    };
+    static const char *whitelist_ro[] = {
+        CONFIG_BDRV_RO_WHITELIST
+    };
+    const char **p;
+
+    if (!whitelist_rw[0] && !whitelist_ro[0]) {
+        return 1;               /* no whitelist, anything goes */
+    }
+
+    for (p = whitelist_rw; *p; p++) {
+        if (!strcmp(drv->format_name, *p)) {
+            return 1;
+        }
+    }
+    if (read_only) {
+        for (p = whitelist_ro; *p; p++) {
+            if (!strcmp(drv->format_name, *p)) {
+                return 1;
+            }
+        }
+    }
+    return 0;
+}
+
+BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
+                                          bool read_only)
+{
+    BlockDriver *drv = bdrv_find_format(format_name);
+    return drv && bdrv_is_whitelisted(drv, read_only) ? drv : NULL;
+}
+
+typedef struct CreateCo {
+    BlockDriver *drv;
+    char *filename;
+    QEMUOptionParameter *options;
+    int ret;
+} CreateCo;
+
+static void coroutine_fn bdrv_create_co_entry(void *opaque)
+{
+    CreateCo *cco = opaque;
+    assert(cco->drv);
+
+    cco->ret = cco->drv->bdrv_create(cco->filename, cco->options);
+}
+
+int bdrv_create(BlockDriver *drv, const char* filename,
+    QEMUOptionParameter *options)
+{
+    int ret;
+
+    Coroutine *co;
+    CreateCo cco = {
+        .drv = drv,
+        .filename = g_strdup(filename),
+        .options = options,
+        .ret = NOT_DONE,
+    };
+
+    if (!drv->bdrv_create) {
+        ret = -ENOTSUP;
+        goto out;
+    }
+
+    if (qemu_in_coroutine()) {
+        /* Fast-path if already in coroutine context */
+        bdrv_create_co_entry(&cco);
+    } else {
+        co = qemu_coroutine_create(bdrv_create_co_entry);
+        qemu_coroutine_enter(co, &cco);
+        while (cco.ret == NOT_DONE) {
+            qemu_aio_wait();
+        }
+    }
+
+    ret = cco.ret;
+
+out:
+    g_free(cco.filename);
+    return ret;
+}
+
+int bdrv_create_file(const char* filename, QEMUOptionParameter *options)
+{
+    BlockDriver *drv;
+
+    drv = bdrv_find_protocol(filename, true);
+    if (drv == NULL) {
+        return -ENOENT;
+    }
+
+    return bdrv_create(drv, filename, options);
+}
+
+/*
+ * Create a uniquely-named empty temporary file.
+ * Return 0 upon success, otherwise a negative errno value.
+ */
+int get_tmp_filename(char *filename, int size)
+{
+#ifdef _WIN32
+    char temp_dir[MAX_PATH];
+    /* GetTempFileName requires that its output buffer (4th param)
+       have length MAX_PATH or greater.  */
+    assert(size >= MAX_PATH);
+    return (GetTempPath(MAX_PATH, temp_dir)
+            && GetTempFileName(temp_dir, "qem", 0, filename)
+            ? 0 : -GetLastError());
+#else
+    int fd;
+    const char *tmpdir;
+    tmpdir = getenv("TMPDIR");
+    if (!tmpdir)
+        tmpdir = "/tmp";
+    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
+        return -EOVERFLOW;
+    }
+    fd = mkstemp(filename);
+    if (fd < 0) {
+        return -errno;
+    }
+    if (close(fd) != 0) {
+        unlink(filename);
+        return -errno;
+    }
+    return 0;
+#endif
+}
+
+/*
+ * Detect host devices. By convention, /dev/cdrom[N] is always
+ * recognized as a host CDROM.
+ */
+static BlockDriver *find_hdev_driver(const char *filename)
+{
+    int score_max = 0, score;
+    BlockDriver *drv = NULL, *d;
+
+    QLIST_FOREACH(d, &bdrv_drivers, list) {
+        if (d->bdrv_probe_device) {
+            score = d->bdrv_probe_device(filename);
+            if (score > score_max) {
+                score_max = score;
+                drv = d;
+            }
+        }
+    }
+
+    return drv;
+}
+
+BlockDriver *bdrv_find_protocol(const char *filename,
+                                bool allow_protocol_prefix)
+{
+    BlockDriver *drv1;
+    char protocol[128];
+    int len;
+    const char *p;
+
+    /* TODO Drivers without bdrv_file_open must be specified explicitly */
+
+    /*
+     * XXX(hch): we really should not let host device detection
+     * override an explicit protocol specification, but moving this
+     * later breaks access to device names with colons in them.
+     * Thanks to the brain-dead persistent naming schemes on udev-
+     * based Linux systems those actually are quite common.
+     */
+    drv1 = find_hdev_driver(filename);
+    if (drv1) {
+        return drv1;
+    }
+
+    if (!path_has_protocol(filename) || !allow_protocol_prefix) {
+        return bdrv_find_format("file");
+    }
+
+    p = strchr(filename, ':');
+    assert(p != NULL);
+    len = p - filename;
+    if (len > sizeof(protocol) - 1)
+        len = sizeof(protocol) - 1;
+    memcpy(protocol, filename, len);
+    protocol[len] = '\0';
+    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
+        if (drv1->protocol_name &&
+            !strcmp(drv1->protocol_name, protocol)) {
+            return drv1;
+        }
+    }
+    return NULL;
+}
+
+static int find_image_format(BlockDriverState *bs, const char *filename,
+                             BlockDriver **pdrv)
+{
+    int score, score_max;
+    BlockDriver *drv1, *drv;
+    uint8_t buf[2048];
+    int ret = 0;
+
+    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
+    if (bs->sg || !bdrv_is_inserted(bs) || bdrv_getlength(bs) == 0) {
+        drv = bdrv_find_format("raw");
+        if (!drv) {
+            ret = -ENOENT;
+        }
+        *pdrv = drv;
+        return ret;
+    }
+
+    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
+    if (ret < 0) {
+        *pdrv = NULL;
+        return ret;
+    }
+
+    score_max = 0;
+    drv = NULL;
+    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
+        if (drv1->bdrv_probe) {
+            score = drv1->bdrv_probe(buf, ret, filename);
+            if (score > score_max) {
+                score_max = score;
+                drv = drv1;
+            }
+        }
+    }
+    if (!drv) {
+        ret = -ENOENT;
+    }
+    *pdrv = drv;
+    return ret;
+}
+
+/**
+ * Set the current 'total_sectors' value
+ */
+static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
+{
+    BlockDriver *drv = bs->drv;
+
+    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
+    if (bs->sg)
+        return 0;
+
+    /* query actual device if possible, otherwise just trust the hint */
+    if (drv->bdrv_getlength) {
+        int64_t length = drv->bdrv_getlength(bs);
+        if (length < 0) {
+            return length;
+        }
+        hint = length >> BDRV_SECTOR_BITS;
+    }
+
+    bs->total_sectors = hint;
+    return 0;
+}
+
+/**
+ * Set open flags for a given discard mode
+ *
+ * Return 0 on success, -1 if the discard mode was invalid.
+ */
+int bdrv_parse_discard_flags(const char *mode, int *flags)
+{
+    *flags &= ~BDRV_O_UNMAP;
+
+    if (!strcmp(mode, "off") || !strcmp(mode, "ignore")) {
+        /* do nothing */
+    } else if (!strcmp(mode, "on") || !strcmp(mode, "unmap")) {
+        *flags |= BDRV_O_UNMAP;
+    } else {
+        return -1;
+    }
+
+    return 0;
+}
+
+/**
+ * Set open flags for a given cache mode
+ *
+ * Return 0 on success, -1 if the cache mode was invalid.
+ */
+int bdrv_parse_cache_flags(const char *mode, int *flags)
+{
+    *flags &= ~BDRV_O_CACHE_MASK;
+
+    if (!strcmp(mode, "off") || !strcmp(mode, "none")) {
+        *flags |= BDRV_O_NOCACHE | BDRV_O_CACHE_WB;
+    } else if (!strcmp(mode, "directsync")) {
+        *flags |= BDRV_O_NOCACHE;
+    } else if (!strcmp(mode, "writeback")) {
+        *flags |= BDRV_O_CACHE_WB;
+    } else if (!strcmp(mode, "unsafe")) {
+        *flags |= BDRV_O_CACHE_WB;
+        *flags |= BDRV_O_NO_FLUSH;
+    } else if (!strcmp(mode, "writethrough")) {
+        /* this is the default */
+    } else {
+        return -1;
+    }
+
+    return 0;
+}
+
+/**
+ * The copy-on-read flag is actually a reference count so multiple users may
+ * use the feature without worrying about clobbering its previous state.
+ * Copy-on-read stays enabled until all users have called to disable it.
+ */
+void bdrv_enable_copy_on_read(BlockDriverState *bs)
+{
+    bs->copy_on_read++;
+}
+
+void bdrv_disable_copy_on_read(BlockDriverState *bs)
+{
+    assert(bs->copy_on_read > 0);
+    bs->copy_on_read--;
+}
+
+static int bdrv_open_flags(BlockDriverState *bs, int flags)
+{
+    int open_flags = flags | BDRV_O_CACHE_WB;
+
+    /*
+     * Clear flags that are internal to the block layer before opening the
+     * image.
+     */
+    open_flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);
+
+    /*
+     * Snapshots should be writable.
+     */
+    if (bs->is_temporary) {
+        open_flags |= BDRV_O_RDWR;
+    }
+
+    return open_flags;
+}
+
+/*
+ * Common part for opening disk images and files
+ *
+ * Removes all processed options from *options.
+ */
+static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
+    QDict *options, int flags, BlockDriver *drv)
+{
+    int ret, open_flags;
+    const char *filename;
+
+    assert(drv != NULL);
+    assert(bs->file == NULL);
+    assert(options != NULL && bs->options != options);
+
+    if (file != NULL) {
+        filename = file->filename;
+    } else {
+        filename = qdict_get_try_str(options, "filename");
+    }
+
+    trace_bdrv_open_common(bs, filename ?: "", flags, drv->format_name);
+
+    /* bdrv_open() with directly using a protocol as drv. This layer is already
+     * opened, so assign it to bs (while file becomes a closed BlockDriverState)
+     * and return immediately. */
+    if (file != NULL && drv->bdrv_file_open) {
+        bdrv_swap(file, bs);
+        return 0;
+    }
+
+    bs->open_flags = flags;
+    bs->buffer_alignment = 512;
+    open_flags = bdrv_open_flags(bs, flags);
+    bs->read_only = !(open_flags & BDRV_O_RDWR);
+
+    if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv, bs->read_only)) {
+        return -ENOTSUP;
+    }
+
+    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
+    if (!bs->read_only && (flags & BDRV_O_COPY_ON_READ)) {
+        bdrv_enable_copy_on_read(bs);
+    }
+
+    if (filename != NULL) {
+        pstrcpy(bs->filename, sizeof(bs->filename), filename);
+    } else {
+        bs->filename[0] = '\0';
+    }
+
+    bs->drv = drv;
+    bs->opaque = g_malloc0(drv->instance_size);
+
+    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
+
+    /* Open the image, either directly or using a protocol */
+    if (drv->bdrv_file_open) {
+        assert(file == NULL);
+        assert(drv->bdrv_parse_filename || filename != NULL);
+        ret = drv->bdrv_file_open(bs, options, open_flags);
+    } else {
+        if (file == NULL) {
+            qerror_report(ERROR_CLASS_GENERIC_ERROR, "Can't use '%s' as a "
+                          "block driver for the protocol level",
+                          drv->format_name);
+            ret = -EINVAL;
+            goto free_and_fail;
+        }
+        assert(file != NULL);
+        bs->file = file;
+        ret = drv->bdrv_open(bs, options, open_flags);
+    }
+
+    if (ret < 0) {
+        goto free_and_fail;
+    }
+
+    ret = refresh_total_sectors(bs, bs->total_sectors);
+    if (ret < 0) {
+        goto free_and_fail;
+    }
+
+#ifndef _WIN32
+    if (bs->is_temporary) {
+        assert(filename != NULL);
+        unlink(filename);
+    }
+#endif
+    return 0;
+
+free_and_fail:
+    bs->file = NULL;
+    g_free(bs->opaque);
+    bs->opaque = NULL;
+    bs->drv = NULL;
+    return ret;
+}
+
+/*
+ * Opens a file using a protocol (file, host_device, nbd, ...)
+ *
+ * options is a QDict of options to pass to the block drivers, or NULL for an
+ * empty set of options. The reference to the QDict belongs to the block layer
+ * after the call (even on failure), so if the caller intends to reuse the
+ * dictionary, it needs to use QINCREF() before calling bdrv_file_open.
+ */
+int bdrv_file_open(BlockDriverState **pbs, const char *filename,
+                   QDict *options, int flags)
+{
+    BlockDriverState *bs;
+    BlockDriver *drv;
+    const char *drvname;
+    bool allow_protocol_prefix = false;
+    int ret;
+
+    /* NULL means an empty set of options */
+    if (options == NULL) {
+        options = qdict_new();
+    }
+
+    bs = bdrv_new("");
+    bs->options = options;
+    options = qdict_clone_shallow(options);
+
+    /* Fetch the file name from the options QDict if necessary */
+    if (!filename) {
+        filename = qdict_get_try_str(options, "filename");
+    } else if (filename && !qdict_haskey(options, "filename")) {
+        qdict_put(options, "filename", qstring_from_str(filename));
+        allow_protocol_prefix = true;
+    } else {
+        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Can't specify 'file' and "
+                      "'filename' options at the same time");
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    /* Find the right block driver */
+    drvname = qdict_get_try_str(options, "driver");
+    if (drvname) {
+        drv = bdrv_find_whitelisted_format(drvname, !(flags & BDRV_O_RDWR));
+        qdict_del(options, "driver");
+    } else if (filename) {
+        drv = bdrv_find_protocol(filename, allow_protocol_prefix);
+        if (!drv) {
+            qerror_report(ERROR_CLASS_GENERIC_ERROR, "Unknown protocol");
+        }
+    } else {
+        qerror_report(ERROR_CLASS_GENERIC_ERROR,
+                      "Must specify either driver or file");
+        drv = NULL;
+    }
+
+    if (!drv) {
+        ret = -ENOENT;
+        goto fail;
+    }
+
+    /* Parse the filename and open it */
+    if (drv->bdrv_parse_filename && filename) {
+        Error *local_err = NULL;
+        drv->bdrv_parse_filename(filename, options, &local_err);
+        if (error_is_set(&local_err)) {
+            qerror_report_err(local_err);
+            error_free(local_err);
+            ret = -EINVAL;
+            goto fail;
+        }
+        qdict_del(options, "filename");
+    } else if (!drv->bdrv_parse_filename && !filename) {
+        qerror_report(ERROR_CLASS_GENERIC_ERROR,
+                      "The '%s' block driver requires a file name",
+                      drv->format_name);
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    ret = bdrv_open_common(bs, NULL, options, flags, drv);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* Check if any unknown options were used */
+    if (qdict_size(options) != 0) {
+        const QDictEntry *entry = qdict_first(options);
+        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Block protocol '%s' doesn't "
+                      "support the option '%s'",
+                      drv->format_name, entry->key);
+        ret = -EINVAL;
+        goto fail;
+    }
+    QDECREF(options);
+
+    bs->growable = 1;
+    *pbs = bs;
+    return 0;
+
+fail:
+    QDECREF(options);
+    if (!bs->drv) {
+        QDECREF(bs->options);
+    }
+    bdrv_delete(bs);
+    return ret;
+}
+
+/*
+ * Opens the backing file for a BlockDriverState if not yet open
+ *
+ * options is a QDict of options to pass to the block drivers, or NULL for an
+ * empty set of options. The reference to the QDict is transferred to this
+ * function (even on failure), so if the caller intends to reuse the dictionary,
+ * it needs to use QINCREF() before calling bdrv_file_open.
+ */
+int bdrv_open_backing_file(BlockDriverState *bs, QDict *options)
+{
+    char backing_filename[PATH_MAX];
+    int back_flags, ret;
+    BlockDriver *back_drv = NULL;
+
+    if (bs->backing_hd != NULL) {
+        QDECREF(options);
+        return 0;
+    }
+
+    /* NULL means an empty set of options */
+    if (options == NULL) {
+        options = qdict_new();
+    }
+
+    bs->open_flags &= ~BDRV_O_NO_BACKING;
+    if (qdict_haskey(options, "file.filename")) {
+        backing_filename[0] = '\0';
+    } else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
+        QDECREF(options);
+        return 0;
+    }
+
+    bs->backing_hd = bdrv_new("");
+    bdrv_get_full_backing_filename(bs, backing_filename,
+                                   sizeof(backing_filename));
+
+    if (bs->backing_format[0] != '\0') {
+        back_drv = bdrv_find_format(bs->backing_format);
+    }
+
+    /* backing files always opened read-only */
+    back_flags = bs->open_flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT);
+
+    ret = bdrv_open(bs->backing_hd,
+                    *backing_filename ? backing_filename : NULL, options,
+                    back_flags, back_drv);
+    if (ret < 0) {
+        bdrv_delete(bs->backing_hd);
+        bs->backing_hd = NULL;
+        bs->open_flags |= BDRV_O_NO_BACKING;
+        return ret;
+    }
+    return 0;
+}
+
+static void extract_subqdict(QDict *src, QDict **dst, const char *start)
+{
+    const QDictEntry *entry, *next;
+    const char *p;
+
+    *dst = qdict_new();
+    entry = qdict_first(src);
+
+    while (entry != NULL) {
+        next = qdict_next(src, entry);
+        if (strstart(entry->key, start, &p)) {
+            qobject_incref(entry->value);
+            qdict_put_obj(*dst, p, entry->value);
+            qdict_del(src, entry->key);
+        }
+        entry = next;
+    }
+}
+
+/*
+ * Opens a disk image (raw, qcow2, vmdk, ...)
+ *
+ * options is a QDict of options to pass to the block drivers, or NULL for an
+ * empty set of options. The reference to the QDict belongs to the block layer
+ * after the call (even on failure), so if the caller intends to reuse the
+ * dictionary, it needs to use QINCREF() before calling bdrv_open.
+ */
+int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
+              int flags, BlockDriver *drv)
+{
+    int ret;
+    /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
+    char tmp_filename[PATH_MAX + 1];
+    BlockDriverState *file = NULL;
+    QDict *file_options = NULL;
+
+    /* NULL means an empty set of options */
+    if (options == NULL) {
+        options = qdict_new();
+    }
+
+    bs->options = options;
+    options = qdict_clone_shallow(options);
+
+    /* For snapshot=on, create a temporary qcow2 overlay */
+    if (flags & BDRV_O_SNAPSHOT) {
+        BlockDriverState *bs1;
+        int64_t total_size;
+        BlockDriver *bdrv_qcow2;
+        QEMUOptionParameter *create_options;
+        char backing_filename[PATH_MAX];
+
+        if (qdict_size(options) != 0) {
+            error_report("Can't use snapshot=on with driver-specific options");
+            ret = -EINVAL;
+            goto fail;
+        }
+        assert(filename != NULL);
+
+        /* if snapshot, we create a temporary backing file and open it
+           instead of opening 'filename' directly */
+
+        /* if there is a backing file, use it */
+        bs1 = bdrv_new("");
+        ret = bdrv_open(bs1, filename, NULL, 0, drv);
+        if (ret < 0) {
+            bdrv_delete(bs1);
+            goto fail;
+        }
+        total_size = bdrv_getlength(bs1) & BDRV_SECTOR_MASK;
+
+        bdrv_delete(bs1);
+
+        ret = get_tmp_filename(tmp_filename, sizeof(tmp_filename));
+        if (ret < 0) {
+            goto fail;
+        }
+
+        /* Real path is meaningless for protocols */
+        if (path_has_protocol(filename)) {
+            snprintf(backing_filename, sizeof(backing_filename),
+                     "%s", filename);
+        } else if (!realpath(filename, backing_filename)) {
+            ret = -errno;
+            goto fail;
+        }
+
+        bdrv_qcow2 = bdrv_find_format("qcow2");
+        create_options = parse_option_parameters("", bdrv_qcow2->create_options,
+                                                 NULL);
+
+        set_option_parameter_int(create_options, BLOCK_OPT_SIZE, total_size);
+        set_option_parameter(create_options, BLOCK_OPT_BACKING_FILE,
+                             backing_filename);
+        if (drv) {
+            set_option_parameter(create_options, BLOCK_OPT_BACKING_FMT,
+                drv->format_name);
+        }
+
+        ret = bdrv_create(bdrv_qcow2, tmp_filename, create_options);
+        free_option_parameters(create_options);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        filename = tmp_filename;
+        drv = bdrv_qcow2;
+        bs->is_temporary = 1;
+    }
+
+    /* Open image file without format layer */
+    if (flags & BDRV_O_RDWR) {
+        flags |= BDRV_O_ALLOW_RDWR;
+    }
+
+    extract_subqdict(options, &file_options, "file.");
+
+    ret = bdrv_file_open(&file, filename, file_options,
+                         bdrv_open_flags(bs, flags | BDRV_O_UNMAP));
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* Find the right image format driver */
+    if (!drv) {
+        ret = find_image_format(file, filename, &drv);
+    }
+
+    if (!drv) {
+        goto unlink_and_fail;
+    }
+
+    /* Open the image */
+    ret = bdrv_open_common(bs, file, options, flags, drv);
+    if (ret < 0) {
+        goto unlink_and_fail;
+    }
+
+    if (bs->file != file) {
+        bdrv_delete(file);
+        file = NULL;
+    }
+
+    /* If there is a backing file, use it */
+    if ((flags & BDRV_O_NO_BACKING) == 0) {
+        QDict *backing_options;
+
+        extract_subqdict(options, &backing_options, "backing.");
+        ret = bdrv_open_backing_file(bs, backing_options);
+        if (ret < 0) {
+            goto close_and_fail;
+        }
+    }
+
+    /* Check if any unknown options were used */
+    if (qdict_size(options) != 0) {
+        const QDictEntry *entry = qdict_first(options);
+        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Block format '%s' used by "
+            "device '%s' doesn't support the option '%s'",
+            drv->format_name, bs->device_name, entry->key);
+
+        ret = -EINVAL;
+        goto close_and_fail;
+    }
+    QDECREF(options);
+
+    if (!bdrv_key_required(bs)) {
+        bdrv_dev_change_media_cb(bs, true);
+    }
+
+    /* throttling disk I/O limits */
+    if (bs->io_limits_enabled) {
+        bdrv_io_limits_enable(bs);
+    }
+
+    return 0;
+
+unlink_and_fail:
+    if (file != NULL) {
+        bdrv_delete(file);
+    }
+    if (bs->is_temporary) {
+        unlink(filename);
+    }
+fail:
+    QDECREF(bs->options);
+    QDECREF(options);
+    bs->options = NULL;
+    return ret;
+
+close_and_fail:
+    bdrv_close(bs);
+    QDECREF(options);
+    return ret;
+}
+
+typedef struct BlockReopenQueueEntry {
+     bool prepared;
+     BDRVReopenState state;
+     QSIMPLEQ_ENTRY(BlockReopenQueueEntry) entry;
+} BlockReopenQueueEntry;
+
+/*
+ * Adds a BlockDriverState to a simple queue for an atomic, transactional
+ * reopen of multiple devices.
+ *
+ * bs_queue can either be an existing BlockReopenQueue that has had QSIMPLE_INIT
+ * already performed, or alternatively may be NULL a new BlockReopenQueue will
+ * be created and initialized. This newly created BlockReopenQueue should be
+ * passed back in for subsequent calls that are intended to be of the same
+ * atomic 'set'.
+ *
+ * bs is the BlockDriverState to add to the reopen queue.
+ *
+ * flags contains the open flags for the associated bs
+ *
+ * returns a pointer to bs_queue, which is either the newly allocated
+ * bs_queue, or the existing bs_queue being used.
+ *
+ */
+BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue,
+                                    BlockDriverState *bs, int flags)
+{
+    assert(bs != NULL);
+
+    BlockReopenQueueEntry *bs_entry;
+    if (bs_queue == NULL) {
+        bs_queue = g_new0(BlockReopenQueue, 1);
+        QSIMPLEQ_INIT(bs_queue);
+    }
+
+    if (bs->file) {
+        bdrv_reopen_queue(bs_queue, bs->file, flags);
+    }
+
+    bs_entry = g_new0(BlockReopenQueueEntry, 1);
+    QSIMPLEQ_INSERT_TAIL(bs_queue, bs_entry, entry);
+
+    bs_entry->state.bs = bs;
+    bs_entry->state.flags = flags;
+
+    return bs_queue;
+}
+
+/*
+ * Reopen multiple BlockDriverStates atomically & transactionally.
+ *
+ * The queue passed in (bs_queue) must have been built up previous
+ * via bdrv_reopen_queue().
+ *
+ * Reopens all BDS specified in the queue, with the appropriate
+ * flags.  All devices are prepared for reopen, and failure of any
+ * device will cause all device changes to be abandonded, and intermediate
+ * data cleaned up.
+ *
+ * If all devices prepare successfully, then the changes are committed
+ * to all devices.
+ *
+ */
+int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp)
+{
+    int ret = -1;
+    BlockReopenQueueEntry *bs_entry, *next;
+    Error *local_err = NULL;
+
+    assert(bs_queue != NULL);
+
+    bdrv_drain_all();
+
+    QSIMPLEQ_FOREACH(bs_entry, bs_queue, entry) {
+        if (bdrv_reopen_prepare(&bs_entry->state, bs_queue, &local_err)) {
+            error_propagate(errp, local_err);
+            goto cleanup;
+        }
+        bs_entry->prepared = true;
+    }
+
+    /* If we reach this point, we have success and just need to apply the
+     * changes
+     */
+    QSIMPLEQ_FOREACH(bs_entry, bs_queue, entry) {
+        bdrv_reopen_commit(&bs_entry->state);
+    }
+
+    ret = 0;
+
+cleanup:
+    QSIMPLEQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) {
+        if (ret && bs_entry->prepared) {
+            bdrv_reopen_abort(&bs_entry->state);
+        }
+        g_free(bs_entry);
+    }
+    g_free(bs_queue);
+    return ret;
+}
+
+
+/* Reopen a single BlockDriverState with the specified flags. */
+int bdrv_reopen(BlockDriverState *bs, int bdrv_flags, Error **errp)
+{
+    int ret = -1;
+    Error *local_err = NULL;
+    BlockReopenQueue *queue = bdrv_reopen_queue(NULL, bs, bdrv_flags);
+
+    ret = bdrv_reopen_multiple(queue, &local_err);
+    if (local_err != NULL) {
+        error_propagate(errp, local_err);
+    }
+    return ret;
+}
+
+
+/*
+ * Prepares a BlockDriverState for reopen. All changes are staged in the
+ * 'opaque' field of the BDRVReopenState, which is used and allocated by
+ * the block driver layer .bdrv_reopen_prepare()
+ *
+ * bs is the BlockDriverState to reopen
+ * flags are the new open flags
+ * queue is the reopen queue
+ *
+ * Returns 0 on success, non-zero on error.  On error errp will be set
+ * as well.
+ *
+ * On failure, bdrv_reopen_abort() will be called to clean up any data.
+ * It is the responsibility of the caller to then call the abort() or
+ * commit() for any other BDS that have been left in a prepare() state
+ *
+ */
+int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue,
+                        Error **errp)
+{
+    int ret = -1;
+    Error *local_err = NULL;
+    BlockDriver *drv;
+
+    assert(reopen_state != NULL);
+    assert(reopen_state->bs->drv != NULL);
+    drv = reopen_state->bs->drv;
+
+    /* if we are to stay read-only, do not allow permission change
+     * to r/w */
+    if (!(reopen_state->bs->open_flags & BDRV_O_ALLOW_RDWR) &&
+        reopen_state->flags & BDRV_O_RDWR) {
+        error_set(errp, QERR_DEVICE_IS_READ_ONLY,
+                  reopen_state->bs->device_name);
+        goto error;
+    }
+
+
+    ret = bdrv_flush(reopen_state->bs);
+    if (ret) {
+        error_set(errp, ERROR_CLASS_GENERIC_ERROR, "Error (%s) flushing drive",
+                  strerror(-ret));
+        goto error;
+    }
+
+    if (drv->bdrv_reopen_prepare) {
+        ret = drv->bdrv_reopen_prepare(reopen_state, queue, &local_err);
+        if (ret) {
+            if (local_err != NULL) {
+                error_propagate(errp, local_err);
+            } else {
+                error_setg(errp, "failed while preparing to reopen image '%s'",
+                           reopen_state->bs->filename);
+            }
+            goto error;
+        }
+    } else {
+        /* It is currently mandatory to have a bdrv_reopen_prepare()
+         * handler for each supported drv. */
+        error_set(errp, QERR_BLOCK_FORMAT_FEATURE_NOT_SUPPORTED,
+                  drv->format_name, reopen_state->bs->device_name,
+                 "reopening of file");
+        ret = -1;
+        goto error;
+    }
+
+    ret = 0;
+
+error:
+    return ret;
+}
+
+/*
+ * Takes the staged changes for the reopen from bdrv_reopen_prepare(), and
+ * makes them final by swapping the staging BlockDriverState contents into
+ * the active BlockDriverState contents.
+ */
+void bdrv_reopen_commit(BDRVReopenState *reopen_state)
+{
+    BlockDriver *drv;
+
+    assert(reopen_state != NULL);
+    drv = reopen_state->bs->drv;
+    assert(drv != NULL);
+
+    /* If there are any driver level actions to take */
+    if (drv->bdrv_reopen_commit) {
+        drv->bdrv_reopen_commit(reopen_state);
+    }
+
+    /* set BDS specific flags now */
+    reopen_state->bs->open_flags         = reopen_state->flags;
+    reopen_state->bs->enable_write_cache = !!(reopen_state->flags &
+                                              BDRV_O_CACHE_WB);
+    reopen_state->bs->read_only = !(reopen_state->flags & BDRV_O_RDWR);
+}
+
+/*
+ * Abort the reopen, and delete and free the staged changes in
+ * reopen_state
+ */
+void bdrv_reopen_abort(BDRVReopenState *reopen_state)
+{
+    BlockDriver *drv;
+
+    assert(reopen_state != NULL);
+    drv = reopen_state->bs->drv;
+    assert(drv != NULL);
+
+    if (drv->bdrv_reopen_abort) {
+        drv->bdrv_reopen_abort(reopen_state);
+    }
+}
+
+
+void bdrv_close(BlockDriverState *bs)
+{
+    if (bs->job) {
+        block_job_cancel_sync(bs->job);
+    }
+    bdrv_drain_all(); /* complete I/O */
+    bdrv_flush(bs);
+    bdrv_drain_all(); /* in case flush left pending I/O */
+    notifier_list_notify(&bs->close_notifiers, bs);
+
+    if (bs->drv) {
+        if (bs->backing_hd) {
+            bdrv_delete(bs->backing_hd);
+            bs->backing_hd = NULL;
+        }
+        bs->drv->bdrv_close(bs);
+        g_free(bs->opaque);
+#ifdef _WIN32
+        if (bs->is_temporary) {
+            unlink(bs->filename);
+        }
+#endif
+        bs->opaque = NULL;
+        bs->drv = NULL;
+        bs->copy_on_read = 0;
+        bs->backing_file[0] = '\0';
+        bs->backing_format[0] = '\0';
+        bs->total_sectors = 0;
+        bs->encrypted = 0;
+        bs->valid_key = 0;
+        bs->sg = 0;
+        bs->growable = 0;
+        QDECREF(bs->options);
+        bs->options = NULL;
+
+        if (bs->file != NULL) {
+            bdrv_delete(bs->file);
+            bs->file = NULL;
+        }
+    }
+
+    bdrv_dev_change_media_cb(bs, false);
+
+    /*throttling disk I/O limits*/
+    if (bs->io_limits_enabled) {
+        bdrv_io_limits_disable(bs);
+    }
+}
+
+void bdrv_close_all(void)
+{
+    BlockDriverState *bs;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        bdrv_close(bs);
+    }
+}
+
+/*
+ * Wait for pending requests to complete across all BlockDriverStates
+ *
+ * This function does not flush data to disk, use bdrv_flush_all() for that
+ * after calling this function.
+ *
+ * Note that completion of an asynchronous I/O operation can trigger any
+ * number of other I/O operations on other devices---for example a coroutine
+ * can be arbitrarily complex and a constant flow of I/O can come until the
+ * coroutine is complete.  Because of this, it is not possible to have a
+ * function to drain a single device's I/O queue.
+ */
+void bdrv_drain_all(void)
+{
+    BlockDriverState *bs;
+    bool busy;
+
+    do {
+        busy = qemu_aio_wait();
+
+        /* FIXME: We do not have timer support here, so this is effectively
+         * a busy wait.
+         */
+        QTAILQ_FOREACH(bs, &bdrv_states, list) {
+            if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
+                qemu_co_queue_restart_all(&bs->throttled_reqs);
+                busy = true;
+            }
+        }
+    } while (busy);
+
+    /* If requests are still pending there is a bug somewhere */
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        assert(QLIST_EMPTY(&bs->tracked_requests));
+        assert(qemu_co_queue_empty(&bs->throttled_reqs));
+    }
+}
+
+/* make a BlockDriverState anonymous by removing from bdrv_state list.
+   Also, NULL terminate the device_name to prevent double remove */
+void bdrv_make_anon(BlockDriverState *bs)
+{
+    if (bs->device_name[0] != '\0') {
+        QTAILQ_REMOVE(&bdrv_states, bs, list);
+    }
+    bs->device_name[0] = '\0';
+}
+
+static void bdrv_rebind(BlockDriverState *bs)
+{
+    if (bs->drv && bs->drv->bdrv_rebind) {
+        bs->drv->bdrv_rebind(bs);
+    }
+}
+
+static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
+                                     BlockDriverState *bs_src)
+{
+    /* move some fields that need to stay attached to the device */
+    bs_dest->open_flags         = bs_src->open_flags;
+
+    /* dev info */
+    bs_dest->dev_ops            = bs_src->dev_ops;
+    bs_dest->dev_opaque         = bs_src->dev_opaque;
+    bs_dest->dev                = bs_src->dev;
+    bs_dest->buffer_alignment   = bs_src->buffer_alignment;
+    bs_dest->copy_on_read       = bs_src->copy_on_read;
+
+    bs_dest->enable_write_cache = bs_src->enable_write_cache;
+
+    /* i/o timing parameters */
+    bs_dest->slice_start        = bs_src->slice_start;
+    bs_dest->slice_end          = bs_src->slice_end;
+    bs_dest->slice_submitted    = bs_src->slice_submitted;
+    bs_dest->io_limits          = bs_src->io_limits;
+    bs_dest->throttled_reqs     = bs_src->throttled_reqs;
+    bs_dest->block_timer        = bs_src->block_timer;
+    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
+
+    /* r/w error */
+    bs_dest->on_read_error      = bs_src->on_read_error;
+    bs_dest->on_write_error     = bs_src->on_write_error;
+
+    /* i/o status */
+    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
+    bs_dest->iostatus           = bs_src->iostatus;
+
+    /* dirty bitmap */
+    bs_dest->dirty_bitmap       = bs_src->dirty_bitmap;
+
+    /* job */
+    bs_dest->in_use             = bs_src->in_use;
+    bs_dest->job                = bs_src->job;
+
+    /* keep the same entry in bdrv_states */
+    pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
+            bs_src->device_name);
+    bs_dest->list = bs_src->list;
+}
+
+/*
+ * Swap bs contents for two image chains while they are live,
+ * while keeping required fields on the BlockDriverState that is
+ * actually attached to a device.
+ *
+ * This will modify the BlockDriverState fields, and swap contents
+ * between bs_new and bs_old. Both bs_new and bs_old are modified.
+ *
+ * bs_new is required to be anonymous.
+ *
+ * This function does not create any image files.
+ */
+void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old)
+{
+    BlockDriverState tmp;
+
+    /* bs_new must be anonymous and shouldn't have anything fancy enabled */
+    assert(bs_new->device_name[0] == '\0');
+    assert(bs_new->dirty_bitmap == NULL);
+    assert(bs_new->job == NULL);
+    assert(bs_new->dev == NULL);
+    assert(bs_new->in_use == 0);
+    assert(bs_new->io_limits_enabled == false);
+    assert(bs_new->block_timer == NULL);
+
+    tmp = *bs_new;
+    *bs_new = *bs_old;
+    *bs_old = tmp;
+
+    /* there are some fields that should not be swapped, move them back */
+    bdrv_move_feature_fields(&tmp, bs_old);
+    bdrv_move_feature_fields(bs_old, bs_new);
+    bdrv_move_feature_fields(bs_new, &tmp);
+
+    /* bs_new shouldn't be in bdrv_states even after the swap!  */
+    assert(bs_new->device_name[0] == '\0');
+
+    /* Check a few fields that should remain attached to the device */
+    assert(bs_new->dev == NULL);
+    assert(bs_new->job == NULL);
+    assert(bs_new->in_use == 0);
+    assert(bs_new->io_limits_enabled == false);
+    assert(bs_new->block_timer == NULL);
+
+    bdrv_rebind(bs_new);
+    bdrv_rebind(bs_old);
+}
+
+/*
+ * Add new bs contents at the top of an image chain while the chain is
+ * live, while keeping required fields on the top layer.
+ *
+ * This will modify the BlockDriverState fields, and swap contents
+ * between bs_new and bs_top. Both bs_new and bs_top are modified.
+ *
+ * bs_new is required to be anonymous.
+ *
+ * This function does not create any image files.
+ */
+void bdrv_append(BlockDriverState *bs_new, BlockDriverState *bs_top)
+{
+    bdrv_swap(bs_new, bs_top);
+
+    /* The contents of 'tmp' will become bs_top, as we are
+     * swapping bs_new and bs_top contents. */
+    bs_top->backing_hd = bs_new;
+    bs_top->open_flags &= ~BDRV_O_NO_BACKING;
+    pstrcpy(bs_top->backing_file, sizeof(bs_top->backing_file),
+            bs_new->filename);
+    pstrcpy(bs_top->backing_format, sizeof(bs_top->backing_format),
+            bs_new->drv ? bs_new->drv->format_name : "");
+}
+
+void bdrv_delete(BlockDriverState *bs)
+{
+    assert(!bs->dev);
+    assert(!bs->job);
+    assert(!bs->in_use);
+
+    /* remove from list, if necessary */
+    bdrv_make_anon(bs);
+
+    bdrv_close(bs);
+
+    g_free(bs);
+}
+
+int bdrv_attach_dev(BlockDriverState *bs, void *dev)
+/* TODO change to DeviceState *dev when all users are qdevified */
+{
+    if (bs->dev) {
+        return -EBUSY;
+    }
+    bs->dev = dev;
+    bdrv_iostatus_reset(bs);
+    return 0;
+}
+
+/* TODO qdevified devices don't use this, remove when devices are qdevified */
+void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
+{
+    if (bdrv_attach_dev(bs, dev) < 0) {
+        abort();
+    }
+}
+
+void bdrv_detach_dev(BlockDriverState *bs, void *dev)
+/* TODO change to DeviceState *dev when all users are qdevified */
+{
+    assert(bs->dev == dev);
+    bs->dev = NULL;
+    bs->dev_ops = NULL;
+    bs->dev_opaque = NULL;
+    bs->buffer_alignment = 512;
+}
+
+/* TODO change to return DeviceState * when all users are qdevified */
+void *bdrv_get_attached_dev(BlockDriverState *bs)
+{
+    return bs->dev;
+}
+
+void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
+                      void *opaque)
+{
+    bs->dev_ops = ops;
+    bs->dev_opaque = opaque;
+}
+
+void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
+                               enum MonitorEvent ev,
+                               BlockErrorAction action, bool is_read)
+{
+    QObject *data;
+    const char *action_str;
+
+    switch (action) {
+    case BDRV_ACTION_REPORT:
+        action_str = "report";
+        break;
+    case BDRV_ACTION_IGNORE:
+        action_str = "ignore";
+        break;
+    case BDRV_ACTION_STOP:
+        action_str = "stop";
+        break;
+    default:
+        abort();
+    }
+
+    data = qobject_from_jsonf("{ 'device': %s, 'action': %s, 'operation': %s }",
+                              bdrv->device_name,
+                              action_str,
+                              is_read ? "read" : "write");
+    monitor_protocol_event(ev, data);
+
+    qobject_decref(data);
+}
+
+static void bdrv_emit_qmp_eject_event(BlockDriverState *bs, bool ejected)
+{
+    QObject *data;
+
+    data = qobject_from_jsonf("{ 'device': %s, 'tray-open': %i }",
+                              bdrv_get_device_name(bs), ejected);
+    monitor_protocol_event(QEVENT_DEVICE_TRAY_MOVED, data);
+
+    qobject_decref(data);
+}
+
+static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
+{
+    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
+        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
+        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
+        if (tray_was_closed) {
+            /* tray open */
+            bdrv_emit_qmp_eject_event(bs, true);
+        }
+        if (load) {
+            /* tray close */
+            bdrv_emit_qmp_eject_event(bs, false);
+        }
+    }
+}
+
+bool bdrv_dev_has_removable_media(BlockDriverState *bs)
+{
+    return !bs->dev || (bs->dev_ops && bs->dev_ops->change_media_cb);
+}
+
+void bdrv_dev_eject_request(BlockDriverState *bs, bool force)
+{
+    if (bs->dev_ops && bs->dev_ops->eject_request_cb) {
+        bs->dev_ops->eject_request_cb(bs->dev_opaque, force);
+    }
+}
+
+bool bdrv_dev_is_tray_open(BlockDriverState *bs)
+{
+    if (bs->dev_ops && bs->dev_ops->is_tray_open) {
+        return bs->dev_ops->is_tray_open(bs->dev_opaque);
+    }
+    return false;
+}
+
+static void bdrv_dev_resize_cb(BlockDriverState *bs)
+{
+    if (bs->dev_ops && bs->dev_ops->resize_cb) {
+        bs->dev_ops->resize_cb(bs->dev_opaque);
+    }
+}
+
+bool bdrv_dev_is_medium_locked(BlockDriverState *bs)
+{
+    if (bs->dev_ops && bs->dev_ops->is_medium_locked) {
+        return bs->dev_ops->is_medium_locked(bs->dev_opaque);
+    }
+    return false;
+}
+
+/*
+ * Run consistency checks on an image
+ *
+ * Returns 0 if the check could be completed (it doesn't mean that the image is
+ * free of errors) or -errno when an internal error occurred. The results of the
+ * check are stored in res.
+ */
+int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
+{
+    if (bs->drv->bdrv_check == NULL) {
+        return -ENOTSUP;
+    }
+
+    memset(res, 0, sizeof(*res));
+    return bs->drv->bdrv_check(bs, res, fix);
+}
+
+#define COMMIT_BUF_SECTORS 2048
+
+/* commit COW file into the raw image */
+int bdrv_commit(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+    int64_t sector, total_sectors;
+    int n, ro, open_flags;
+    int ret = 0;
+    uint8_t *buf;
+    char filename[PATH_MAX];
+
+    if (!drv)
+        return -ENOMEDIUM;
+    
+    if (!bs->backing_hd) {
+        return -ENOTSUP;
+    }
+
+    if (bdrv_in_use(bs) || bdrv_in_use(bs->backing_hd)) {
+        return -EBUSY;
+    }
+
+    ro = bs->backing_hd->read_only;
+    /* Use pstrcpy (not strncpy): filename must be NUL-terminated. */
+    pstrcpy(filename, sizeof(filename), bs->backing_hd->filename);
+    open_flags =  bs->backing_hd->open_flags;
+
+    if (ro) {
+        if (bdrv_reopen(bs->backing_hd, open_flags | BDRV_O_RDWR, NULL)) {
+            return -EACCES;
+        }
+    }
+
+    total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
+    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
+
+    for (sector = 0; sector < total_sectors; sector += n) {
+        if (bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n)) {
+
+            if (bdrv_read(bs, sector, buf, n) != 0) {
+                ret = -EIO;
+                goto ro_cleanup;
+            }
+
+            if (bdrv_write(bs->backing_hd, sector, buf, n) != 0) {
+                ret = -EIO;
+                goto ro_cleanup;
+            }
+        }
+    }
+
+    if (drv->bdrv_make_empty) {
+        ret = drv->bdrv_make_empty(bs);
+        bdrv_flush(bs);
+    }
+
+    /*
+     * Make sure all data we wrote to the backing device is actually
+     * stable on disk.
+     */
+    if (bs->backing_hd)
+        bdrv_flush(bs->backing_hd);
+
+ro_cleanup:
+    g_free(buf);
+
+    if (ro) {
+        /* ignoring error return here */
+        bdrv_reopen(bs->backing_hd, open_flags & ~BDRV_O_RDWR, NULL);
+    }
+
+    return ret;
+}
+
+int bdrv_commit_all(void)
+{
+    BlockDriverState *bs;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        if (bs->drv && bs->backing_hd) {
+            int ret = bdrv_commit(bs);
+            if (ret < 0) {
+                return ret;
+            }
+        }
+    }
+    return 0;
+}
+
+/**
+ * Remove an active request from the tracked requests list
+ *
+ * This function should be called when a tracked request is completing.
+ */
+static void tracked_request_end(BdrvTrackedRequest *req)
+{
+    QLIST_REMOVE(req, list);
+    qemu_co_queue_restart_all(&req->wait_queue);
+}
+
+/**
+ * Add an active request to the tracked requests list
+ */
+static void tracked_request_begin(BdrvTrackedRequest *req,
+                                  BlockDriverState *bs,
+                                  int64_t sector_num,
+                                  int nb_sectors, bool is_write)
+{
+    *req = (BdrvTrackedRequest){
+        .bs = bs,
+        .sector_num = sector_num,
+        .nb_sectors = nb_sectors,
+        .is_write = is_write,
+        .co = qemu_coroutine_self(),
+    };
+
+    qemu_co_queue_init(&req->wait_queue);
+
+    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
+}
+
+/**
+ * Round a region to cluster boundaries
+ */
+void bdrv_round_to_clusters(BlockDriverState *bs,
+                            int64_t sector_num, int nb_sectors,
+                            int64_t *cluster_sector_num,
+                            int *cluster_nb_sectors)
+{
+    BlockDriverInfo bdi;
+
+    if (bdrv_get_info(bs, &bdi) < 0 || bdi.cluster_size == 0) {
+        *cluster_sector_num = sector_num;
+        *cluster_nb_sectors = nb_sectors;
+    } else {
+        int64_t c = bdi.cluster_size / BDRV_SECTOR_SIZE;
+        *cluster_sector_num = QEMU_ALIGN_DOWN(sector_num, c);
+        *cluster_nb_sectors = QEMU_ALIGN_UP(sector_num - *cluster_sector_num +
+                                            nb_sectors, c);
+    }
+}
+
+static bool tracked_request_overlaps(BdrvTrackedRequest *req,
+                                     int64_t sector_num, int nb_sectors) {
+    /*        aaaa   bbbb */
+    if (sector_num >= req->sector_num + req->nb_sectors) {
+        return false;
+    }
+    /* bbbb   aaaa        */
+    if (req->sector_num >= sector_num + nb_sectors) {
+        return false;
+    }
+    return true;
+}
+
+static void coroutine_fn wait_for_overlapping_requests(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors)
+{
+    BdrvTrackedRequest *req;
+    int64_t cluster_sector_num;
+    int cluster_nb_sectors;
+    bool retry;
+
+    /* If we touch the same cluster it counts as an overlap.  This guarantees
+     * that allocating writes will be serialized and not race with each other
+     * for the same cluster.  For example, in copy-on-read it ensures that the
+     * CoR read and write operations are atomic and guest writes cannot
+     * interleave between them.
+     */
+    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
+                           &cluster_sector_num, &cluster_nb_sectors);
+
+    do {
+        retry = false;
+        QLIST_FOREACH(req, &bs->tracked_requests, list) {
+            if (tracked_request_overlaps(req, cluster_sector_num,
+                                         cluster_nb_sectors)) {
+                /* Hitting this means there was a reentrant request, for
+                 * example, a block driver issuing nested requests.  This must
+                 * never happen since it means deadlock.
+                 */
+                assert(qemu_coroutine_self() != req->co);
+
+                qemu_co_queue_wait(&req->wait_queue);
+                retry = true;
+                break;
+            }
+        }
+    } while (retry);
+}
+
+/*
+ * Return values:
+ * 0        - success
+ * -EINVAL  - backing format specified, but no file
+ * -ENOSPC  - can't update the backing file because no space is left in the
+ *            image file header
+ * -ENOTSUP - format driver doesn't support changing the backing file
+ */
+int bdrv_change_backing_file(BlockDriverState *bs,
+    const char *backing_file, const char *backing_fmt)
+{
+    BlockDriver *drv = bs->drv;
+    int ret;
+
+    /* Backing file format doesn't make sense without a backing file */
+    if (backing_fmt && !backing_file) {
+        return -EINVAL;
+    }
+
+    if (drv->bdrv_change_backing_file != NULL) {
+        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
+    } else {
+        ret = -ENOTSUP;
+    }
+
+    if (ret == 0) {
+        pstrcpy(bs->backing_file, sizeof(bs->backing_file), backing_file ?: "");
+        pstrcpy(bs->backing_format, sizeof(bs->backing_format), backing_fmt ?: "");
+    }
+    return ret;
+}
+
+/*
+ * Finds the image layer in the chain that has 'bs' as its backing file.
+ *
+ * active is the current topmost image.
+ *
+ * Returns NULL if bs is not found in active's image chain,
+ * or if active == bs.
+ */
+BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
+                                    BlockDriverState *bs)
+{
+    BlockDriverState *overlay = NULL;
+    BlockDriverState *intermediate;
+
+    assert(active != NULL);
+    assert(bs != NULL);
+
+    /* if bs is the same as active, then by definition it has no overlay
+     */
+    if (active == bs) {
+        return NULL;
+    }
+
+    intermediate = active;
+    while (intermediate->backing_hd) {
+        if (intermediate->backing_hd == bs) {
+            overlay = intermediate;
+            break;
+        }
+        intermediate = intermediate->backing_hd;
+    }
+
+    return overlay;
+}
+
+typedef struct BlkIntermediateStates {
+    BlockDriverState *bs;
+    QSIMPLEQ_ENTRY(BlkIntermediateStates) entry;
+} BlkIntermediateStates;
+
+
+/*
+ * Drops images above 'base' up to and including 'top', and sets the image
+ * above 'top' to have base as its backing file.
+ *
+ * Requires that the overlay to 'top' is opened r/w, so that the backing file
+ * information in 'bs' can be properly updated.
+ *
+ * E.g., this will convert the following chain:
+ * bottom <- base <- intermediate <- top <- active
+ *
+ * to
+ *
+ * bottom <- base <- active
+ *
+ * It is allowed for bottom==base, in which case it converts:
+ *
+ * base <- intermediate <- top <- active
+ *
+ * to
+ *
+ * base <- active
+ *
+ * Error conditions:
+ *  if active == top, that is considered an error
+ *
+ */
+int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
+                           BlockDriverState *base)
+{
+    BlockDriverState *intermediate;
+    BlockDriverState *base_bs = NULL;
+    BlockDriverState *new_top_bs = NULL;
+    BlkIntermediateStates *intermediate_state, *next;
+    int ret = -EIO;
+
+    QSIMPLEQ_HEAD(states_to_delete, BlkIntermediateStates) states_to_delete;
+    QSIMPLEQ_INIT(&states_to_delete);
+
+    if (!top->drv || !base->drv) {
+        goto exit;
+    }
+
+    new_top_bs = bdrv_find_overlay(active, top);
+
+    if (new_top_bs == NULL) {
+        /* we could not find the image above 'top', this is an error */
+        goto exit;
+    }
+
+    /* special case of new_top_bs->backing_hd already pointing to base - nothing
+     * to do, no intermediate images */
+    if (new_top_bs->backing_hd == base) {
+        ret = 0;
+        goto exit;
+    }
+
+    intermediate = top;
+
+    /* now we will go down through the list, and add each BDS we find
+     * into our deletion queue, until we hit the 'base'
+     */
+    while (intermediate) {
+        intermediate_state = g_malloc0(sizeof(BlkIntermediateStates));
+        intermediate_state->bs = intermediate;
+        QSIMPLEQ_INSERT_TAIL(&states_to_delete, intermediate_state, entry);
+
+        if (intermediate->backing_hd == base) {
+            base_bs = intermediate->backing_hd;
+            break;
+        }
+        intermediate = intermediate->backing_hd;
+    }
+    if (base_bs == NULL) {
+        /* something went wrong, we did not end at the base. safely
+         * unravel everything, and exit with error */
+        goto exit;
+    }
+
+    /* success - we can delete the intermediate states, and link top->base */
+    ret = bdrv_change_backing_file(new_top_bs, base_bs->filename,
+                                   base_bs->drv ? base_bs->drv->format_name : "");
+    if (ret) {
+        goto exit;
+    }
+    new_top_bs->backing_hd = base_bs;
+
+
+    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
+        /* so that bdrv_close() does not recursively close the chain */
+        intermediate_state->bs->backing_hd = NULL;
+        bdrv_delete(intermediate_state->bs);
+    }
+    ret = 0;
+
+exit:
+    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
+        g_free(intermediate_state);
+    }
+    return ret;
+}
+
+
+static int bdrv_check_byte_request(BlockDriverState *bs, int64_t offset,
+                                   size_t size)
+{
+    int64_t len;
+
+    if (!bdrv_is_inserted(bs))
+        return -ENOMEDIUM;
+
+    if (bs->growable)
+        return 0;
+
+    len = bdrv_getlength(bs);
+
+    if (offset < 0)
+        return -EIO;
+
+    if ((offset > len) || (len - offset < size))
+        return -EIO;
+
+    return 0;
+}
+
+static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
+                              int nb_sectors)
+{
+    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
+                                   nb_sectors * BDRV_SECTOR_SIZE);
+}
+
+typedef struct RwCo {
+    BlockDriverState *bs;
+    int64_t sector_num;
+    int nb_sectors;
+    QEMUIOVector *qiov;
+    bool is_write;
+    int ret;
+} RwCo;
+
+static void coroutine_fn bdrv_rw_co_entry(void *opaque)
+{
+    RwCo *rwco = opaque;
+
+    if (!rwco->is_write) {
+        rwco->ret = bdrv_co_do_readv(rwco->bs, rwco->sector_num,
+                                     rwco->nb_sectors, rwco->qiov, 0);
+    } else {
+        rwco->ret = bdrv_co_do_writev(rwco->bs, rwco->sector_num,
+                                      rwco->nb_sectors, rwco->qiov, 0);
+    }
+}
+
+/*
+ * Process a vectored synchronous request using coroutines
+ */
+static int bdrv_rwv_co(BlockDriverState *bs, int64_t sector_num,
+                       QEMUIOVector *qiov, bool is_write)
+{
+    Coroutine *co;
+    RwCo rwco = {
+        .bs = bs,
+        .sector_num = sector_num,
+        .nb_sectors = qiov->size >> BDRV_SECTOR_BITS,
+        .qiov = qiov,
+        .is_write = is_write,
+        .ret = NOT_DONE,
+    };
+    assert((qiov->size & (BDRV_SECTOR_SIZE - 1)) == 0);
+
+    /**
+     * In sync call context, when the vcpu is blocked, this throttling timer
+     * will not fire; so the I/O throttling function has to be disabled here
+     * if it has been enabled.
+     */
+    if (bs->io_limits_enabled) {
+        fprintf(stderr, "Disabling I/O throttling on '%s' due "
+                        "to synchronous I/O.\n", bdrv_get_device_name(bs));
+        bdrv_io_limits_disable(bs);
+    }
+
+    if (qemu_in_coroutine()) {
+        /* Fast-path if already in coroutine context */
+        bdrv_rw_co_entry(&rwco);
+    } else {
+        co = qemu_coroutine_create(bdrv_rw_co_entry);
+        qemu_coroutine_enter(co, &rwco);
+        while (rwco.ret == NOT_DONE) {
+            qemu_aio_wait();
+        }
+    }
+    return rwco.ret;
+}
+
+/*
+ * Process a synchronous request using coroutines
+ */
+static int bdrv_rw_co(BlockDriverState *bs, int64_t sector_num, uint8_t *buf,
+                      int nb_sectors, bool is_write)
+{
+    QEMUIOVector qiov;
+    struct iovec iov = {
+        .iov_base = (void *)buf,
+        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
+    };
+
+    qemu_iovec_init_external(&qiov, &iov, 1);
+    return bdrv_rwv_co(bs, sector_num, &qiov, is_write);
+}
+
+/* return < 0 if error. See bdrv_write() for the return codes */
+int bdrv_read(BlockDriverState *bs, int64_t sector_num,
+              uint8_t *buf, int nb_sectors)
+{
+    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false);
+}
+
+/* Just like bdrv_read(), but with I/O throttling temporarily disabled */
+int bdrv_read_unthrottled(BlockDriverState *bs, int64_t sector_num,
+                          uint8_t *buf, int nb_sectors)
+{
+    bool enabled;
+    int ret;
+
+    enabled = bs->io_limits_enabled;
+    bs->io_limits_enabled = false;
+    ret = bdrv_read(bs, 0, buf, 1);
+    bs->io_limits_enabled = enabled;
+    return ret;
+}
+
+/* Return < 0 if error. Important errors are:
+  -EIO         generic I/O error (may happen for all errors)
+  -ENOMEDIUM   No media inserted.
+  -EINVAL      Invalid sector number or nb_sectors
+  -EACCES      Trying to write a read-only device
+*/
+int bdrv_write(BlockDriverState *bs, int64_t sector_num,
+               const uint8_t *buf, int nb_sectors)
+{
+    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true);
+}
+
+int bdrv_writev(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov)
+{
+    return bdrv_rwv_co(bs, sector_num, qiov, true);
+}
+
+int bdrv_pread(BlockDriverState *bs, int64_t offset,
+               void *buf, int count1)
+{
+    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
+    int len, nb_sectors, count;
+    int64_t sector_num;
+    int ret;
+
+    count = count1;
+    /* first read to align to sector start */
+    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
+    if (len > count)
+        len = count;
+    sector_num = offset >> BDRV_SECTOR_BITS;
+    if (len > 0) {
+        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
+            return ret;
+        memcpy(buf, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), len);
+        count -= len;
+        if (count == 0)
+            return count1;
+        sector_num++;
+        buf += len;
+    }
+
+    /* read the sectors "in place" */
+    nb_sectors = count >> BDRV_SECTOR_BITS;
+    if (nb_sectors > 0) {
+        if ((ret = bdrv_read(bs, sector_num, buf, nb_sectors)) < 0)
+            return ret;
+        sector_num += nb_sectors;
+        len = nb_sectors << BDRV_SECTOR_BITS;
+        buf += len;
+        count -= len;
+    }
+
+    /* add data from the last sector */
+    if (count > 0) {
+        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
+            return ret;
+        memcpy(buf, tmp_buf, count);
+    }
+    return count1;
+}
+
+int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
+{
+    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
+    int len, nb_sectors, count;
+    int64_t sector_num;
+    int ret;
+
+    count = qiov->size;
+
+    /* first write to align to sector start */
+    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
+    if (len > count)
+        len = count;
+    sector_num = offset >> BDRV_SECTOR_BITS;
+    if (len > 0) {
+        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
+            return ret;
+        qemu_iovec_to_buf(qiov, 0, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)),
+                          len);
+        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
+            return ret;
+        count -= len;
+        if (count == 0)
+            return qiov->size;
+        sector_num++;
+    }
+
+    /* write the sectors "in place" */
+    nb_sectors = count >> BDRV_SECTOR_BITS;
+    if (nb_sectors > 0) {
+        QEMUIOVector qiov_inplace;
+
+        qemu_iovec_init(&qiov_inplace, qiov->niov);
+        qemu_iovec_concat(&qiov_inplace, qiov, len,
+                          nb_sectors << BDRV_SECTOR_BITS);
+        ret = bdrv_writev(bs, sector_num, &qiov_inplace);
+        qemu_iovec_destroy(&qiov_inplace);
+        if (ret < 0) {
+            return ret;
+        }
+
+        sector_num += nb_sectors;
+        len = nb_sectors << BDRV_SECTOR_BITS;
+        count -= len;
+    }
+
+    /* add data from the last sector */
+    if (count > 0) {
+        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
+            return ret;
+        qemu_iovec_to_buf(qiov, qiov->size - count, tmp_buf, count);
+        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
+            return ret;
+    }
+    return qiov->size;
+}
+
+int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
+                const void *buf, int count1)
+{
+    QEMUIOVector qiov;
+    struct iovec iov = {
+        .iov_base   = (void *) buf,
+        .iov_len    = count1,
+    };
+
+    qemu_iovec_init_external(&qiov, &iov, 1);
+    return bdrv_pwritev(bs, offset, &qiov);
+}
+
+/*
+ * Writes to the file and ensures that no writes are reordered across this
+ * request (acts as a barrier)
+ *
+ * Returns 0 on success, -errno in error cases.
+ */
+int bdrv_pwrite_sync(BlockDriverState *bs, int64_t offset,
+    const void *buf, int count)
+{
+    int ret;
+
+    ret = bdrv_pwrite(bs, offset, buf, count);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* No flush needed for cache modes that already do it */
+    if (bs->enable_write_cache) {
+        bdrv_flush(bs);
+    }
+
+    return 0;
+}
+
+static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors, QEMUIOVector *qiov)
+{
+    /* Perform I/O through a temporary buffer so that users who scribble over
+     * their read buffer while the operation is in progress do not end up
+     * modifying the image file.  This is critical for zero-copy guest I/O
+     * where anything might happen inside guest memory.
+     */
+    void *bounce_buffer;
+
+    BlockDriver *drv = bs->drv;
+    struct iovec iov;
+    QEMUIOVector bounce_qiov;
+    int64_t cluster_sector_num;
+    int cluster_nb_sectors;
+    size_t skip_bytes;
+    int ret;
+
+    /* Cover entire cluster so no additional backing file I/O is required when
+     * allocating cluster in the image file.
+     */
+    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
+                           &cluster_sector_num, &cluster_nb_sectors);
+
+    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
+                                   cluster_sector_num, cluster_nb_sectors);
+
+    iov.iov_len = cluster_nb_sectors * BDRV_SECTOR_SIZE;
+    iov.iov_base = bounce_buffer = qemu_blockalign(bs, iov.iov_len);
+    qemu_iovec_init_external(&bounce_qiov, &iov, 1);
+
+    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
+                             &bounce_qiov);
+    if (ret < 0) {
+        goto err;
+    }
+
+    if (drv->bdrv_co_write_zeroes &&
+        buffer_is_zero(bounce_buffer, iov.iov_len)) {
+        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
+                                      cluster_nb_sectors);
+    } else {
+        /* This does not change the data on the disk, it is not necessary
+         * to flush even in cache=writethrough mode.
+         */
+        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
+                                  &bounce_qiov);
+    }
+
+    if (ret < 0) {
+        /* It might be okay to ignore write errors for guest requests.  If this
+         * is a deliberate copy-on-read then we don't want to ignore the error.
+         * Simply report it in all cases.
+         */
+        goto err;
+    }
+
+    skip_bytes = (sector_num - cluster_sector_num) * BDRV_SECTOR_SIZE;
+    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
+                        nb_sectors * BDRV_SECTOR_SIZE);
+
+err:
+    qemu_vfree(bounce_buffer);
+    return ret;
+}
+
+/*
+ * Handle a read request in coroutine context
+ */
+static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+    BdrvRequestFlags flags)
+{
+    BlockDriver *drv = bs->drv;
+    BdrvTrackedRequest req;
+    int ret;
+
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (bdrv_check_request(bs, sector_num, nb_sectors)) {
+        return -EIO;
+    }
+
+    /* throttling disk read I/O */
+    if (bs->io_limits_enabled) {
+        bdrv_io_limits_intercept(bs, false, nb_sectors);
+    }
+
+    if (bs->copy_on_read) {
+        flags |= BDRV_REQ_COPY_ON_READ;
+    }
+    if (flags & BDRV_REQ_COPY_ON_READ) {
+        bs->copy_on_read_in_flight++;
+    }
+
+    if (bs->copy_on_read_in_flight) {
+        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
+    }
+
+    tracked_request_begin(&req, bs, sector_num, nb_sectors, false);
+
+    if (flags & BDRV_REQ_COPY_ON_READ) {
+        int pnum;
+
+        ret = bdrv_co_is_allocated(bs, sector_num, nb_sectors, &pnum);
+        if (ret < 0) {
+            goto out;
+        }
+
+        if (!ret || pnum != nb_sectors) {
+            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
+            goto out;
+        }
+    }
+
+    ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
+
+out:
+    tracked_request_end(&req);
+
+    if (flags & BDRV_REQ_COPY_ON_READ) {
+        bs->copy_on_read_in_flight--;
+    }
+
+    return ret;
+}
+
+int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
+    int nb_sectors, QEMUIOVector *qiov)
+{
+    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
+
+    return bdrv_co_do_readv(bs, sector_num, nb_sectors, qiov, 0);
+}
+
+int coroutine_fn bdrv_co_copy_on_readv(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov)
+{
+    trace_bdrv_co_copy_on_readv(bs, sector_num, nb_sectors);
+
+    return bdrv_co_do_readv(bs, sector_num, nb_sectors, qiov,
+                            BDRV_REQ_COPY_ON_READ);
+}
+
+static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors)
+{
+    BlockDriver *drv = bs->drv;
+    QEMUIOVector qiov;
+    struct iovec iov;
+    int ret;
+
+    /* TODO Emulate only part of misaligned requests instead of letting block
+     * drivers return -ENOTSUP and emulate everything */
+
+    /* First try the efficient write zeroes operation */
+    if (drv->bdrv_co_write_zeroes) {
+        ret = drv->bdrv_co_write_zeroes(bs, sector_num, nb_sectors);
+        if (ret != -ENOTSUP) {
+            return ret;
+        }
+    }
+
+    /* Fall back to bounce buffer if write zeroes is unsupported */
+    iov.iov_len  = nb_sectors * BDRV_SECTOR_SIZE;
+    iov.iov_base = qemu_blockalign(bs, iov.iov_len);
+    memset(iov.iov_base, 0, iov.iov_len);
+    qemu_iovec_init_external(&qiov, &iov, 1);
+
+    ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, &qiov);
+
+    qemu_vfree(iov.iov_base);
+    return ret;
+}
+
+/*
+ * Handle a write request in coroutine context
+ */
+static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+    BdrvRequestFlags flags)
+{
+    BlockDriver *drv = bs->drv;
+    BdrvTrackedRequest req;
+    int ret;
+
+    if (!bs->drv) {
+        return -ENOMEDIUM;
+    }
+    if (bs->read_only) {
+        return -EACCES;
+    }
+    if (bdrv_check_request(bs, sector_num, nb_sectors)) {
+        return -EIO;
+    }
+
+    /* throttling disk write I/O */
+    if (bs->io_limits_enabled) {
+        bdrv_io_limits_intercept(bs, true, nb_sectors);
+    }
+
+    if (bs->copy_on_read_in_flight) {
+        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
+    }
+
+    tracked_request_begin(&req, bs, sector_num, nb_sectors, true);
+
+    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, &req);
+
+    if (ret < 0) {
+        /* Do nothing, write notifier decided to fail this request */
+    } else if (flags & BDRV_REQ_ZERO_WRITE) {
+        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors);
+    } else {
+        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
+    }
+
+    if (ret == 0 && !bs->enable_write_cache) {
+        ret = bdrv_co_flush(bs);
+    }
+
+    if (bs->dirty_bitmap) {
+        bdrv_set_dirty(bs, sector_num, nb_sectors);
+    }
+
+    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
+        bs->wr_highest_sector = sector_num + nb_sectors - 1;
+    }
+
+    tracked_request_end(&req);
+
+    return ret;
+}
+
+int coroutine_fn bdrv_co_writev(BlockDriverState *bs, int64_t sector_num,
+    int nb_sectors, QEMUIOVector *qiov)
+{
+    trace_bdrv_co_writev(bs, sector_num, nb_sectors);
+
+    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
+}
+
+int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
+                                      int64_t sector_num, int nb_sectors)
+{
+    trace_bdrv_co_write_zeroes(bs, sector_num, nb_sectors);
+
+    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
+                             BDRV_REQ_ZERO_WRITE);
+}
+
+/**
+ * Truncate file to 'offset' bytes (needed only for file protocols)
+ */
+int bdrv_truncate(BlockDriverState *bs, int64_t offset)
+{
+    BlockDriver *drv = bs->drv;
+    int ret;
+    if (!drv)
+        return -ENOMEDIUM;
+    if (!drv->bdrv_truncate)
+        return -ENOTSUP;
+    if (bs->read_only)
+        return -EACCES;
+    if (bdrv_in_use(bs))
+        return -EBUSY;
+    ret = drv->bdrv_truncate(bs, offset);
+    if (ret == 0) {
+        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
+        bdrv_dev_resize_cb(bs);
+    }
+    return ret;
+}
+
+/**
+ * Length of a allocated file in bytes. Sparse files are counted by actual
+ * allocated space. Return < 0 if error or unknown.
+ */
+int64_t bdrv_get_allocated_file_size(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (drv->bdrv_get_allocated_file_size) {
+        return drv->bdrv_get_allocated_file_size(bs);
+    }
+    if (bs->file) {
+        return bdrv_get_allocated_file_size(bs->file);
+    }
+    return -ENOTSUP;
+}
+
+/**
+ * Length of a file in bytes. Return < 0 if error or unknown.
+ */
+int64_t bdrv_getlength(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv)
+        return -ENOMEDIUM;
+
+    if (bs->growable || bdrv_dev_has_removable_media(bs)) {
+        if (drv->bdrv_getlength) {
+            return drv->bdrv_getlength(bs);
+        }
+    }
+    return bs->total_sectors * BDRV_SECTOR_SIZE;
+}
+
+/* return 0 as number of sectors if no device present or error */
+void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
+{
+    int64_t length;
+    length = bdrv_getlength(bs);
+    if (length < 0)
+        length = 0;
+    else
+        length = length >> BDRV_SECTOR_BITS;
+    *nb_sectors_ptr = length;
+}
+
+/* throttling disk io limits */
+void bdrv_set_io_limits(BlockDriverState *bs,
+                        BlockIOLimit *io_limits)
+{
+    bs->io_limits = *io_limits;
+    bs->io_limits_enabled = bdrv_io_limits_enabled(bs);
+}
+
+void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
+                       BlockdevOnError on_write_error)
+{
+    bs->on_read_error = on_read_error;
+    bs->on_write_error = on_write_error;
+}
+
+BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
+{
+    return is_read ? bs->on_read_error : bs->on_write_error;
+}
+
+BlockErrorAction bdrv_get_error_action(BlockDriverState *bs, bool is_read, int error)
+{
+    BlockdevOnError on_err = is_read ? bs->on_read_error : bs->on_write_error;
+
+    switch (on_err) {
+    case BLOCKDEV_ON_ERROR_ENOSPC:
+        return (error == ENOSPC) ? BDRV_ACTION_STOP : BDRV_ACTION_REPORT;
+    case BLOCKDEV_ON_ERROR_STOP:
+        return BDRV_ACTION_STOP;
+    case BLOCKDEV_ON_ERROR_REPORT:
+        return BDRV_ACTION_REPORT;
+    case BLOCKDEV_ON_ERROR_IGNORE:
+        return BDRV_ACTION_IGNORE;
+    default:
+        abort();
+    }
+}
+
+/* This is done by device models because, while the block layer knows
+ * about the error, it does not know whether an operation comes from
+ * the device or the block layer (from a job, for example).
+ */
+void bdrv_error_action(BlockDriverState *bs, BlockErrorAction action,
+                       bool is_read, int error)
+{
+    assert(error >= 0);
+    bdrv_emit_qmp_error_event(bs, QEVENT_BLOCK_IO_ERROR, action, is_read);
+    if (action == BDRV_ACTION_STOP) {
+        vm_stop(RUN_STATE_IO_ERROR);
+        bdrv_iostatus_set_err(bs, error);
+    }
+}
+
+int bdrv_is_read_only(BlockDriverState *bs)
+{
+    return bs->read_only;
+}
+
+int bdrv_is_sg(BlockDriverState *bs)
+{
+    return bs->sg;
+}
+
+int bdrv_enable_write_cache(BlockDriverState *bs)
+{
+    return bs->enable_write_cache;
+}
+
+void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
+{
+    bs->enable_write_cache = wce;
+
+    /* so a reopen() will preserve wce */
+    if (wce) {
+        bs->open_flags |= BDRV_O_CACHE_WB;
+    } else {
+        bs->open_flags &= ~BDRV_O_CACHE_WB;
+    }
+}
+
+int bdrv_is_encrypted(BlockDriverState *bs)
+{
+    if (bs->backing_hd && bs->backing_hd->encrypted)
+        return 1;
+    return bs->encrypted;
+}
+
+int bdrv_key_required(BlockDriverState *bs)
+{
+    BlockDriverState *backing_hd = bs->backing_hd;
+
+    if (backing_hd && backing_hd->encrypted && !backing_hd->valid_key)
+        return 1;
+    return (bs->encrypted && !bs->valid_key);
+}
+
+int bdrv_set_key(BlockDriverState *bs, const char *key)
+{
+    int ret;
+    if (bs->backing_hd && bs->backing_hd->encrypted) {
+        ret = bdrv_set_key(bs->backing_hd, key);
+        if (ret < 0)
+            return ret;
+        if (!bs->encrypted)
+            return 0;
+    }
+    if (!bs->encrypted) {
+        return -EINVAL;
+    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
+        return -ENOMEDIUM;
+    }
+    ret = bs->drv->bdrv_set_key(bs, key);
+    if (ret < 0) {
+        bs->valid_key = 0;
+    } else if (!bs->valid_key) {
+        bs->valid_key = 1;
+        /* call the change callback now, we skipped it on open */
+        bdrv_dev_change_media_cb(bs, true);
+    }
+    return ret;
+}
+
+const char *bdrv_get_format_name(BlockDriverState *bs)
+{
+    return bs->drv ? bs->drv->format_name : NULL;
+}
+
+void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
+                         void *opaque)
+{
+    BlockDriver *drv;
+
+    QLIST_FOREACH(drv, &bdrv_drivers, list) {
+        it(opaque, drv->format_name);
+    }
+}
+
+BlockDriverState *bdrv_find(const char *name)
+{
+    BlockDriverState *bs;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        if (!strcmp(name, bs->device_name)) {
+            return bs;
+        }
+    }
+    return NULL;
+}
+
+BlockDriverState *bdrv_next(BlockDriverState *bs)
+{
+    if (!bs) {
+        return QTAILQ_FIRST(&bdrv_states);
+    }
+    return QTAILQ_NEXT(bs, list);
+}
+
+void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
+{
+    BlockDriverState *bs;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        it(opaque, bs);
+    }
+}
+
+const char *bdrv_get_device_name(BlockDriverState *bs)
+{
+    return bs->device_name;
+}
+
+int bdrv_get_flags(BlockDriverState *bs)
+{
+    return bs->open_flags;
+}
+
+int bdrv_flush_all(void)
+{
+    BlockDriverState *bs;
+    int result = 0;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        int ret = bdrv_flush(bs);
+        if (ret < 0 && !result) {
+            result = ret;
+        }
+    }
+
+    return result;
+}
+
+int bdrv_has_zero_init_1(BlockDriverState *bs)
+{
+    return 1;
+}
+
+int bdrv_has_zero_init(BlockDriverState *bs)
+{
+    assert(bs->drv);
+
+    if (bs->drv->bdrv_has_zero_init) {
+        return bs->drv->bdrv_has_zero_init(bs);
+    }
+
+    /* safe default */
+    return 0;
+}
+
+typedef struct BdrvCoIsAllocatedData {
+    BlockDriverState *bs;
+    BlockDriverState *base;
+    int64_t sector_num;
+    int nb_sectors;
+    int *pnum;
+    int ret;
+    bool done;
+} BdrvCoIsAllocatedData;
+
+/*
+ * Returns true iff the specified sector is present in the disk image. Drivers
+ * not implementing the functionality are assumed to not support backing files,
+ * hence all their sectors are reported as allocated.
+ *
+ * If 'sector_num' is beyond the end of the disk image the return value is 0
+ * and 'pnum' is set to 0.
+ *
+ * 'pnum' is set to the number of sectors (including and immediately following
+ * the specified sector) that are known to be in the same
+ * allocated/unallocated state.
+ *
+ * 'nb_sectors' is the max value 'pnum' should be set to.  If nb_sectors goes
+ * beyond the end of the disk image it will be clamped.
+ */
+int coroutine_fn bdrv_co_is_allocated(BlockDriverState *bs, int64_t sector_num,
+                                      int nb_sectors, int *pnum)
+{
+    int64_t n;
+
+    if (sector_num >= bs->total_sectors) {
+        *pnum = 0;
+        return 0;
+    }
+
+    n = bs->total_sectors - sector_num;
+    if (n < nb_sectors) {
+        nb_sectors = n;
+    }
+
+    if (!bs->drv->bdrv_co_is_allocated) {
+        *pnum = nb_sectors;
+        return 1;
+    }
+
+    return bs->drv->bdrv_co_is_allocated(bs, sector_num, nb_sectors, pnum);
+}
+
+/* Coroutine wrapper for bdrv_is_allocated() */
+static void coroutine_fn bdrv_is_allocated_co_entry(void *opaque)
+{
+    BdrvCoIsAllocatedData *data = opaque;
+    BlockDriverState *bs = data->bs;
+
+    data->ret = bdrv_co_is_allocated(bs, data->sector_num, data->nb_sectors,
+                                     data->pnum);
+    data->done = true;
+}
+
+/*
+ * Synchronous wrapper around bdrv_co_is_allocated().
+ *
+ * See bdrv_co_is_allocated() for details.
+ */
+int bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num, int nb_sectors,
+                      int *pnum)
+{
+    Coroutine *co;
+    BdrvCoIsAllocatedData data = {
+        .bs = bs,
+        .sector_num = sector_num,
+        .nb_sectors = nb_sectors,
+        .pnum = pnum,
+        .done = false,
+    };
+
+    co = qemu_coroutine_create(bdrv_is_allocated_co_entry);
+    qemu_coroutine_enter(co, &data);
+    while (!data.done) {
+        qemu_aio_wait();
+    }
+    return data.ret;
+}
+
+/*
+ * Given an image chain: ... -> [BASE] -> [INTER1] -> [INTER2] -> [TOP]
+ *
+ * Return true if the given sector is allocated in any image between
+ * BASE and TOP (inclusive).  BASE can be NULL to check if the given
+ * sector is allocated in any image of the chain.  Return false otherwise.
+ *
+ * 'pnum' is set to the number of sectors (including and immediately following
+ *  the specified sector) that are known to be in the same
+ *  allocated/unallocated state.
+ *
+ */
+int coroutine_fn bdrv_co_is_allocated_above(BlockDriverState *top,
+                                            BlockDriverState *base,
+                                            int64_t sector_num,
+                                            int nb_sectors, int *pnum)
+{
+    BlockDriverState *intermediate;
+    int ret, n = nb_sectors;
+
+    intermediate = top;
+    while (intermediate && intermediate != base) {
+        int pnum_inter;
+        ret = bdrv_co_is_allocated(intermediate, sector_num, nb_sectors,
+                                   &pnum_inter);
+        if (ret < 0) {
+            return ret;
+        } else if (ret) {
+            *pnum = pnum_inter;
+            return 1;
+        }
+
+        /*
+         * [sector_num, nb_sectors] is unallocated on top but intermediate
+         * might have
+         *
+         * [sector_num+x, nr_sectors] allocated.
+         */
+        if (n > pnum_inter &&
+            (intermediate == top ||
+             sector_num + pnum_inter < intermediate->total_sectors)) {
+            n = pnum_inter;
+        }
+
+        intermediate = intermediate->backing_hd;
+    }
+
+    *pnum = n;
+    return 0;
+}
+
+/* Coroutine wrapper for bdrv_is_allocated_above() */
+static void coroutine_fn bdrv_is_allocated_above_co_entry(void *opaque)
+{
+    BdrvCoIsAllocatedData *data = opaque;
+    BlockDriverState *top = data->bs;
+    BlockDriverState *base = data->base;
+
+    data->ret = bdrv_co_is_allocated_above(top, base, data->sector_num,
+                                           data->nb_sectors, data->pnum);
+    data->done = true;
+}
+
+/*
+ * Synchronous wrapper around bdrv_co_is_allocated_above().
+ *
+ * See bdrv_co_is_allocated_above() for details.
+ */
+int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base,
+                            int64_t sector_num, int nb_sectors, int *pnum)
+{
+    Coroutine *co;
+    BdrvCoIsAllocatedData data = {
+        .bs = top,
+        .base = base,
+        .sector_num = sector_num,
+        .nb_sectors = nb_sectors,
+        .pnum = pnum,
+        .done = false,
+    };
+
+    co = qemu_coroutine_create(bdrv_is_allocated_above_co_entry);
+    qemu_coroutine_enter(co, &data);
+    while (!data.done) {
+        qemu_aio_wait();
+    }
+    return data.ret;
+}
+
+const char *bdrv_get_encrypted_filename(BlockDriverState *bs)
+{
+    if (bs->backing_hd && bs->backing_hd->encrypted)
+        return bs->backing_file;
+    else if (bs->encrypted)
+        return bs->filename;
+    else
+        return NULL;
+}
+
+void bdrv_get_backing_filename(BlockDriverState *bs,
+                               char *filename, int filename_size)
+{
+    pstrcpy(filename, filename_size, bs->backing_file);
+}
+
+int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
+                          const uint8_t *buf, int nb_sectors)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv)
+        return -ENOMEDIUM;
+    if (!drv->bdrv_write_compressed)
+        return -ENOTSUP;
+    if (bdrv_check_request(bs, sector_num, nb_sectors))
+        return -EIO;
+
+    assert(!bs->dirty_bitmap);
+
+    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
+}
+
+int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv)
+        return -ENOMEDIUM;
+    if (!drv->bdrv_get_info)
+        return -ENOTSUP;
+    memset(bdi, 0, sizeof(*bdi));
+    return drv->bdrv_get_info(bs, bdi);
+}
+
+int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
+                      int64_t pos, int size)
+{
+    QEMUIOVector qiov;
+    struct iovec iov = {
+        .iov_base   = (void *) buf,
+        .iov_len    = size,
+    };
+
+    qemu_iovec_init_external(&qiov, &iov, 1);
+    return bdrv_writev_vmstate(bs, &qiov, pos);
+}
+
+int bdrv_writev_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos)
+{
+    BlockDriver *drv = bs->drv;
+
+    if (!drv) {
+        return -ENOMEDIUM;
+    } else if (drv->bdrv_save_vmstate) {
+        return drv->bdrv_save_vmstate(bs, qiov, pos);
+    } else if (bs->file) {
+        return bdrv_writev_vmstate(bs->file, qiov, pos);
+    }
+
+    return -ENOTSUP;
+}
+
+int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
+                      int64_t pos, int size)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv)
+        return -ENOMEDIUM;
+    if (drv->bdrv_load_vmstate)
+        return drv->bdrv_load_vmstate(bs, buf, pos, size);
+    if (bs->file)
+        return bdrv_load_vmstate(bs->file, buf, pos, size);
+    return -ENOTSUP;
+}
+
+void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
+{
+    if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
+        return;
+    }
+
+    bs->drv->bdrv_debug_event(bs, event);
+}
+
+int bdrv_debug_breakpoint(BlockDriverState *bs, const char *event,
+                          const char *tag)
+{
+    while (bs && bs->drv && !bs->drv->bdrv_debug_breakpoint) {
+        bs = bs->file;
+    }
+
+    if (bs && bs->drv && bs->drv->bdrv_debug_breakpoint) {
+        return bs->drv->bdrv_debug_breakpoint(bs, event, tag);
+    }
+
+    return -ENOTSUP;
+}
+
+int bdrv_debug_resume(BlockDriverState *bs, const char *tag)
+{
+    while (bs && bs->drv && !bs->drv->bdrv_debug_resume) {
+        bs = bs->file;
+    }
+
+    if (bs && bs->drv && bs->drv->bdrv_debug_resume) {
+        return bs->drv->bdrv_debug_resume(bs, tag);
+    }
+
+    return -ENOTSUP;
+}
+
+bool bdrv_debug_is_suspended(BlockDriverState *bs, const char *tag)
+{
+    while (bs && bs->drv && !bs->drv->bdrv_debug_is_suspended) {
+        bs = bs->file;
+    }
+
+    if (bs && bs->drv && bs->drv->bdrv_debug_is_suspended) {
+        return bs->drv->bdrv_debug_is_suspended(bs, tag);
+    }
+
+    return false;
+}
+
+int bdrv_is_snapshot(BlockDriverState *bs)
+{
+    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
+}
+
+/* backing_file can either be relative, or absolute, or a protocol.  If it is
+ * relative, it must be relative to the chain.  So, passing in bs->filename
+ * from a BDS as backing_file should not be done, as that may be relative to
+ * the CWD rather than the chain. */
+BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
+        const char *backing_file)
+{
+    char *filename_full = NULL;
+    char *backing_file_full = NULL;
+    char *filename_tmp = NULL;
+    int is_protocol = 0;
+    BlockDriverState *curr_bs = NULL;
+    BlockDriverState *retval = NULL;
+
+    if (!bs || !bs->drv || !backing_file) {
+        return NULL;
+    }
+
+    filename_full     = g_malloc(PATH_MAX);
+    backing_file_full = g_malloc(PATH_MAX);
+    filename_tmp      = g_malloc(PATH_MAX);
+
+    is_protocol = path_has_protocol(backing_file);
+
+    for (curr_bs = bs; curr_bs->backing_hd; curr_bs = curr_bs->backing_hd) {
+
+        /* If either of the filename paths is actually a protocol, then
+         * compare unmodified paths; otherwise make paths relative */
+        if (is_protocol || path_has_protocol(curr_bs->backing_file)) {
+            if (strcmp(backing_file, curr_bs->backing_file) == 0) {
+                retval = curr_bs->backing_hd;
+                break;
+            }
+        } else {
+            /* If not an absolute filename path, make it relative to the current
+             * image's filename path */
+            path_combine(filename_tmp, PATH_MAX, curr_bs->filename,
+                         backing_file);
+
+            /* We are going to compare absolute pathnames */
+            if (!realpath(filename_tmp, filename_full)) {
+                continue;
+            }
+
+            /* We need to make sure the backing filename we are comparing against
+             * is relative to the current image filename (or absolute) */
+            path_combine(filename_tmp, PATH_MAX, curr_bs->filename,
+                         curr_bs->backing_file);
+
+            if (!realpath(filename_tmp, backing_file_full)) {
+                continue;
+            }
+
+            if (strcmp(backing_file_full, filename_full) == 0) {
+                retval = curr_bs->backing_hd;
+                break;
+            }
+        }
+    }
+
+    g_free(filename_full);
+    g_free(backing_file_full);
+    g_free(filename_tmp);
+    return retval;
+}
+
+int bdrv_get_backing_file_depth(BlockDriverState *bs)
+{
+    if (!bs->drv) {
+        return 0;
+    }
+
+    if (!bs->backing_hd) {
+        return 0;
+    }
+
+    return 1 + bdrv_get_backing_file_depth(bs->backing_hd);
+}
+
+BlockDriverState *bdrv_find_base(BlockDriverState *bs)
+{
+    BlockDriverState *curr_bs = NULL;
+
+    if (!bs) {
+        return NULL;
+    }
+
+    curr_bs = bs;
+
+    while (curr_bs->backing_hd) {
+        curr_bs = curr_bs->backing_hd;
+    }
+    return curr_bs;
+}
+
+/**************************************************************/
+/* async I/Os */
+
+BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
+                                 QEMUIOVector *qiov, int nb_sectors,
+                                 BlockDriverCompletionFunc *cb, void *opaque)
+{
+    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);
+
+    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
+                                 cb, opaque, false);
+}
+
+BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
+                                  QEMUIOVector *qiov, int nb_sectors,
+                                  BlockDriverCompletionFunc *cb, void *opaque)
+{
+    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);
+
+    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
+                                 cb, opaque, true);
+}
+
+
+typedef struct MultiwriteCB {
+    int error;
+    int num_requests;
+    int num_callbacks;
+    struct {
+        BlockDriverCompletionFunc *cb;
+        void *opaque;
+        QEMUIOVector *free_qiov;
+    } callbacks[];
+} MultiwriteCB;
+
+static void multiwrite_user_cb(MultiwriteCB *mcb)
+{
+    int i;
+
+    for (i = 0; i < mcb->num_callbacks; i++) {
+        mcb->callbacks[i].cb(mcb->callbacks[i].opaque, mcb->error);
+        if (mcb->callbacks[i].free_qiov) {
+            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
+        }
+        g_free(mcb->callbacks[i].free_qiov);
+    }
+}
+
+static void multiwrite_cb(void *opaque, int ret)
+{
+    MultiwriteCB *mcb = opaque;
+
+    trace_multiwrite_cb(mcb, ret);
+
+    if (ret < 0 && !mcb->error) {
+        mcb->error = ret;
+    }
+
+    mcb->num_requests--;
+    if (mcb->num_requests == 0) {
+        multiwrite_user_cb(mcb);
+        g_free(mcb);
+    }
+}
+
+static int multiwrite_req_compare(const void *a, const void *b)
+{
+    const BlockRequest *req1 = a, *req2 = b;
+
+    /*
+     * Note that we can't simply subtract req2->sector from req1->sector
+     * here as that could overflow the return value.
+     */
+    if (req1->sector > req2->sector) {
+        return 1;
+    } else if (req1->sector < req2->sector) {
+        return -1;
+    } else {
+        return 0;
+    }
+}
+
+/*
+ * Takes a bunch of requests and tries to merge them. Returns the number of
+ * requests that remain after merging.
+ */
+static int multiwrite_merge(BlockDriverState *bs, BlockRequest *reqs,
+    int num_reqs, MultiwriteCB *mcb)
+{
+    int i, outidx;
+
+    // Sort requests by start sector
+    qsort(reqs, num_reqs, sizeof(*reqs), &multiwrite_req_compare);
+
+    // Check if adjacent requests touch the same clusters. If so, combine them,
+    // filling up gaps with zero sectors.
+    outidx = 0;
+    for (i = 1; i < num_reqs; i++) {
+        int merge = 0;
+        int64_t oldreq_last = reqs[outidx].sector + reqs[outidx].nb_sectors;
+
+        // Handle exactly sequential writes and overlapping writes.
+        if (reqs[i].sector <= oldreq_last) {
+            merge = 1;
+        }
+
+        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
+            merge = 0;
+        }
+
+        if (merge) {
+            size_t size;
+            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
+            qemu_iovec_init(qiov,
+                reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1);
+
+            // Add the first request to the merged one. If the requests are
+            // overlapping, drop the last sectors of the first request.
+            size = (reqs[i].sector - reqs[outidx].sector) << 9;
+            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
+
+            // We should need to add any zeros between the two requests
+            assert (reqs[i].sector <= oldreq_last);
+
+            // Add the second request
+            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
+
+            reqs[outidx].nb_sectors = qiov->size >> 9;
+            reqs[outidx].qiov = qiov;
+
+            mcb->callbacks[i].free_qiov = reqs[outidx].qiov;
+        } else {
+            outidx++;
+            reqs[outidx].sector     = reqs[i].sector;
+            reqs[outidx].nb_sectors = reqs[i].nb_sectors;
+            reqs[outidx].qiov       = reqs[i].qiov;
+        }
+    }
+
+    return outidx + 1;
+}
+
+/*
+ * Submit multiple AIO write requests at once.
+ *
+ * On success, the function returns 0 and all requests in the reqs array have
+ * been submitted. In error case this function returns -1, and any of the
+ * requests may or may not be submitted yet. In particular, this means that the
+ * callback will be called for some of the requests, for others it won't. The
+ * caller must check the error field of the BlockRequest to wait for the right
+ * callbacks (if error != 0, no callback will be called).
+ *
+ * The implementation may modify the contents of the reqs array, e.g. to merge
+ * requests. However, the fields opaque and error are left unmodified as they
+ * are used to signal failure for a single request to the caller.
+ */
+int bdrv_aio_multiwrite(BlockDriverState *bs, BlockRequest *reqs, int num_reqs)
+{
+    MultiwriteCB *mcb;
+    int i;
+
+    /* don't submit writes if we don't have a medium */
+    if (bs->drv == NULL) {
+        for (i = 0; i < num_reqs; i++) {
+            reqs[i].error = -ENOMEDIUM;
+        }
+        return -1;
+    }
+
+    if (num_reqs == 0) {
+        return 0;
+    }
+
+    // Create MultiwriteCB structure
+    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
+    mcb->num_requests = 0;
+    mcb->num_callbacks = num_reqs;
+
+    for (i = 0; i < num_reqs; i++) {
+        mcb->callbacks[i].cb = reqs[i].cb;
+        mcb->callbacks[i].opaque = reqs[i].opaque;
+    }
+
+    // Check for mergable requests
+    num_reqs = multiwrite_merge(bs, reqs, num_reqs, mcb);
+
+    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);
+
+    /* Run the aio requests. */
+    mcb->num_requests = num_reqs;
+    for (i = 0; i < num_reqs; i++) {
+        bdrv_aio_writev(bs, reqs[i].sector, reqs[i].qiov,
+            reqs[i].nb_sectors, multiwrite_cb, mcb);
+    }
+
+    return 0;
+}
+
+void bdrv_aio_cancel(BlockDriverAIOCB *acb)
+{
+    acb->aiocb_info->cancel(acb);
+}
+
+/* block I/O throttling */
+static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
+                 bool is_write, double elapsed_time, uint64_t *wait)
+{
+    uint64_t bps_limit = 0;
+    uint64_t extension;
+    double   bytes_limit, bytes_base, bytes_res;
+    double   slice_time, wait_time;
+
+    if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
+        bps_limit = bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL];
+    } else if (bs->io_limits.bps[is_write]) {
+        bps_limit = bs->io_limits.bps[is_write];
+    } else {
+        if (wait) {
+            *wait = 0;
+        }
+
+        return false;
+    }
+
+    slice_time = bs->slice_end - bs->slice_start;
+    slice_time /= (NANOSECONDS_PER_SECOND);
+    bytes_limit = bps_limit * slice_time;
+    bytes_base  = bs->slice_submitted.bytes[is_write];
+    if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
+        bytes_base += bs->slice_submitted.bytes[!is_write];
+    }
+
+    /* bytes_base: the bytes of data which have been read/written; and
+     *             it is obtained from the history statistic info.
+     * bytes_res: the remaining bytes of data which need to be read/written.
+     * (bytes_base + bytes_res) / bps_limit: used to calcuate
+     *             the total time for completing reading/writting all data.
+     */
+    bytes_res   = (unsigned) nb_sectors * BDRV_SECTOR_SIZE;
+
+    if (bytes_base + bytes_res <= bytes_limit) {
+        if (wait) {
+            *wait = 0;
+        }
+
+        return false;
+    }
+
+    /* Calc approx time to dispatch */
+    wait_time = (bytes_base + bytes_res) / bps_limit - elapsed_time;
+
+    /* When the I/O rate at runtime exceeds the limits,
+     * bs->slice_end need to be extended in order that the current statistic
+     * info can be kept until the timer fire, so it is increased and tuned
+     * based on the result of experiment.
+     */
+    extension = wait_time * NANOSECONDS_PER_SECOND;
+    extension = DIV_ROUND_UP(extension, BLOCK_IO_SLICE_TIME) *
+                BLOCK_IO_SLICE_TIME;
+    bs->slice_end += extension;
+    if (wait) {
+        *wait = wait_time * NANOSECONDS_PER_SECOND;
+    }
+
+    return true;
+}
+
+static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
+                             double elapsed_time, uint64_t *wait)
+{
+    uint64_t iops_limit = 0;
+    double   ios_limit, ios_base;
+    double   slice_time, wait_time;
+
+    if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
+        iops_limit = bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL];
+    } else if (bs->io_limits.iops[is_write]) {
+        iops_limit = bs->io_limits.iops[is_write];
+    } else {
+        if (wait) {
+            *wait = 0;
+        }
+
+        return false;
+    }
+
+    slice_time = bs->slice_end - bs->slice_start;
+    slice_time /= (NANOSECONDS_PER_SECOND);
+    ios_limit  = iops_limit * slice_time;
+    ios_base   = bs->slice_submitted.ios[is_write];
+    if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
+        ios_base += bs->slice_submitted.ios[!is_write];
+    }
+
+    if (ios_base + 1 <= ios_limit) {
+        if (wait) {
+            *wait = 0;
+        }
+
+        return false;
+    }
+
+    /* Calc approx time to dispatch, in seconds */
+    wait_time = (ios_base + 1) / iops_limit;
+    if (wait_time > elapsed_time) {
+        wait_time = wait_time - elapsed_time;
+    } else {
+        wait_time = 0;
+    }
+
+    /* Exceeded current slice, extend it by another slice time */
+    bs->slice_end += BLOCK_IO_SLICE_TIME;
+    if (wait) {
+        *wait = wait_time * NANOSECONDS_PER_SECOND;
+    }
+
+    return true;
+}
+
+static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
+                           bool is_write, int64_t *wait)
+{
+    int64_t  now, max_wait;
+    uint64_t bps_wait = 0, iops_wait = 0;
+    double   elapsed_time;
+    int      bps_ret, iops_ret;
+
+    now = qemu_get_clock_ns(vm_clock);
+    if (now > bs->slice_end) {
+        bs->slice_start = now;
+        bs->slice_end   = now + BLOCK_IO_SLICE_TIME;
+        memset(&bs->slice_submitted, 0, sizeof(bs->slice_submitted));
+    }
+
+    elapsed_time  = now - bs->slice_start;
+    elapsed_time  /= (NANOSECONDS_PER_SECOND);
+
+    bps_ret  = bdrv_exceed_bps_limits(bs, nb_sectors,
+                                      is_write, elapsed_time, &bps_wait);
+    iops_ret = bdrv_exceed_iops_limits(bs, is_write,
+                                      elapsed_time, &iops_wait);
+    if (bps_ret || iops_ret) {
+        max_wait = bps_wait > iops_wait ? bps_wait : iops_wait;
+        if (wait) {
+            *wait = max_wait;
+        }
+
+        now = qemu_get_clock_ns(vm_clock);
+        if (bs->slice_end < now + max_wait) {
+            bs->slice_end = now + max_wait;
+        }
+
+        return true;
+    }
+
+    if (wait) {
+        *wait = 0;
+    }
+
+    bs->slice_submitted.bytes[is_write] += (int64_t)nb_sectors *
+                                           BDRV_SECTOR_SIZE;
+    bs->slice_submitted.ios[is_write]++;
+
+    return false;
+}
+
+/**************************************************************/
+/* async block device emulation */
+
+typedef struct BlockDriverAIOCBSync {
+    BlockDriverAIOCB common;
+    QEMUBH *bh;
+    int ret;
+    /* vector translation state */
+    QEMUIOVector *qiov;
+    uint8_t *bounce;
+    int is_write;
+} BlockDriverAIOCBSync;
+
+static void bdrv_aio_cancel_em(BlockDriverAIOCB *blockacb)
+{
+    BlockDriverAIOCBSync *acb =
+        container_of(blockacb, BlockDriverAIOCBSync, common);
+    qemu_bh_delete(acb->bh);
+    acb->bh = NULL;
+    qemu_aio_release(acb);
+}
+
+static const AIOCBInfo bdrv_em_aiocb_info = {
+    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
+    .cancel             = bdrv_aio_cancel_em,
+};
+
+static void bdrv_aio_bh_cb(void *opaque)
+{
+    BlockDriverAIOCBSync *acb = opaque;
+
+    if (!acb->is_write)
+        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
+    qemu_vfree(acb->bounce);
+    acb->common.cb(acb->common.opaque, acb->ret);
+    qemu_bh_delete(acb->bh);
+    acb->bh = NULL;
+    qemu_aio_release(acb);
+}
+
+static BlockDriverAIOCB *bdrv_aio_rw_vector(BlockDriverState *bs,
+                                            int64_t sector_num,
+                                            QEMUIOVector *qiov,
+                                            int nb_sectors,
+                                            BlockDriverCompletionFunc *cb,
+                                            void *opaque,
+                                            int is_write)
+
+{
+    BlockDriverAIOCBSync *acb;
+
+    acb = qemu_aio_get(&bdrv_em_aiocb_info, bs, cb, opaque);
+    acb->is_write = is_write;
+    acb->qiov = qiov;
+    acb->bounce = qemu_blockalign(bs, qiov->size);
+    acb->bh = qemu_bh_new(bdrv_aio_bh_cb, acb);
+
+    if (is_write) {
+        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
+        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
+    } else {
+        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
+    }
+
+    qemu_bh_schedule(acb->bh);
+
+    return &acb->common;
+}
+
+static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
+}
+
+static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 1);
+}
+
+
+typedef struct BlockDriverAIOCBCoroutine {
+    BlockDriverAIOCB common;
+    BlockRequest req;
+    bool is_write;
+    bool *done;
+    QEMUBH* bh;
+} BlockDriverAIOCBCoroutine;
+
+static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
+{
+    BlockDriverAIOCBCoroutine *acb =
+        container_of(blockacb, BlockDriverAIOCBCoroutine, common);
+    bool done = false;
+
+    acb->done = &done;
+    while (!done) {
+        qemu_aio_wait();
+    }
+}
+
+static const AIOCBInfo bdrv_em_co_aiocb_info = {
+    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
+    .cancel             = bdrv_aio_co_cancel_em,
+};
+
+static void bdrv_co_em_bh(void *opaque)
+{
+    BlockDriverAIOCBCoroutine *acb = opaque;
+
+    acb->common.cb(acb->common.opaque, acb->req.error);
+
+    if (acb->done) {
+        *acb->done = true;
+    }
+
+    qemu_bh_delete(acb->bh);
+    qemu_aio_release(acb);
+}
+
+/* Invoke bdrv_co_do_readv/bdrv_co_do_writev */
+static void coroutine_fn bdrv_co_do_rw(void *opaque)
+{
+    BlockDriverAIOCBCoroutine *acb = opaque;
+    BlockDriverState *bs = acb->common.bs;
+
+    if (!acb->is_write) {
+        acb->req.error = bdrv_co_do_readv(bs, acb->req.sector,
+            acb->req.nb_sectors, acb->req.qiov, 0);
+    } else {
+        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
+            acb->req.nb_sectors, acb->req.qiov, 0);
+    }
+
+    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
+    qemu_bh_schedule(acb->bh);
+}
+
+static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
+                                               int64_t sector_num,
+                                               QEMUIOVector *qiov,
+                                               int nb_sectors,
+                                               BlockDriverCompletionFunc *cb,
+                                               void *opaque,
+                                               bool is_write)
+{
+    Coroutine *co;
+    BlockDriverAIOCBCoroutine *acb;
+
+    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
+    acb->req.sector = sector_num;
+    acb->req.nb_sectors = nb_sectors;
+    acb->req.qiov = qiov;
+    acb->is_write = is_write;
+    acb->done = NULL;
+
+    co = qemu_coroutine_create(bdrv_co_do_rw);
+    qemu_coroutine_enter(co, acb);
+
+    return &acb->common;
+}
+
+static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
+{
+    BlockDriverAIOCBCoroutine *acb = opaque;
+    BlockDriverState *bs = acb->common.bs;
+
+    acb->req.error = bdrv_co_flush(bs);
+    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
+    qemu_bh_schedule(acb->bh);
+}
+
+BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    trace_bdrv_aio_flush(bs, opaque);
+
+    Coroutine *co;
+    BlockDriverAIOCBCoroutine *acb;
+
+    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
+    acb->done = NULL;
+
+    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
+    qemu_coroutine_enter(co, acb);
+
+    return &acb->common;
+}
+
+static void coroutine_fn bdrv_aio_discard_co_entry(void *opaque)
+{
+    BlockDriverAIOCBCoroutine *acb = opaque;
+    BlockDriverState *bs = acb->common.bs;
+
+    acb->req.error = bdrv_co_discard(bs, acb->req.sector, acb->req.nb_sectors);
+    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
+    qemu_bh_schedule(acb->bh);
+}
+
+BlockDriverAIOCB *bdrv_aio_discard(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    Coroutine *co;
+    BlockDriverAIOCBCoroutine *acb;
+
+    trace_bdrv_aio_discard(bs, sector_num, nb_sectors, opaque);
+
+    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
+    acb->req.sector = sector_num;
+    acb->req.nb_sectors = nb_sectors;
+    acb->done = NULL;
+    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
+    qemu_coroutine_enter(co, acb);
+
+    return &acb->common;
+}
+
+void bdrv_init(void)
+{
+    module_call_init(MODULE_INIT_BLOCK);
+}
+
+void bdrv_init_with_whitelist(void)
+{
+    use_bdrv_whitelist = 1;
+    bdrv_init();
+}
+
+void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
+                   BlockDriverCompletionFunc *cb, void *opaque)
+{
+    BlockDriverAIOCB *acb;
+
+    acb = g_slice_alloc(aiocb_info->aiocb_size);
+    acb->aiocb_info = aiocb_info;
+    acb->bs = bs;
+    acb->cb = cb;
+    acb->opaque = opaque;
+    return acb;
+}
+
+void qemu_aio_release(void *p)
+{
+    BlockDriverAIOCB *acb = p;
+    g_slice_free1(acb->aiocb_info->aiocb_size, acb);
+}
+
+/**************************************************************/
+/* Coroutine block device emulation */
+
+typedef struct CoroutineIOCompletion {
+    Coroutine *coroutine;
+    int ret;
+} CoroutineIOCompletion;
+
+static void bdrv_co_io_em_complete(void *opaque, int ret)
+{
+    CoroutineIOCompletion *co = opaque;
+
+    co->ret = ret;
+    qemu_coroutine_enter(co->coroutine, NULL);
+}
+
+static int coroutine_fn bdrv_co_io_em(BlockDriverState *bs, int64_t sector_num,
+                                      int nb_sectors, QEMUIOVector *iov,
+                                      bool is_write)
+{
+    CoroutineIOCompletion co = {
+        .coroutine = qemu_coroutine_self(),
+    };
+    BlockDriverAIOCB *acb;
+
+    if (is_write) {
+        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
+                                       bdrv_co_io_em_complete, &co);
+    } else {
+        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
+                                      bdrv_co_io_em_complete, &co);
+    }
+
+    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
+    if (!acb) {
+        return -EIO;
+    }
+    qemu_coroutine_yield();
+
+    return co.ret;
+}
+
+static int coroutine_fn bdrv_co_readv_em(BlockDriverState *bs,
+                                         int64_t sector_num, int nb_sectors,
+                                         QEMUIOVector *iov)
+{
+    return bdrv_co_io_em(bs, sector_num, nb_sectors, iov, false);
+}
+
+static int coroutine_fn bdrv_co_writev_em(BlockDriverState *bs,
+                                         int64_t sector_num, int nb_sectors,
+                                         QEMUIOVector *iov)
+{
+    return bdrv_co_io_em(bs, sector_num, nb_sectors, iov, true);
+}
+
+static void coroutine_fn bdrv_flush_co_entry(void *opaque)
+{
+    RwCo *rwco = opaque;
+
+    rwco->ret = bdrv_co_flush(rwco->bs);
+}
+
+int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
+{
+    int ret;
+
+    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
+        return 0;
+    }
+
+    /* Write back cached data to the OS even with cache=unsafe */
+    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
+    if (bs->drv->bdrv_co_flush_to_os) {
+        ret = bs->drv->bdrv_co_flush_to_os(bs);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    /* But don't actually force it to the disk with cache=unsafe */
+    if (bs->open_flags & BDRV_O_NO_FLUSH) {
+        goto flush_parent;
+    }
+
+    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_DISK);
+    if (bs->drv->bdrv_co_flush_to_disk) {
+        ret = bs->drv->bdrv_co_flush_to_disk(bs);
+    } else if (bs->drv->bdrv_aio_flush) {
+        BlockDriverAIOCB *acb;
+        CoroutineIOCompletion co = {
+            .coroutine = qemu_coroutine_self(),
+        };
+
+        acb = bs->drv->bdrv_aio_flush(bs, bdrv_co_io_em_complete, &co);
+        if (acb == NULL) {
+            ret = -EIO;
+        } else {
+            qemu_coroutine_yield();
+            ret = co.ret;
+        }
+    } else {
+        /*
+         * Some block drivers always operate in either writethrough or unsafe
+         * mode and don't support bdrv_flush therefore. Usually qemu doesn't
+         * know how the server works (because the behaviour is hardcoded or
+         * depends on server-side configuration), so we can't ensure that
+         * everything is safe on disk. Returning an error doesn't work because
+         * that would break guests even if the server operates in writethrough
+         * mode.
+         *
+         * Let's hope the user knows what he's doing.
+         */
+        ret = 0;
+    }
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Now flush the underlying protocol.  It will also have BDRV_O_NO_FLUSH
+     * in the case of cache=unsafe, so there are no useless flushes.
+     */
+flush_parent:
+    return bdrv_co_flush(bs->file);
+}
+
+void bdrv_invalidate_cache(BlockDriverState *bs)
+{
+    if (bs->drv && bs->drv->bdrv_invalidate_cache) {
+        bs->drv->bdrv_invalidate_cache(bs);
+    }
+}
+
+void bdrv_invalidate_cache_all(void)
+{
+    BlockDriverState *bs;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        bdrv_invalidate_cache(bs);
+    }
+}
+
+void bdrv_clear_incoming_migration_all(void)
+{
+    BlockDriverState *bs;
+
+    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        bs->open_flags = bs->open_flags & ~(BDRV_O_INCOMING);
+    }
+}
+
+int bdrv_flush(BlockDriverState *bs)
+{
+    Coroutine *co;
+    RwCo rwco = {
+        .bs = bs,
+        .ret = NOT_DONE,
+    };
+
+    if (qemu_in_coroutine()) {
+        /* Fast-path if already in coroutine context */
+        bdrv_flush_co_entry(&rwco);
+    } else {
+        co = qemu_coroutine_create(bdrv_flush_co_entry);
+        qemu_coroutine_enter(co, &rwco);
+        while (rwco.ret == NOT_DONE) {
+            qemu_aio_wait();
+        }
+    }
+
+    return rwco.ret;
+}
+
+static void coroutine_fn bdrv_discard_co_entry(void *opaque)
+{
+    RwCo *rwco = opaque;
+
+    rwco->ret = bdrv_co_discard(rwco->bs, rwco->sector_num, rwco->nb_sectors);
+}
+
+int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
+                                 int nb_sectors)
+{
+    if (!bs->drv) {
+        return -ENOMEDIUM;
+    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
+        return -EIO;
+    } else if (bs->read_only) {
+        return -EROFS;
+    }
+
+    if (bs->dirty_bitmap) {
+        bdrv_reset_dirty(bs, sector_num, nb_sectors);
+    }
+
+    /* Do nothing if disabled.  */
+    if (!(bs->open_flags & BDRV_O_UNMAP)) {
+        return 0;
+    }
+
+    if (bs->drv->bdrv_co_discard) {
+        return bs->drv->bdrv_co_discard(bs, sector_num, nb_sectors);
+    } else if (bs->drv->bdrv_aio_discard) {
+        BlockDriverAIOCB *acb;
+        CoroutineIOCompletion co = {
+            .coroutine = qemu_coroutine_self(),
+        };
+
+        acb = bs->drv->bdrv_aio_discard(bs, sector_num, nb_sectors,
+                                        bdrv_co_io_em_complete, &co);
+        if (acb == NULL) {
+            return -EIO;
+        } else {
+            qemu_coroutine_yield();
+            return co.ret;
+        }
+    } else {
+        return 0;
+    }
+}
+
+int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
+{
+    Coroutine *co;
+    RwCo rwco = {
+        .bs = bs,
+        .sector_num = sector_num,
+        .nb_sectors = nb_sectors,
+        .ret = NOT_DONE,
+    };
+
+    if (qemu_in_coroutine()) {
+        /* Fast-path if already in coroutine context */
+        bdrv_discard_co_entry(&rwco);
+    } else {
+        co = qemu_coroutine_create(bdrv_discard_co_entry);
+        qemu_coroutine_enter(co, &rwco);
+        while (rwco.ret == NOT_DONE) {
+            qemu_aio_wait();
+        }
+    }
+
+    return rwco.ret;
+}
+
+/**************************************************************/
+/* removable device support */
+
+/**
+ * Return TRUE if the media is present
+ */
+int bdrv_is_inserted(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+
+    if (!drv)
+        return 0;
+    if (!drv->bdrv_is_inserted)
+        return 1;
+    return drv->bdrv_is_inserted(bs);
+}
+
+/**
+ * Return whether the media changed since the last call to this
+ * function, or -ENOTSUP if we don't know.  Most drivers don't know.
+ */
+int bdrv_media_changed(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+
+    if (drv && drv->bdrv_media_changed) {
+        return drv->bdrv_media_changed(bs);
+    }
+    return -ENOTSUP;
+}
+
+/**
+ * If eject_flag is TRUE, eject the media. Otherwise, close the tray
+ */
+void bdrv_eject(BlockDriverState *bs, bool eject_flag)
+{
+    BlockDriver *drv = bs->drv;
+
+    if (drv && drv->bdrv_eject) {
+        drv->bdrv_eject(bs, eject_flag);
+    }
+
+    if (bs->device_name[0] != '\0') {
+        bdrv_emit_qmp_eject_event(bs, eject_flag);
+    }
+}
+
+/**
+ * Lock or unlock the media (if it is locked, the user won't be able
+ * to eject it manually).
+ */
+void bdrv_lock_medium(BlockDriverState *bs, bool locked)
+{
+    BlockDriver *drv = bs->drv;
+
+    trace_bdrv_lock_medium(bs, locked);
+
+    if (drv && drv->bdrv_lock_medium) {
+        drv->bdrv_lock_medium(bs, locked);
+    }
+}
+
+/* needed for generic scsi interface */
+
+int bdrv_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
+{
+    BlockDriver *drv = bs->drv;
+
+    if (drv && drv->bdrv_ioctl)
+        return drv->bdrv_ioctl(bs, req, buf);
+    return -ENOTSUP;
+}
+
+BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
+        unsigned long int req, void *buf,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    BlockDriver *drv = bs->drv;
+
+    if (drv && drv->bdrv_aio_ioctl)
+        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
+    return NULL;
+}
+
+void bdrv_set_buffer_alignment(BlockDriverState *bs, int align)
+{
+    bs->buffer_alignment = align;
+}
+
+void *qemu_blockalign(BlockDriverState *bs, size_t size)
+{
+    return qemu_memalign((bs && bs->buffer_alignment) ? bs->buffer_alignment : 512, size);
+}
+
+/*
+ * Check if all memory in this vector is sector aligned.
+ */
+bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
+{
+    int i;
+
+    for (i = 0; i < qiov->niov; i++) {
+        if ((uintptr_t) qiov->iov[i].iov_base % bs->buffer_alignment) {
+            return false;
+        }
+    }
+
+    return true;
+}
+
+void bdrv_set_dirty_tracking(BlockDriverState *bs, int granularity)
+{
+    int64_t bitmap_size;
+
+    assert((granularity & (granularity - 1)) == 0);
+
+    if (granularity) {
+        granularity >>= BDRV_SECTOR_BITS;
+        assert(!bs->dirty_bitmap);
+        bitmap_size = (bdrv_getlength(bs) >> BDRV_SECTOR_BITS);
+        bs->dirty_bitmap = hbitmap_alloc(bitmap_size, ffs(granularity) - 1);
+    } else {
+        if (bs->dirty_bitmap) {
+            hbitmap_free(bs->dirty_bitmap);
+            bs->dirty_bitmap = NULL;
+        }
+    }
+}
+
+int bdrv_get_dirty(BlockDriverState *bs, int64_t sector)
+{
+    if (bs->dirty_bitmap) {
+        return hbitmap_get(bs->dirty_bitmap, sector);
+    } else {
+        return 0;
+    }
+}
+
+void bdrv_dirty_iter_init(BlockDriverState *bs, HBitmapIter *hbi)
+{
+    hbitmap_iter_init(hbi, bs->dirty_bitmap, 0);
+}
+
+void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
+                    int nr_sectors)
+{
+    hbitmap_set(bs->dirty_bitmap, cur_sector, nr_sectors);
+}
+
+void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector,
+                      int nr_sectors)
+{
+    hbitmap_reset(bs->dirty_bitmap, cur_sector, nr_sectors);
+}
+
+int64_t bdrv_get_dirty_count(BlockDriverState *bs)
+{
+    if (bs->dirty_bitmap) {
+        return hbitmap_count(bs->dirty_bitmap);
+    } else {
+        return 0;
+    }
+}
+
+void bdrv_set_in_use(BlockDriverState *bs, int in_use)
+{
+    assert(bs->in_use != in_use);
+    bs->in_use = in_use;
+}
+
+int bdrv_in_use(BlockDriverState *bs)
+{
+    return bs->in_use;
+}
+
+void bdrv_iostatus_enable(BlockDriverState *bs)
+{
+    bs->iostatus_enabled = true;
+    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
+}
+
+/* The I/O status is only enabled if the drive explicitly
+ * enables it _and_ the VM is configured to stop on errors */
+bool bdrv_iostatus_is_enabled(const BlockDriverState *bs)
+{
+    return (bs->iostatus_enabled &&
+           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
+            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
+            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
+}
+
+void bdrv_iostatus_disable(BlockDriverState *bs)
+{
+    bs->iostatus_enabled = false;
+}
+
+void bdrv_iostatus_reset(BlockDriverState *bs)
+{
+    if (bdrv_iostatus_is_enabled(bs)) {
+        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
+        if (bs->job) {
+            block_job_iostatus_reset(bs->job);
+        }
+    }
+}
+
+void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
+{
+    assert(bdrv_iostatus_is_enabled(bs));
+    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
+        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
+                                         BLOCK_DEVICE_IO_STATUS_FAILED;
+    }
+}
+
+void
+bdrv_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie, int64_t bytes,
+        enum BlockAcctType type)
+{
+    assert(type < BDRV_MAX_IOTYPE);
+
+    cookie->bytes = bytes;
+    cookie->start_time_ns = get_clock();
+    cookie->type = type;
+}
+
+void
+bdrv_acct_done(BlockDriverState *bs, BlockAcctCookie *cookie)
+{
+    assert(cookie->type < BDRV_MAX_IOTYPE);
+
+    bs->nr_bytes[cookie->type] += cookie->bytes;
+    bs->nr_ops[cookie->type]++;
+    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
+}
+
+void bdrv_img_create(const char *filename, const char *fmt,
+                     const char *base_filename, const char *base_fmt,
+                     char *options, uint64_t img_size, int flags,
+                     Error **errp, bool quiet)
+{
+    QEMUOptionParameter *param = NULL, *create_options = NULL;
+    QEMUOptionParameter *backing_fmt, *backing_file, *size;
+    BlockDriverState *bs = NULL;
+    BlockDriver *drv, *proto_drv;
+    BlockDriver *backing_drv = NULL;
+    int ret = 0;
+
+    /* Find driver and parse its options */
+    drv = bdrv_find_format(fmt);
+    if (!drv) {
+        error_setg(errp, "Unknown file format '%s'", fmt);
+        return;
+    }
+
+    proto_drv = bdrv_find_protocol(filename, true);
+    if (!proto_drv) {
+        error_setg(errp, "Unknown protocol '%s'", filename);
+        return;
+    }
+
+    create_options = append_option_parameters(create_options,
+                                              drv->create_options);
+    create_options = append_option_parameters(create_options,
+                                              proto_drv->create_options);
+
+    /* Create parameter list with default values */
+    param = parse_option_parameters("", create_options, param);
+
+    set_option_parameter_int(param, BLOCK_OPT_SIZE, img_size);
+
+    /* Parse -o options */
+    if (options) {
+        param = parse_option_parameters(options, create_options, param);
+        if (param == NULL) {
+            error_setg(errp, "Invalid options for file format '%s'.", fmt);
+            goto out;
+        }
+    }
+
+    if (base_filename) {
+        if (set_option_parameter(param, BLOCK_OPT_BACKING_FILE,
+                                 base_filename)) {
+            error_setg(errp, "Backing file not supported for file format '%s'",
+                       fmt);
+            goto out;
+        }
+    }
+
+    if (base_fmt) {
+        if (set_option_parameter(param, BLOCK_OPT_BACKING_FMT, base_fmt)) {
+            error_setg(errp, "Backing file format not supported for file "
+                             "format '%s'", fmt);
+            goto out;
+        }
+    }
+
+    backing_file = get_option_parameter(param, BLOCK_OPT_BACKING_FILE);
+    if (backing_file && backing_file->value.s) {
+        if (!strcmp(filename, backing_file->value.s)) {
+            error_setg(errp, "Error: Trying to create an image with the "
+                             "same filename as the backing file");
+            goto out;
+        }
+    }
+
+    backing_fmt = get_option_parameter(param, BLOCK_OPT_BACKING_FMT);
+    if (backing_fmt && backing_fmt->value.s) {
+        backing_drv = bdrv_find_format(backing_fmt->value.s);
+        if (!backing_drv) {
+            error_setg(errp, "Unknown backing file format '%s'",
+                       backing_fmt->value.s);
+            goto out;
+        }
+    }
+
+    // The size for the image must always be specified, with one exception:
+    // If we are using a backing file, we can obtain the size from there
+    size = get_option_parameter(param, BLOCK_OPT_SIZE);
+    if (size && size->value.n == -1) {
+        if (backing_file && backing_file->value.s) {
+            uint64_t size;
+            char buf[32];
+            int back_flags;
+
+            /* backing files always opened read-only */
+            back_flags =
+                flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);
+
+            bs = bdrv_new("");
+
+            ret = bdrv_open(bs, backing_file->value.s, NULL, back_flags,
+                            backing_drv);
+            if (ret < 0) {
+                error_setg_errno(errp, -ret, "Could not open '%s'",
+                                 backing_file->value.s);
+                goto out;
+            }
+            bdrv_get_geometry(bs, &size);
+            size *= 512;
+
+            snprintf(buf, sizeof(buf), "%" PRId64, size);
+            set_option_parameter(param, BLOCK_OPT_SIZE, buf);
+        } else {
+            error_setg(errp, "Image creation needs a size parameter");
+            goto out;
+        }
+    }
+
+    if (!quiet) {
+        printf("Formatting '%s', fmt=%s ", filename, fmt);
+        print_option_parameters(param);
+        puts("");
+    }
+    ret = bdrv_create(drv, filename, param);
+    if (ret < 0) {
+        if (ret == -ENOTSUP) {
+            error_setg(errp,"Formatting or formatting option not supported for "
+                            "file format '%s'", fmt);
+        } else if (ret == -EFBIG) {
+            const char *cluster_size_hint = "";
+            if (get_option_parameter(create_options, BLOCK_OPT_CLUSTER_SIZE)) {
+                cluster_size_hint = " (try using a larger cluster size)";
+            }
+            error_setg(errp, "The image size is too large for file format '%s'%s",
+                       fmt, cluster_size_hint);
+        } else {
+            error_setg(errp, "%s: error while creating %s: %s", filename, fmt,
+                       strerror(-ret));
+        }
+    }
+
+out:
+    free_option_parameters(create_options);
+    free_option_parameters(param);
+
+    if (bs) {
+        bdrv_delete(bs);
+    }
+}
+
+AioContext *bdrv_get_aio_context(BlockDriverState *bs)
+{
+    /* Currently BlockDriverState always uses the main loop AioContext */
+    return qemu_get_aio_context();
+}
+
+void bdrv_add_before_write_notifier(BlockDriverState *bs,
+                                    NotifierWithReturn *notifier)
+{
+    notifier_with_return_list_add(&bs->before_write_notifiers, notifier);
+}
diff --git a/contrib/qemu/block/qcow.c b/contrib/qemu/block/qcow.c
new file mode 100644
index 000000000..5239bd68f
--- /dev/null
+++ b/contrib/qemu/block/qcow.c
@@ -0,0 +1,914 @@
+/*
+ * Block driver for the QCOW format
+ *
+ * Copyright (c) 2004-2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "qemu-common.h"
+#include "block/block_int.h"
+#include "qemu/module.h"
+#include <zlib.h>
+#include "qemu/aes.h"
+#include "migration/migration.h"
+
+/**************************************************************/
+/* QEMU COW block driver with compression and encryption support */
+
+#define QCOW_MAGIC (('Q' << 24) | ('F' << 16) | ('I' << 8) | 0xfb)
+#define QCOW_VERSION 1
+
+#define QCOW_CRYPT_NONE 0
+#define QCOW_CRYPT_AES  1
+
+#define QCOW_OFLAG_COMPRESSED (1LL << 63)
+
+typedef struct QCowHeader {
+    uint32_t magic;
+    uint32_t version;
+    uint64_t backing_file_offset;
+    uint32_t backing_file_size;
+    uint32_t mtime;
+    uint64_t size; /* in bytes */
+    uint8_t cluster_bits;
+    uint8_t l2_bits;
+    uint32_t crypt_method;
+    uint64_t l1_table_offset;
+} QCowHeader;
+
+#define L2_CACHE_SIZE 16
+
+typedef struct BDRVQcowState {
+    int cluster_bits;
+    int cluster_size;
+    int cluster_sectors;
+    int l2_bits;
+    int l2_size;
+    int l1_size;
+    uint64_t cluster_offset_mask;
+    uint64_t l1_table_offset;
+    uint64_t *l1_table;
+    uint64_t *l2_cache;
+    uint64_t l2_cache_offsets[L2_CACHE_SIZE];
+    uint32_t l2_cache_counts[L2_CACHE_SIZE];
+    uint8_t *cluster_cache;
+    uint8_t *cluster_data;
+    uint64_t cluster_cache_offset;
+    uint32_t crypt_method; /* current crypt method, 0 if no key yet */
+    uint32_t crypt_method_header;
+    AES_KEY aes_encrypt_key;
+    AES_KEY aes_decrypt_key;
+    CoMutex lock;
+    Error *migration_blocker;
+} BDRVQcowState;
+
+static int decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset);
+
+static int qcow_probe(const uint8_t *buf, int buf_size, const char *filename)
+{
+    const QCowHeader *cow_header = (const void *)buf;
+
+    if (buf_size >= sizeof(QCowHeader) &&
+        be32_to_cpu(cow_header->magic) == QCOW_MAGIC &&
+        be32_to_cpu(cow_header->version) == QCOW_VERSION)
+        return 100;
+    else
+        return 0;
+}
+
+static int qcow_open(BlockDriverState *bs, QDict *options, int flags)
+{
+    BDRVQcowState *s = bs->opaque;
+    int len, i, shift, ret;
+    QCowHeader header;
+
+    ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
+    if (ret < 0) {
+        goto fail;
+    }
+    be32_to_cpus(&header.magic);
+    be32_to_cpus(&header.version);
+    be64_to_cpus(&header.backing_file_offset);
+    be32_to_cpus(&header.backing_file_size);
+    be32_to_cpus(&header.mtime);
+    be64_to_cpus(&header.size);
+    be32_to_cpus(&header.crypt_method);
+    be64_to_cpus(&header.l1_table_offset);
+
+    if (header.magic != QCOW_MAGIC) {
+        ret = -EMEDIUMTYPE;
+        goto fail;
+    }
+    if (header.version != QCOW_VERSION) {
+        char version[64];
+        snprintf(version, sizeof(version), "QCOW version %d", header.version);
+        qerror_report(QERR_UNKNOWN_BLOCK_FORMAT_FEATURE,
+            bs->device_name, "qcow", version);
+        ret = -ENOTSUP;
+        goto fail;
+    }
+
+    if (header.size <= 1 || header.cluster_bits < 9) {
+        ret = -EINVAL;
+        goto fail;
+    }
+    if (header.crypt_method > QCOW_CRYPT_AES) {
+        ret = -EINVAL;
+        goto fail;
+    }
+    s->crypt_method_header = header.crypt_method;
+    if (s->crypt_method_header) {
+        bs->encrypted = 1;
+    }
+    s->cluster_bits = header.cluster_bits;
+    s->cluster_size = 1 << s->cluster_bits;
+    s->cluster_sectors = 1 << (s->cluster_bits - 9);
+    s->l2_bits = header.l2_bits;
+    s->l2_size = 1 << s->l2_bits;
+    bs->total_sectors = header.size / 512;
+    s->cluster_offset_mask = (1LL << (63 - s->cluster_bits)) - 1;
+
+    /* read the level 1 table */
+    shift = s->cluster_bits + s->l2_bits;
+    s->l1_size = (header.size + (1LL << shift) - 1) >> shift;
+
+    s->l1_table_offset = header.l1_table_offset;
+    s->l1_table = g_malloc(s->l1_size * sizeof(uint64_t));
+
+    ret = bdrv_pread(bs->file, s->l1_table_offset, s->l1_table,
+               s->l1_size * sizeof(uint64_t));
+    if (ret < 0) {
+        goto fail;
+    }
+
+    for(i = 0;i < s->l1_size; i++) {
+        be64_to_cpus(&s->l1_table[i]);
+    }
+    /* alloc L2 cache */
+    s->l2_cache = g_malloc(s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));
+    s->cluster_cache = g_malloc(s->cluster_size);
+    s->cluster_data = g_malloc(s->cluster_size);
+    s->cluster_cache_offset = -1;
+
+    /* read the backing file name */
+    if (header.backing_file_offset != 0) {
+        len = header.backing_file_size;
+        if (len > 1023) {
+            len = 1023;
+        }
+        ret = bdrv_pread(bs->file, header.backing_file_offset,
+                   bs->backing_file, len);
+        if (ret < 0) {
+            goto fail;
+        }
+        bs->backing_file[len] = '\0';
+    }
+
+    /* Disable migration when qcow images are used */
+    error_set(&s->migration_blocker,
+              QERR_BLOCK_FORMAT_FEATURE_NOT_SUPPORTED,
+              "qcow", bs->device_name, "live migration");
+    migrate_add_blocker(s->migration_blocker);
+
+    qemu_co_mutex_init(&s->lock);
+    return 0;
+
+ fail:
+    g_free(s->l1_table);
+    g_free(s->l2_cache);
+    g_free(s->cluster_cache);
+    g_free(s->cluster_data);
+    return ret;
+}
+
+
+/* We have nothing to do for QCOW reopen, stubs just return
+ * success */
+static int qcow_reopen_prepare(BDRVReopenState *state,
+                               BlockReopenQueue *queue, Error **errp)
+{
+    return 0;
+}
+
+static int qcow_set_key(BlockDriverState *bs, const char *key)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint8_t keybuf[16];
+    int len, i;
+
+    memset(keybuf, 0, 16);
+    len = strlen(key);
+    if (len > 16)
+        len = 16;
+    /* XXX: we could compress the chars to 7 bits to increase
+       entropy */
+    for(i = 0;i < len;i++) {
+        keybuf[i] = key[i];
+    }
+    s->crypt_method = s->crypt_method_header;
+
+    if (AES_set_encrypt_key(keybuf, 128, &s->aes_encrypt_key) != 0)
+        return -1;
+    if (AES_set_decrypt_key(keybuf, 128, &s->aes_decrypt_key) != 0)
+        return -1;
+    return 0;
+}
+
+/* The crypt function is compatible with the linux cryptoloop
+   algorithm for < 4 GB images. NOTE: out_buf == in_buf is
+   supported */
+static void encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
+                            uint8_t *out_buf, const uint8_t *in_buf,
+                            int nb_sectors, int enc,
+                            const AES_KEY *key)
+{
+    union {
+        uint64_t ll[2];
+        uint8_t b[16];
+    } ivec;
+    int i;
+
+    for(i = 0; i < nb_sectors; i++) {
+        ivec.ll[0] = cpu_to_le64(sector_num);
+        ivec.ll[1] = 0;
+        AES_cbc_encrypt(in_buf, out_buf, 512, key,
+                        ivec.b, enc);
+        sector_num++;
+        in_buf += 512;
+        out_buf += 512;
+    }
+}
+
+/* 'allocate' is:
+ *
+ * 0 to not allocate.
+ *
+ * 1 to allocate a normal cluster (for sector indexes 'n_start' to
+ * 'n_end')
+ *
+ * 2 to allocate a compressed cluster of size
+ * 'compressed_size'. 'compressed_size' must be > 0 and <
+ * cluster_size
+ *
+ * return 0 if not allocated.
+ */
+static uint64_t get_cluster_offset(BlockDriverState *bs,
+                                   uint64_t offset, int allocate,
+                                   int compressed_size,
+                                   int n_start, int n_end)
+{
+    BDRVQcowState *s = bs->opaque;
+    int min_index, i, j, l1_index, l2_index;
+    uint64_t l2_offset, *l2_table, cluster_offset, tmp;
+    uint32_t min_count;
+    int new_l2_table;
+
+    l1_index = offset >> (s->l2_bits + s->cluster_bits);
+    l2_offset = s->l1_table[l1_index];
+    new_l2_table = 0;
+    if (!l2_offset) {
+        if (!allocate)
+            return 0;
+        /* allocate a new l2 entry */
+        l2_offset = bdrv_getlength(bs->file);
+        /* round to cluster size */
+        l2_offset = (l2_offset + s->cluster_size - 1) & ~(s->cluster_size - 1);
+        /* update the L1 entry */
+        s->l1_table[l1_index] = l2_offset;
+        tmp = cpu_to_be64(l2_offset);
+        if (bdrv_pwrite_sync(bs->file,
+                s->l1_table_offset + l1_index * sizeof(tmp),
+                &tmp, sizeof(tmp)) < 0)
+            return 0;
+        new_l2_table = 1;
+    }
+    for(i = 0; i < L2_CACHE_SIZE; i++) {
+        if (l2_offset == s->l2_cache_offsets[i]) {
+            /* increment the hit count */
+            if (++s->l2_cache_counts[i] == 0xffffffff) {
+                for(j = 0; j < L2_CACHE_SIZE; j++) {
+                    s->l2_cache_counts[j] >>= 1;
+                }
+            }
+            l2_table = s->l2_cache + (i << s->l2_bits);
+            goto found;
+        }
+    }
+    /* not found: load a new entry in the least used one */
+    min_index = 0;
+    min_count = 0xffffffff;
+    for(i = 0; i < L2_CACHE_SIZE; i++) {
+        if (s->l2_cache_counts[i] < min_count) {
+            min_count = s->l2_cache_counts[i];
+            min_index = i;
+        }
+    }
+    l2_table = s->l2_cache + (min_index << s->l2_bits);
+    if (new_l2_table) {
+        memset(l2_table, 0, s->l2_size * sizeof(uint64_t));
+        if (bdrv_pwrite_sync(bs->file, l2_offset, l2_table,
+                s->l2_size * sizeof(uint64_t)) < 0)
+            return 0;
+    } else {
+        if (bdrv_pread(bs->file, l2_offset, l2_table, s->l2_size * sizeof(uint64_t)) !=
+            s->l2_size * sizeof(uint64_t))
+            return 0;
+    }
+    s->l2_cache_offsets[min_index] = l2_offset;
+    s->l2_cache_counts[min_index] = 1;
+ found:
+    l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
+    cluster_offset = be64_to_cpu(l2_table[l2_index]);
+    if (!cluster_offset ||
+        ((cluster_offset & QCOW_OFLAG_COMPRESSED) && allocate == 1)) {
+        if (!allocate)
+            return 0;
+        /* allocate a new cluster */
+        if ((cluster_offset & QCOW_OFLAG_COMPRESSED) &&
+            (n_end - n_start) < s->cluster_sectors) {
+            /* if the cluster is already compressed, we must
+               decompress it in the case it is not completely
+               overwritten */
+            if (decompress_cluster(bs, cluster_offset) < 0)
+                return 0;
+            cluster_offset = bdrv_getlength(bs->file);
+            cluster_offset = (cluster_offset + s->cluster_size - 1) &
+                ~(s->cluster_size - 1);
+            /* write the cluster content */
+            if (bdrv_pwrite(bs->file, cluster_offset, s->cluster_cache, s->cluster_size) !=
+                s->cluster_size)
+                return -1;
+        } else {
+            cluster_offset = bdrv_getlength(bs->file);
+            if (allocate == 1) {
+                /* round to cluster size */
+                cluster_offset = (cluster_offset + s->cluster_size - 1) &
+                    ~(s->cluster_size - 1);
+                bdrv_truncate(bs->file, cluster_offset + s->cluster_size);
+                /* if encrypted, we must initialize the cluster
+                   content which won't be written */
+                if (s->crypt_method &&
+                    (n_end - n_start) < s->cluster_sectors) {
+                    uint64_t start_sect;
+                    start_sect = (offset & ~(s->cluster_size - 1)) >> 9;
+                    memset(s->cluster_data + 512, 0x00, 512);
+                    for(i = 0; i < s->cluster_sectors; i++) {
+                        if (i < n_start || i >= n_end) {
+                            encrypt_sectors(s, start_sect + i,
+                                            s->cluster_data,
+                                            s->cluster_data + 512, 1, 1,
+                                            &s->aes_encrypt_key);
+                            if (bdrv_pwrite(bs->file, cluster_offset + i * 512,
+                                            s->cluster_data, 512) != 512)
+                                return -1;
+                        }
+                    }
+                }
+            } else if (allocate == 2) {
+                cluster_offset |= QCOW_OFLAG_COMPRESSED |
+                    (uint64_t)compressed_size << (63 - s->cluster_bits);
+            }
+        }
+        /* update L2 table */
+        tmp = cpu_to_be64(cluster_offset);
+        l2_table[l2_index] = tmp;
+        if (bdrv_pwrite_sync(bs->file, l2_offset + l2_index * sizeof(tmp),
+                &tmp, sizeof(tmp)) < 0)
+            return 0;
+    }
+    return cluster_offset;
+}
+
+static int coroutine_fn qcow_co_is_allocated(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors, int *pnum)
+{
+    BDRVQcowState *s = bs->opaque;
+    int index_in_cluster, n;
+    uint64_t cluster_offset;
+
+    qemu_co_mutex_lock(&s->lock);
+    cluster_offset = get_cluster_offset(bs, sector_num << 9, 0, 0, 0, 0);
+    qemu_co_mutex_unlock(&s->lock);
+    index_in_cluster = sector_num & (s->cluster_sectors - 1);
+    n = s->cluster_sectors - index_in_cluster;
+    if (n > nb_sectors)
+        n = nb_sectors;
+    *pnum = n;
+    return (cluster_offset != 0);
+}
+
+static int decompress_buffer(uint8_t *out_buf, int out_buf_size,
+                             const uint8_t *buf, int buf_size)
+{
+    z_stream strm1, *strm = &strm1;
+    int ret, out_len;
+
+    memset(strm, 0, sizeof(*strm));
+
+    strm->next_in = (uint8_t *)buf;
+    strm->avail_in = buf_size;
+    strm->next_out = out_buf;
+    strm->avail_out = out_buf_size;
+
+    ret = inflateInit2(strm, -12);
+    if (ret != Z_OK)
+        return -1;
+    ret = inflate(strm, Z_FINISH);
+    out_len = strm->next_out - out_buf;
+    if ((ret != Z_STREAM_END && ret != Z_BUF_ERROR) ||
+        out_len != out_buf_size) {
+        inflateEnd(strm);
+        return -1;
+    }
+    inflateEnd(strm);
+    return 0;
+}
+
+static int decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret, csize;
+    uint64_t coffset;
+
+    coffset = cluster_offset & s->cluster_offset_mask;
+    if (s->cluster_cache_offset != coffset) {
+        csize = cluster_offset >> (63 - s->cluster_bits);
+        csize &= (s->cluster_size - 1);
+        ret = bdrv_pread(bs->file, coffset, s->cluster_data, csize);
+        if (ret != csize)
+            return -1;
+        if (decompress_buffer(s->cluster_cache, s->cluster_size,
+                              s->cluster_data, csize) < 0) {
+            return -1;
+        }
+        s->cluster_cache_offset = coffset;
+    }
+    return 0;
+}
+
+static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num,
+                         int nb_sectors, QEMUIOVector *qiov)
+{
+    BDRVQcowState *s = bs->opaque;
+    int index_in_cluster;
+    int ret = 0, n;
+    uint64_t cluster_offset;
+    struct iovec hd_iov;
+    QEMUIOVector hd_qiov;
+    uint8_t *buf;
+    void *orig_buf;
+
+    if (qiov->niov > 1) {
+        buf = orig_buf = qemu_blockalign(bs, qiov->size);
+    } else {
+        orig_buf = NULL;
+        buf = (uint8_t *)qiov->iov->iov_base;
+    }
+
+    qemu_co_mutex_lock(&s->lock);
+
+    while (nb_sectors != 0) {
+        /* prepare next request */
+        cluster_offset = get_cluster_offset(bs, sector_num << 9,
+                                                 0, 0, 0, 0);
+        index_in_cluster = sector_num & (s->cluster_sectors - 1);
+        n = s->cluster_sectors - index_in_cluster;
+        if (n > nb_sectors) {
+            n = nb_sectors;
+        }
+
+        if (!cluster_offset) {
+            if (bs->backing_hd) {
+                /* read from the base image */
+                hd_iov.iov_base = (void *)buf;
+                hd_iov.iov_len = n * 512;
+                qemu_iovec_init_external(&hd_qiov, &hd_iov, 1);
+                qemu_co_mutex_unlock(&s->lock);
+                ret = bdrv_co_readv(bs->backing_hd, sector_num,
+                                    n, &hd_qiov);
+                qemu_co_mutex_lock(&s->lock);
+                if (ret < 0) {
+                    goto fail;
+                }
+            } else {
+                /* Note: in this case, no need to wait */
+                memset(buf, 0, 512 * n);
+            }
+        } else if (cluster_offset & QCOW_OFLAG_COMPRESSED) {
+            /* add AIO support for compressed blocks ? */
+            if (decompress_cluster(bs, cluster_offset) < 0) {
+                goto fail;
+            }
+            memcpy(buf,
+                   s->cluster_cache + index_in_cluster * 512, 512 * n);
+        } else {
+            if ((cluster_offset & 511) != 0) {
+                goto fail;
+            }
+            hd_iov.iov_base = (void *)buf;
+            hd_iov.iov_len = n * 512;
+            qemu_iovec_init_external(&hd_qiov, &hd_iov, 1);
+            qemu_co_mutex_unlock(&s->lock);
+            ret = bdrv_co_readv(bs->file,
+                                (cluster_offset >> 9) + index_in_cluster,
+                                n, &hd_qiov);
+            qemu_co_mutex_lock(&s->lock);
+            if (ret < 0) {
+                break;
+            }
+            if (s->crypt_method) {
+                encrypt_sectors(s, sector_num, buf, buf,
+                                n, 0,
+                                &s->aes_decrypt_key);
+            }
+        }
+        ret = 0;
+
+        nb_sectors -= n;
+        sector_num += n;
+        buf += n * 512;
+    }
+
+done:
+    qemu_co_mutex_unlock(&s->lock);
+
+    if (qiov->niov > 1) {
+        qemu_iovec_from_buf(qiov, 0, orig_buf, qiov->size);
+        qemu_vfree(orig_buf);
+    }
+
+    return ret;
+
+fail:
+    ret = -EIO;
+    goto done;
+}
+
+static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num,
+                          int nb_sectors, QEMUIOVector *qiov)
+{
+    BDRVQcowState *s = bs->opaque;
+    int index_in_cluster;
+    uint64_t cluster_offset;
+    const uint8_t *src_buf;
+    int ret = 0, n;
+    uint8_t *cluster_data = NULL;
+    struct iovec hd_iov;
+    QEMUIOVector hd_qiov;
+    uint8_t *buf;
+    void *orig_buf;
+
+    s->cluster_cache_offset = -1; /* disable compressed cache */
+
+    if (qiov->niov > 1) {
+        buf = orig_buf = qemu_blockalign(bs, qiov->size);
+        qemu_iovec_to_buf(qiov, 0, buf, qiov->size);
+    } else {
+        orig_buf = NULL;
+        buf = (uint8_t *)qiov->iov->iov_base;
+    }
+
+    qemu_co_mutex_lock(&s->lock);
+
+    while (nb_sectors != 0) {
+
+        index_in_cluster = sector_num & (s->cluster_sectors - 1);
+        n = s->cluster_sectors - index_in_cluster;
+        if (n > nb_sectors) {
+            n = nb_sectors;
+        }
+        cluster_offset = get_cluster_offset(bs, sector_num << 9, 1, 0,
+                                            index_in_cluster,
+                                            index_in_cluster + n);
+        if (!cluster_offset || (cluster_offset & 511) != 0) {
+            ret = -EIO;
+            break;
+        }
+        if (s->crypt_method) {
+            if (!cluster_data) {
+                cluster_data = g_malloc0(s->cluster_size);
+            }
+            encrypt_sectors(s, sector_num, cluster_data, buf,
+                            n, 1, &s->aes_encrypt_key);
+            src_buf = cluster_data;
+        } else {
+            src_buf = buf;
+        }
+
+        hd_iov.iov_base = (void *)src_buf;
+        hd_iov.iov_len = n * 512;
+        qemu_iovec_init_external(&hd_qiov, &hd_iov, 1);
+        qemu_co_mutex_unlock(&s->lock);
+        ret = bdrv_co_writev(bs->file,
+                             (cluster_offset >> 9) + index_in_cluster,
+                             n, &hd_qiov);
+        qemu_co_mutex_lock(&s->lock);
+        if (ret < 0) {
+            break;
+        }
+        ret = 0;
+
+        nb_sectors -= n;
+        sector_num += n;
+        buf += n * 512;
+    }
+    qemu_co_mutex_unlock(&s->lock);
+
+    if (qiov->niov > 1) {
+        qemu_vfree(orig_buf);
+    }
+    g_free(cluster_data);
+
+    return ret;
+}
+
+static void qcow_close(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+
+    g_free(s->l1_table);
+    g_free(s->l2_cache);
+    g_free(s->cluster_cache);
+    g_free(s->cluster_data);
+
+    migrate_del_blocker(s->migration_blocker);
+    error_free(s->migration_blocker);
+}
+
+static int qcow_create(const char *filename, QEMUOptionParameter *options)
+{
+    int header_size, backing_filename_len, l1_size, shift, i;
+    QCowHeader header;
+    uint8_t *tmp;
+    int64_t total_size = 0;
+    const char *backing_file = NULL;
+    int flags = 0;
+    int ret;
+    BlockDriverState *qcow_bs;
+
+    /* Read out options */
+    while (options && options->name) {
+        if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
+            total_size = options->value.n / 512;
+        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
+            backing_file = options->value.s;
+        } else if (!strcmp(options->name, BLOCK_OPT_ENCRYPT)) {
+            flags |= options->value.n ? BLOCK_FLAG_ENCRYPT : 0;
+        }
+        options++;
+    }
+
+    ret = bdrv_create_file(filename, options);
+    if (ret < 0) {
+        return ret;
+    }
+
+    ret = bdrv_file_open(&qcow_bs, filename, NULL, BDRV_O_RDWR);
+    if (ret < 0) {
+        return ret;
+    }
+
+    ret = bdrv_truncate(qcow_bs, 0);
+    if (ret < 0) {
+        goto exit;
+    }
+
+    memset(&header, 0, sizeof(header));
+    header.magic = cpu_to_be32(QCOW_MAGIC);
+    header.version = cpu_to_be32(QCOW_VERSION);
+    header.size = cpu_to_be64(total_size * 512);
+    header_size = sizeof(header);
+    backing_filename_len = 0;
+    if (backing_file) {
+        if (strcmp(backing_file, "fat:")) {
+            header.backing_file_offset = cpu_to_be64(header_size);
+            backing_filename_len = strlen(backing_file);
+            header.backing_file_size = cpu_to_be32(backing_filename_len);
+            header_size += backing_filename_len;
+        } else {
+            /* special backing file for vvfat */
+            backing_file = NULL;
+        }
+        header.cluster_bits = 9; /* 512 byte cluster to avoid copying
+                                    unmodifyed sectors */
+        header.l2_bits = 12; /* 32 KB L2 tables */
+    } else {
+        header.cluster_bits = 12; /* 4 KB clusters */
+        header.l2_bits = 9; /* 4 KB L2 tables */
+    }
+    header_size = (header_size + 7) & ~7;
+    shift = header.cluster_bits + header.l2_bits;
+    l1_size = ((total_size * 512) + (1LL << shift) - 1) >> shift;
+
+    header.l1_table_offset = cpu_to_be64(header_size);
+    if (flags & BLOCK_FLAG_ENCRYPT) {
+        header.crypt_method = cpu_to_be32(QCOW_CRYPT_AES);
+    } else {
+        header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
+    }
+
+    /* write all the data */
+    ret = bdrv_pwrite(qcow_bs, 0, &header, sizeof(header));
+    if (ret != sizeof(header)) {
+        goto exit;
+    }
+
+    if (backing_file) {
+        ret = bdrv_pwrite(qcow_bs, sizeof(header),
+            backing_file, backing_filename_len);
+        if (ret != backing_filename_len) {
+            goto exit;
+        }
+    }
+
+    tmp = g_malloc0(BDRV_SECTOR_SIZE);
+    for (i = 0; i < ((sizeof(uint64_t)*l1_size + BDRV_SECTOR_SIZE - 1)/
+        BDRV_SECTOR_SIZE); i++) {
+        ret = bdrv_pwrite(qcow_bs, header_size +
+            BDRV_SECTOR_SIZE*i, tmp, BDRV_SECTOR_SIZE);
+        if (ret != BDRV_SECTOR_SIZE) {
+            g_free(tmp);
+            goto exit;
+        }
+    }
+
+    g_free(tmp);
+    ret = 0;
+exit:
+    bdrv_delete(qcow_bs);
+    return ret;
+}
+
+static int qcow_make_empty(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint32_t l1_length = s->l1_size * sizeof(uint64_t);
+    int ret;
+
+    memset(s->l1_table, 0, l1_length);
+    if (bdrv_pwrite_sync(bs->file, s->l1_table_offset, s->l1_table,
+            l1_length) < 0)
+        return -1;
+    ret = bdrv_truncate(bs->file, s->l1_table_offset + l1_length);
+    if (ret < 0)
+        return ret;
+
+    memset(s->l2_cache, 0, s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));
+    memset(s->l2_cache_offsets, 0, L2_CACHE_SIZE * sizeof(uint64_t));
+    memset(s->l2_cache_counts, 0, L2_CACHE_SIZE * sizeof(uint32_t));
+
+    return 0;
+}
+
+/* XXX: put compressed sectors first, then all the cluster aligned
+   tables to avoid losing bytes in alignment */
+static int qcow_write_compressed(BlockDriverState *bs, int64_t sector_num,
+                                 const uint8_t *buf, int nb_sectors)
+{
+    BDRVQcowState *s = bs->opaque;
+    z_stream strm;
+    int ret, out_len;
+    uint8_t *out_buf;
+    uint64_t cluster_offset;
+
+    if (nb_sectors != s->cluster_sectors) {
+        ret = -EINVAL;
+
+        /* Zero-pad last write if image size is not cluster aligned */
+        if (sector_num + nb_sectors == bs->total_sectors &&
+            nb_sectors < s->cluster_sectors) {
+            uint8_t *pad_buf = qemu_blockalign(bs, s->cluster_size);
+            memset(pad_buf, 0, s->cluster_size);
+            memcpy(pad_buf, buf, nb_sectors * BDRV_SECTOR_SIZE);
+            ret = qcow_write_compressed(bs, sector_num,
+                                        pad_buf, s->cluster_sectors);
+            qemu_vfree(pad_buf);
+        }
+        return ret;
+    }
+
+    out_buf = g_malloc(s->cluster_size + (s->cluster_size / 1000) + 128);
+
+    /* best compression, small window, no zlib header */
+    memset(&strm, 0, sizeof(strm));
+    ret = deflateInit2(&strm, Z_DEFAULT_COMPRESSION,
+                       Z_DEFLATED, -12,
+                       9, Z_DEFAULT_STRATEGY);
+    if (ret != 0) {
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    strm.avail_in = s->cluster_size;
+    strm.next_in = (uint8_t *)buf;
+    strm.avail_out = s->cluster_size;
+    strm.next_out = out_buf;
+
+    ret = deflate(&strm, Z_FINISH);
+    if (ret != Z_STREAM_END && ret != Z_OK) {
+        deflateEnd(&strm);
+        ret = -EINVAL;
+        goto fail;
+    }
+    out_len = strm.next_out - out_buf;
+
+    deflateEnd(&strm);
+
+    if (ret != Z_STREAM_END || out_len >= s->cluster_size) {
+        /* could not compress: write normal cluster */
+        ret = bdrv_write(bs, sector_num, buf, s->cluster_sectors);
+        if (ret < 0) {
+            goto fail;
+        }
+    } else {
+        cluster_offset = get_cluster_offset(bs, sector_num << 9, 2,
+                                            out_len, 0, 0);
+        if (cluster_offset == 0) {
+            ret = -EIO;
+            goto fail;
+        }
+
+        cluster_offset &= s->cluster_offset_mask;
+        ret = bdrv_pwrite(bs->file, cluster_offset, out_buf, out_len);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    ret = 0;
+fail:
+    g_free(out_buf);
+    return ret;
+}
+
+static int qcow_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
+{
+    BDRVQcowState *s = bs->opaque;
+    bdi->cluster_size = s->cluster_size;
+    return 0;
+}
+
+
+static QEMUOptionParameter qcow_create_options[] = {
+    {
+        .name = BLOCK_OPT_SIZE,
+        .type = OPT_SIZE,
+        .help = "Virtual disk size"
+    },
+    {
+        .name = BLOCK_OPT_BACKING_FILE,
+        .type = OPT_STRING,
+        .help = "File name of a base image"
+    },
+    {
+        .name = BLOCK_OPT_ENCRYPT,
+        .type = OPT_FLAG,
+        .help = "Encrypt the image"
+    },
+    { NULL }
+};
+
+static BlockDriver bdrv_qcow = {
+    .format_name	= "qcow",
+    .instance_size	= sizeof(BDRVQcowState),
+    .bdrv_probe		= qcow_probe,
+    .bdrv_open		= qcow_open,
+    .bdrv_close		= qcow_close,
+    .bdrv_reopen_prepare = qcow_reopen_prepare,
+    .bdrv_create	= qcow_create,
+    .bdrv_has_zero_init     = bdrv_has_zero_init_1,
+
+    .bdrv_co_readv          = qcow_co_readv,
+    .bdrv_co_writev         = qcow_co_writev,
+    .bdrv_co_is_allocated   = qcow_co_is_allocated,
+
+    .bdrv_set_key           = qcow_set_key,
+    .bdrv_make_empty        = qcow_make_empty,
+    .bdrv_write_compressed  = qcow_write_compressed,
+    .bdrv_get_info          = qcow_get_info,
+
+    .create_options = qcow_create_options,
+};
+
+static void bdrv_qcow_init(void)
+{
+    bdrv_register(&bdrv_qcow);
+}
+
+block_init(bdrv_qcow_init);
diff --git a/contrib/qemu/block/qcow2-cache.c b/contrib/qemu/block/qcow2-cache.c
new file mode 100644
index 000000000..2f3114ecc
--- /dev/null
+++ b/contrib/qemu/block/qcow2-cache.c
@@ -0,0 +1,323 @@
+/*
+ * L2/refcount table cache for the QCOW2 format
+ *
+ * Copyright (c) 2010 Kevin Wolf <kwolf@redhat.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "block/block_int.h"
+#include "qemu-common.h"
+#include "qcow2.h"
+#include "trace.h"
+
+typedef struct Qcow2CachedTable {
+    void*   table;
+    int64_t offset;
+    bool    dirty;
+    int     cache_hits;
+    int     ref;
+} Qcow2CachedTable;
+
+struct Qcow2Cache {
+    Qcow2CachedTable*       entries;
+    struct Qcow2Cache*      depends;
+    int                     size;
+    bool                    depends_on_flush;
+};
+
+Qcow2Cache *qcow2_cache_create(BlockDriverState *bs, int num_tables)
+{
+    BDRVQcowState *s = bs->opaque;
+    Qcow2Cache *c;
+    int i;
+
+    c = g_malloc0(sizeof(*c));
+    c->size = num_tables;
+    c->entries = g_malloc0(sizeof(*c->entries) * num_tables);
+
+    for (i = 0; i < c->size; i++) {
+        c->entries[i].table = qemu_blockalign(bs, s->cluster_size);
+    }
+
+    return c;
+}
+
+int qcow2_cache_destroy(BlockDriverState* bs, Qcow2Cache *c)
+{
+    int i;
+
+    for (i = 0; i < c->size; i++) {
+        assert(c->entries[i].ref == 0);
+        qemu_vfree(c->entries[i].table);
+    }
+
+    g_free(c->entries);
+    g_free(c);
+
+    return 0;
+}
+
+static int qcow2_cache_flush_dependency(BlockDriverState *bs, Qcow2Cache *c)
+{
+    int ret;
+
+    ret = qcow2_cache_flush(bs, c->depends);
+    if (ret < 0) {
+        return ret;
+    }
+
+    c->depends = NULL;
+    c->depends_on_flush = false;
+
+    return 0;
+}
+
+static int qcow2_cache_entry_flush(BlockDriverState *bs, Qcow2Cache *c, int i)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret = 0;
+
+    if (!c->entries[i].dirty || !c->entries[i].offset) {
+        return 0;
+    }
+
+    trace_qcow2_cache_entry_flush(qemu_coroutine_self(),
+                                  c == s->l2_table_cache, i);
+
+    if (c->depends) {
+        ret = qcow2_cache_flush_dependency(bs, c);
+    } else if (c->depends_on_flush) {
+        ret = bdrv_flush(bs->file);
+        if (ret >= 0) {
+            c->depends_on_flush = false;
+        }
+    }
+
+    if (ret < 0) {
+        return ret;
+    }
+
+    if (c == s->refcount_block_cache) {
+        BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_UPDATE_PART);
+    } else if (c == s->l2_table_cache) {
+        BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE);
+    }
+
+    ret = bdrv_pwrite(bs->file, c->entries[i].offset, c->entries[i].table,
+        s->cluster_size);
+    if (ret < 0) {
+        return ret;
+    }
+
+    c->entries[i].dirty = false;
+
+    return 0;
+}
+
+int qcow2_cache_flush(BlockDriverState *bs, Qcow2Cache *c)
+{
+    BDRVQcowState *s = bs->opaque;
+    int result = 0;
+    int ret;
+    int i;
+
+    trace_qcow2_cache_flush(qemu_coroutine_self(), c == s->l2_table_cache);
+
+    for (i = 0; i < c->size; i++) {
+        ret = qcow2_cache_entry_flush(bs, c, i);
+        if (ret < 0 && result != -ENOSPC) {
+            result = ret;
+        }
+    }
+
+    if (result == 0) {
+        ret = bdrv_flush(bs->file);
+        if (ret < 0) {
+            result = ret;
+        }
+    }
+
+    return result;
+}
+
+int qcow2_cache_set_dependency(BlockDriverState *bs, Qcow2Cache *c,
+    Qcow2Cache *dependency)
+{
+    int ret;
+
+    if (dependency->depends) {
+        ret = qcow2_cache_flush_dependency(bs, dependency);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    if (c->depends && (c->depends != dependency)) {
+        ret = qcow2_cache_flush_dependency(bs, c);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    c->depends = dependency;
+    return 0;
+}
+
+void qcow2_cache_depends_on_flush(Qcow2Cache *c)
+{
+    c->depends_on_flush = true;
+}
+
+static int qcow2_cache_find_entry_to_replace(Qcow2Cache *c)
+{
+    int i;
+    int min_count = INT_MAX;
+    int min_index = -1;
+
+
+    for (i = 0; i < c->size; i++) {
+        if (c->entries[i].ref) {
+            continue;
+        }
+
+        if (c->entries[i].cache_hits < min_count) {
+            min_index = i;
+            min_count = c->entries[i].cache_hits;
+        }
+
+        /* Give newer hits priority */
+        /* TODO Check how to optimize the replacement strategy */
+        c->entries[i].cache_hits /= 2;
+    }
+
+    if (min_index == -1) {
+        /* This can't happen in current synchronous code, but leave the check
+         * here as a reminder for whoever starts using AIO with the cache */
+        abort();
+    }
+    return min_index;
+}
+
+static int qcow2_cache_do_get(BlockDriverState *bs, Qcow2Cache *c,
+    uint64_t offset, void **table, bool read_from_disk)
+{
+    BDRVQcowState *s = bs->opaque;
+    int i;
+    int ret;
+
+    trace_qcow2_cache_get(qemu_coroutine_self(), c == s->l2_table_cache,
+                          offset, read_from_disk);
+
+    /* Check if the table is already cached */
+    for (i = 0; i < c->size; i++) {
+        if (c->entries[i].offset == offset) {
+            goto found;
+        }
+    }
+
+    /* If not, write a table back and replace it */
+    i = qcow2_cache_find_entry_to_replace(c);
+    trace_qcow2_cache_get_replace_entry(qemu_coroutine_self(),
+                                        c == s->l2_table_cache, i);
+    if (i < 0) {
+        return i;
+    }
+
+    ret = qcow2_cache_entry_flush(bs, c, i);
+    if (ret < 0) {
+        return ret;
+    }
+
+    trace_qcow2_cache_get_read(qemu_coroutine_self(),
+                               c == s->l2_table_cache, i);
+    c->entries[i].offset = 0;
+    if (read_from_disk) {
+        if (c == s->l2_table_cache) {
+            BLKDBG_EVENT(bs->file, BLKDBG_L2_LOAD);
+        }
+
+        ret = bdrv_pread(bs->file, offset, c->entries[i].table, s->cluster_size);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    /* Give the table some hits for the start so that it won't be replaced
+     * immediately. The number 32 is completely arbitrary. */
+    c->entries[i].cache_hits = 32;
+    c->entries[i].offset = offset;
+
+    /* And return the right table */
+found:
+    c->entries[i].cache_hits++;
+    c->entries[i].ref++;
+    *table = c->entries[i].table;
+
+    trace_qcow2_cache_get_done(qemu_coroutine_self(),
+                               c == s->l2_table_cache, i);
+
+    return 0;
+}
+
+int qcow2_cache_get(BlockDriverState *bs, Qcow2Cache *c, uint64_t offset,
+    void **table)
+{
+    return qcow2_cache_do_get(bs, c, offset, table, true);
+}
+
+int qcow2_cache_get_empty(BlockDriverState *bs, Qcow2Cache *c, uint64_t offset,
+    void **table)
+{
+    return qcow2_cache_do_get(bs, c, offset, table, false);
+}
+
+int qcow2_cache_put(BlockDriverState *bs, Qcow2Cache *c, void **table)
+{
+    int i;
+
+    for (i = 0; i < c->size; i++) {
+        if (c->entries[i].table == *table) {
+            goto found;
+        }
+    }
+    return -ENOENT;
+
+found:
+    c->entries[i].ref--;
+    *table = NULL;
+
+    assert(c->entries[i].ref >= 0);
+    return 0;
+}
+
+void qcow2_cache_entry_mark_dirty(Qcow2Cache *c, void *table)
+{
+    int i;
+
+    for (i = 0; i < c->size; i++) {
+        if (c->entries[i].table == table) {
+            goto found;
+        }
+    }
+    abort();
+
+found:
+    c->entries[i].dirty = true;
+}
diff --git a/contrib/qemu/block/qcow2-cluster.c b/contrib/qemu/block/qcow2-cluster.c
new file mode 100644
index 000000000..cca76d4fc
--- /dev/null
+++ b/contrib/qemu/block/qcow2-cluster.c
@@ -0,0 +1,1478 @@
+/*
+ * Block driver for the QCOW version 2 format
+ *
+ * Copyright (c) 2004-2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include <zlib.h>
+
+#include "qemu-common.h"
+#include "block/block_int.h"
+#include "block/qcow2.h"
+#include "trace.h"
+
+int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
+                        bool exact_size)
+{
+    BDRVQcowState *s = bs->opaque;
+    int new_l1_size2, ret, i;
+    uint64_t *new_l1_table;
+    int64_t new_l1_table_offset, new_l1_size;
+    uint8_t data[12];
+
+    if (min_size <= s->l1_size)
+        return 0;
+
+    if (exact_size) {
+        new_l1_size = min_size;
+    } else {
+        /* Bump size up to reduce the number of times we have to grow */
+        new_l1_size = s->l1_size;
+        if (new_l1_size == 0) {
+            new_l1_size = 1;
+        }
+        while (min_size > new_l1_size) {
+            new_l1_size = (new_l1_size * 3 + 1) / 2;
+        }
+    }
+
+    if (new_l1_size > INT_MAX) {
+        return -EFBIG;
+    }
+
+#ifdef DEBUG_ALLOC2
+    fprintf(stderr, "grow l1_table from %d to %" PRId64 "\n",
+            s->l1_size, new_l1_size);
+#endif
+
+    new_l1_size2 = sizeof(uint64_t) * new_l1_size;
+    new_l1_table = g_malloc0(align_offset(new_l1_size2, 512));
+    memcpy(new_l1_table, s->l1_table, s->l1_size * sizeof(uint64_t));
+
+    /* write new table (align to cluster) */
+    BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ALLOC_TABLE);
+    new_l1_table_offset = qcow2_alloc_clusters(bs, new_l1_size2);
+    if (new_l1_table_offset < 0) {
+        g_free(new_l1_table);
+        return new_l1_table_offset;
+    }
+
+    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_WRITE_TABLE);
+    for(i = 0; i < s->l1_size; i++)
+        new_l1_table[i] = cpu_to_be64(new_l1_table[i]);
+    ret = bdrv_pwrite_sync(bs->file, new_l1_table_offset, new_l1_table, new_l1_size2);
+    if (ret < 0)
+        goto fail;
+    for(i = 0; i < s->l1_size; i++)
+        new_l1_table[i] = be64_to_cpu(new_l1_table[i]);
+
+    /* set new table */
+    BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ACTIVATE_TABLE);
+    cpu_to_be32w((uint32_t*)data, new_l1_size);
+    cpu_to_be64wu((uint64_t*)(data + 4), new_l1_table_offset);
+    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, l1_size), data,sizeof(data));
+    if (ret < 0) {
+        goto fail;
+    }
+    g_free(s->l1_table);
+    qcow2_free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t),
+                        QCOW2_DISCARD_OTHER);
+    s->l1_table_offset = new_l1_table_offset;
+    s->l1_table = new_l1_table;
+    s->l1_size = new_l1_size;
+    return 0;
+ fail:
+    g_free(new_l1_table);
+    qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2,
+                        QCOW2_DISCARD_OTHER);
+    return ret;
+}
+
+/*
+ * l2_load
+ *
+ * Loads a L2 table into memory. If the table is in the cache, the cache
+ * is used; otherwise the L2 table is loaded from the image file.
+ *
+ * Returns a pointer to the L2 table on success, or NULL if the read from
+ * the image file failed.
+ */
+
+static int l2_load(BlockDriverState *bs, uint64_t l2_offset,
+    uint64_t **l2_table)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret;
+
+    ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset, (void**) l2_table);
+
+    return ret;
+}
+
+/*
+ * Writes one sector of the L1 table to the disk (can't update single entries
+ * and we really don't want bdrv_pread to perform a read-modify-write)
+ */
+#define L1_ENTRIES_PER_SECTOR (512 / 8)
+static int write_l1_entry(BlockDriverState *bs, int l1_index)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t buf[L1_ENTRIES_PER_SECTOR];
+    int l1_start_index;
+    int i, ret;
+
+    l1_start_index = l1_index & ~(L1_ENTRIES_PER_SECTOR - 1);
+    for (i = 0; i < L1_ENTRIES_PER_SECTOR; i++) {
+        buf[i] = cpu_to_be64(s->l1_table[l1_start_index + i]);
+    }
+
+    BLKDBG_EVENT(bs->file, BLKDBG_L1_UPDATE);
+    ret = bdrv_pwrite_sync(bs->file, s->l1_table_offset + 8 * l1_start_index,
+        buf, sizeof(buf));
+    if (ret < 0) {
+        return ret;
+    }
+
+    return 0;
+}
+
+/*
+ * l2_allocate
+ *
+ * Allocate a new l2 entry in the file. If l1_index points to an already
+ * used entry in the L2 table (i.e. we are doing a copy on write for the L2
+ * table) copy the contents of the old L2 table into the newly allocated one.
+ * Otherwise the new table is initialized with zeros.
+ *
+ */
+
+static int l2_allocate(BlockDriverState *bs, int l1_index, uint64_t **table)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t old_l2_offset;
+    uint64_t *l2_table;
+    int64_t l2_offset;
+    int ret;
+
+    old_l2_offset = s->l1_table[l1_index];
+
+    trace_qcow2_l2_allocate(bs, l1_index);
+
+    /* allocate a new l2 entry */
+
+    l2_offset = qcow2_alloc_clusters(bs, s->l2_size * sizeof(uint64_t));
+    if (l2_offset < 0) {
+        return l2_offset;
+    }
+
+    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* allocate a new entry in the l2 cache */
+
+    trace_qcow2_l2_allocate_get_empty(bs, l1_index);
+    ret = qcow2_cache_get_empty(bs, s->l2_table_cache, l2_offset, (void**) table);
+    if (ret < 0) {
+        return ret;
+    }
+
+    l2_table = *table;
+
+    if ((old_l2_offset & L1E_OFFSET_MASK) == 0) {
+        /* if there was no old l2 table, clear the new table */
+        memset(l2_table, 0, s->l2_size * sizeof(uint64_t));
+    } else {
+        uint64_t* old_table;
+
+        /* if there was an old l2 table, read it from the disk */
+        BLKDBG_EVENT(bs->file, BLKDBG_L2_ALLOC_COW_READ);
+        ret = qcow2_cache_get(bs, s->l2_table_cache,
+            old_l2_offset & L1E_OFFSET_MASK,
+            (void**) &old_table);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        memcpy(l2_table, old_table, s->cluster_size);
+
+        ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &old_table);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    /* write the l2 table to the file */
+    BLKDBG_EVENT(bs->file, BLKDBG_L2_ALLOC_WRITE);
+
+    trace_qcow2_l2_allocate_write_l2(bs, l1_index);
+    qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+    ret = qcow2_cache_flush(bs, s->l2_table_cache);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* update the L1 entry */
+    trace_qcow2_l2_allocate_write_l1(bs, l1_index);
+    s->l1_table[l1_index] = l2_offset | QCOW_OFLAG_COPIED;
+    ret = write_l1_entry(bs, l1_index);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    *table = l2_table;
+    trace_qcow2_l2_allocate_done(bs, l1_index, 0);
+    return 0;
+
+fail:
+    trace_qcow2_l2_allocate_done(bs, l1_index, ret);
+    qcow2_cache_put(bs, s->l2_table_cache, (void**) table);
+    s->l1_table[l1_index] = old_l2_offset;
+    return ret;
+}
+
+/*
+ * Checks how many clusters in a given L2 table are contiguous in the image
+ * file. As soon as one of the flags in the bitmask stop_flags changes compared
+ * to the first cluster, the search is stopped and the cluster is not counted
+ * as contiguous. (This allows it, for example, to stop at the first compressed
+ * cluster which may require a different handling)
+ */
+static int count_contiguous_clusters(uint64_t nb_clusters, int cluster_size,
+        uint64_t *l2_table, uint64_t start, uint64_t stop_flags)
+{
+    int i;
+    uint64_t mask = stop_flags | L2E_OFFSET_MASK;
+    uint64_t offset = be64_to_cpu(l2_table[0]) & mask;
+
+    if (!offset)
+        return 0;
+
+    for (i = start; i < start + nb_clusters; i++) {
+        uint64_t l2_entry = be64_to_cpu(l2_table[i]) & mask;
+        if (offset + (uint64_t) i * cluster_size != l2_entry) {
+            break;
+        }
+    }
+
+	return (i - start);
+}
+
+static int count_contiguous_free_clusters(uint64_t nb_clusters, uint64_t *l2_table)
+{
+    int i;
+
+    for (i = 0; i < nb_clusters; i++) {
+        int type = qcow2_get_cluster_type(be64_to_cpu(l2_table[i]));
+
+        if (type != QCOW2_CLUSTER_UNALLOCATED) {
+            break;
+        }
+    }
+
+    return i;
+}
+
+/* The crypt function is compatible with the linux cryptoloop
+   algorithm for < 4 GB images. NOTE: out_buf == in_buf is
+   supported */
+void qcow2_encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
+                           uint8_t *out_buf, const uint8_t *in_buf,
+                           int nb_sectors, int enc,
+                           const AES_KEY *key)
+{
+    union {
+        uint64_t ll[2];
+        uint8_t b[16];
+    } ivec;
+    int i;
+
+    for(i = 0; i < nb_sectors; i++) {
+        ivec.ll[0] = cpu_to_le64(sector_num);
+        ivec.ll[1] = 0;
+        AES_cbc_encrypt(in_buf, out_buf, 512, key,
+                        ivec.b, enc);
+        sector_num++;
+        in_buf += 512;
+        out_buf += 512;
+    }
+}
+
+static int coroutine_fn copy_sectors(BlockDriverState *bs,
+                                     uint64_t start_sect,
+                                     uint64_t cluster_offset,
+                                     int n_start, int n_end)
+{
+    BDRVQcowState *s = bs->opaque;
+    QEMUIOVector qiov;
+    struct iovec iov;
+    int n, ret;
+
+    /*
+     * If this is the last cluster and it is only partially used, we must only
+     * copy until the end of the image, or bdrv_check_request will fail for the
+     * bdrv_read/write calls below.
+     */
+    if (start_sect + n_end > bs->total_sectors) {
+        n_end = bs->total_sectors - start_sect;
+    }
+
+    n = n_end - n_start;
+    if (n <= 0) {
+        return 0;
+    }
+
+    iov.iov_len = n * BDRV_SECTOR_SIZE;
+    iov.iov_base = qemu_blockalign(bs, iov.iov_len);
+
+    qemu_iovec_init_external(&qiov, &iov, 1);
+
+    BLKDBG_EVENT(bs->file, BLKDBG_COW_READ);
+
+    /* Call .bdrv_co_readv() directly instead of using the public block-layer
+     * interface.  This avoids double I/O throttling and request tracking,
+     * which can lead to deadlock when block layer copy-on-read is enabled.
+     */
+    ret = bs->drv->bdrv_co_readv(bs, start_sect + n_start, n, &qiov);
+    if (ret < 0) {
+        goto out;
+    }
+
+    if (s->crypt_method) {
+        qcow2_encrypt_sectors(s, start_sect + n_start,
+                        iov.iov_base, iov.iov_base, n, 1,
+                        &s->aes_encrypt_key);
+    }
+
+    BLKDBG_EVENT(bs->file, BLKDBG_COW_WRITE);
+    ret = bdrv_co_writev(bs->file, (cluster_offset >> 9) + n_start, n, &qiov);
+    if (ret < 0) {
+        goto out;
+    }
+
+    ret = 0;
+out:
+    qemu_vfree(iov.iov_base);
+    return ret;
+}
+
+
+/*
+ * get_cluster_offset
+ *
+ * For a given offset of the disk image, find the cluster offset in
+ * qcow2 file. The offset is stored in *cluster_offset.
+ *
+ * on entry, *num is the number of contiguous sectors we'd like to
+ * access following offset.
+ *
+ * on exit, *num is the number of contiguous sectors we can read.
+ *
+ * Returns the cluster type (QCOW2_CLUSTER_*) on success, -errno in error
+ * cases.
+ */
+int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
+    int *num, uint64_t *cluster_offset)
+{
+    BDRVQcowState *s = bs->opaque;
+    unsigned int l2_index;
+    uint64_t l1_index, l2_offset, *l2_table;
+    int l1_bits, c;
+    unsigned int index_in_cluster, nb_clusters;
+    uint64_t nb_available, nb_needed;
+    int ret;
+
+    index_in_cluster = (offset >> 9) & (s->cluster_sectors - 1);
+    nb_needed = *num + index_in_cluster;
+
+    l1_bits = s->l2_bits + s->cluster_bits;
+
+    /* compute how many bytes there are between the offset and
+     * the end of the l1 entry
+     */
+
+    nb_available = (1ULL << l1_bits) - (offset & ((1ULL << l1_bits) - 1));
+
+    /* compute the number of available sectors */
+
+    nb_available = (nb_available >> 9) + index_in_cluster;
+
+    if (nb_needed > nb_available) {
+        nb_needed = nb_available;
+    }
+
+    *cluster_offset = 0;
+
+    /* seek the the l2 offset in the l1 table */
+
+    l1_index = offset >> l1_bits;
+    if (l1_index >= s->l1_size) {
+        ret = QCOW2_CLUSTER_UNALLOCATED;
+        goto out;
+    }
+
+    l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
+    if (!l2_offset) {
+        ret = QCOW2_CLUSTER_UNALLOCATED;
+        goto out;
+    }
+
+    /* load the l2 table in memory */
+
+    ret = l2_load(bs, l2_offset, &l2_table);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* find the cluster offset for the given disk offset */
+
+    l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
+    *cluster_offset = be64_to_cpu(l2_table[l2_index]);
+    nb_clusters = size_to_clusters(s, nb_needed << 9);
+
+    ret = qcow2_get_cluster_type(*cluster_offset);
+    switch (ret) {
+    case QCOW2_CLUSTER_COMPRESSED:
+        /* Compressed clusters can only be processed one by one */
+        c = 1;
+        *cluster_offset &= L2E_COMPRESSED_OFFSET_SIZE_MASK;
+        break;
+    case QCOW2_CLUSTER_ZERO:
+        if (s->qcow_version < 3) {
+            return -EIO;
+        }
+        c = count_contiguous_clusters(nb_clusters, s->cluster_size,
+                &l2_table[l2_index], 0,
+                QCOW_OFLAG_COMPRESSED | QCOW_OFLAG_ZERO);
+        *cluster_offset = 0;
+        break;
+    case QCOW2_CLUSTER_UNALLOCATED:
+        /* how many empty clusters ? */
+        c = count_contiguous_free_clusters(nb_clusters, &l2_table[l2_index]);
+        *cluster_offset = 0;
+        break;
+    case QCOW2_CLUSTER_NORMAL:
+        /* how many allocated clusters ? */
+        c = count_contiguous_clusters(nb_clusters, s->cluster_size,
+                &l2_table[l2_index], 0,
+                QCOW_OFLAG_COMPRESSED | QCOW_OFLAG_ZERO);
+        *cluster_offset &= L2E_OFFSET_MASK;
+        break;
+    default:
+        abort();
+    }
+
+    qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+
+    nb_available = (c * s->cluster_sectors);
+
+out:
+    if (nb_available > nb_needed)
+        nb_available = nb_needed;
+
+    *num = nb_available - index_in_cluster;
+
+    return ret;
+}
+
+/*
+ * get_cluster_table
+ *
+ * for a given disk offset, load (and allocate if needed)
+ * the l2 table.
+ *
+ * the l2 table offset in the qcow2 file and the cluster index
+ * in the l2 table are given to the caller.
+ *
+ * Returns 0 on success, -errno in failure case
+ */
+static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
+                             uint64_t **new_l2_table,
+                             int *new_l2_index)
+{
+    BDRVQcowState *s = bs->opaque;
+    unsigned int l2_index;
+    uint64_t l1_index, l2_offset;
+    uint64_t *l2_table = NULL;
+    int ret;
+
+    /* seek the the l2 offset in the l1 table */
+
+    l1_index = offset >> (s->l2_bits + s->cluster_bits);
+    if (l1_index >= s->l1_size) {
+        ret = qcow2_grow_l1_table(bs, l1_index + 1, false);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    assert(l1_index < s->l1_size);
+    l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
+
+    /* seek the l2 table of the given l2 offset */
+
+    if (s->l1_table[l1_index] & QCOW_OFLAG_COPIED) {
+        /* load the l2 table in memory */
+        ret = l2_load(bs, l2_offset, &l2_table);
+        if (ret < 0) {
+            return ret;
+        }
+    } else {
+        /* First allocate a new L2 table (and do COW if needed) */
+        ret = l2_allocate(bs, l1_index, &l2_table);
+        if (ret < 0) {
+            return ret;
+        }
+
+        /* Then decrease the refcount of the old table */
+        if (l2_offset) {
+            qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
+                                QCOW2_DISCARD_OTHER);
+        }
+    }
+
+    /* find the cluster offset for the given disk offset */
+
+    l2_index = (offset >> s->cluster_bits) & (s->l2_size - 1);
+
+    *new_l2_table = l2_table;
+    *new_l2_index = l2_index;
+
+    return 0;
+}
+
+/*
+ * alloc_compressed_cluster_offset
+ *
+ * For a given offset of the disk image, return cluster offset in
+ * qcow2 file.
+ *
+ * If the offset is not found, allocate a new compressed cluster.
+ *
+ * Return the cluster offset if successful,
+ * Return 0, otherwise.
+ *
+ */
+
+uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
+                                               uint64_t offset,
+                                               int compressed_size)
+{
+    BDRVQcowState *s = bs->opaque;
+    int l2_index, ret;
+    uint64_t *l2_table;
+    int64_t cluster_offset;
+    int nb_csectors;
+
+    ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
+    if (ret < 0) {
+        return 0;
+    }
+
+    /* Compression can't overwrite anything. Fail if the cluster was already
+     * allocated. */
+    cluster_offset = be64_to_cpu(l2_table[l2_index]);
+    if (cluster_offset & L2E_OFFSET_MASK) {
+        qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+        return 0;
+    }
+
+    cluster_offset = qcow2_alloc_bytes(bs, compressed_size);
+    if (cluster_offset < 0) {
+        qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+        return 0;
+    }
+
+    nb_csectors = ((cluster_offset + compressed_size - 1) >> 9) -
+                  (cluster_offset >> 9);
+
+    cluster_offset |= QCOW_OFLAG_COMPRESSED |
+                      ((uint64_t)nb_csectors << s->csize_shift);
+
+    /* update L2 table */
+
+    /* compressed clusters never have the copied flag */
+
+    BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE_COMPRESSED);
+    qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+    l2_table[l2_index] = cpu_to_be64(cluster_offset);
+    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    if (ret < 0) {
+        return 0;
+    }
+
+    return cluster_offset;
+}
+
+static int perform_cow(BlockDriverState *bs, QCowL2Meta *m, Qcow2COWRegion *r)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret;
+
+    if (r->nb_sectors == 0) {
+        return 0;
+    }
+
+    qemu_co_mutex_unlock(&s->lock);
+    ret = copy_sectors(bs, m->offset / BDRV_SECTOR_SIZE, m->alloc_offset,
+                       r->offset / BDRV_SECTOR_SIZE,
+                       r->offset / BDRV_SECTOR_SIZE + r->nb_sectors);
+    qemu_co_mutex_lock(&s->lock);
+
+    if (ret < 0) {
+        return ret;
+    }
+
+    /*
+     * Before we update the L2 table to actually point to the new cluster, we
+     * need to be sure that the refcounts have been increased and COW was
+     * handled.
+     */
+    qcow2_cache_depends_on_flush(s->l2_table_cache);
+
+    return 0;
+}
+
+int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
+{
+    BDRVQcowState *s = bs->opaque;
+    int i, j = 0, l2_index, ret;
+    uint64_t *old_cluster, *l2_table;
+    uint64_t cluster_offset = m->alloc_offset;
+
+    trace_qcow2_cluster_link_l2(qemu_coroutine_self(), m->nb_clusters);
+    assert(m->nb_clusters > 0);
+
+    old_cluster = g_malloc(m->nb_clusters * sizeof(uint64_t));
+
+    /* copy content of unmodified sectors */
+    ret = perform_cow(bs, m, &m->cow_start);
+    if (ret < 0) {
+        goto err;
+    }
+
+    ret = perform_cow(bs, m, &m->cow_end);
+    if (ret < 0) {
+        goto err;
+    }
+
+    /* Update L2 table. */
+    if (s->use_lazy_refcounts) {
+        qcow2_mark_dirty(bs);
+    }
+    if (qcow2_need_accurate_refcounts(s)) {
+        qcow2_cache_set_dependency(bs, s->l2_table_cache,
+                                   s->refcount_block_cache);
+    }
+
+    ret = get_cluster_table(bs, m->offset, &l2_table, &l2_index);
+    if (ret < 0) {
+        goto err;
+    }
+    qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+
+    for (i = 0; i < m->nb_clusters; i++) {
+        /* if two concurrent writes happen to the same unallocated cluster
+	 * each write allocates separate cluster and writes data concurrently.
+	 * The first one to complete updates l2 table with pointer to its
+	 * cluster the second one has to do RMW (which is done above by
+	 * copy_sectors()), update l2 table with its cluster pointer and free
+	 * old cluster. This is what this loop does */
+        if(l2_table[l2_index + i] != 0)
+            old_cluster[j++] = l2_table[l2_index + i];
+
+        l2_table[l2_index + i] = cpu_to_be64((cluster_offset +
+                    (i << s->cluster_bits)) | QCOW_OFLAG_COPIED);
+     }
+
+
+    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    if (ret < 0) {
+        goto err;
+    }
+
+    /*
+     * If this was a COW, we need to decrease the refcount of the old cluster.
+     * Also flush bs->file to get the right order for L2 and refcount update.
+     *
+     * Don't discard clusters that reach a refcount of 0 (e.g. compressed
+     * clusters), the next write will reuse them anyway.
+     */
+    if (j != 0) {
+        for (i = 0; i < j; i++) {
+            qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1,
+                                    QCOW2_DISCARD_NEVER);
+        }
+    }
+
+    ret = 0;
+err:
+    g_free(old_cluster);
+    return ret;
+ }
+
+/*
+ * Returns the number of contiguous clusters that can be used for an allocating
+ * write, but require COW to be performed (this includes yet unallocated space,
+ * which must copy from the backing file)
+ */
+static int count_cow_clusters(BDRVQcowState *s, int nb_clusters,
+    uint64_t *l2_table, int l2_index)
+{
+    int i;
+
+    for (i = 0; i < nb_clusters; i++) {
+        uint64_t l2_entry = be64_to_cpu(l2_table[l2_index + i]);
+        int cluster_type = qcow2_get_cluster_type(l2_entry);
+
+        switch(cluster_type) {
+        case QCOW2_CLUSTER_NORMAL:
+            if (l2_entry & QCOW_OFLAG_COPIED) {
+                goto out;
+            }
+            break;
+        case QCOW2_CLUSTER_UNALLOCATED:
+        case QCOW2_CLUSTER_COMPRESSED:
+        case QCOW2_CLUSTER_ZERO:
+            break;
+        default:
+            abort();
+        }
+    }
+
+out:
+    assert(i <= nb_clusters);
+    return i;
+}
+
+/*
+ * Check if there already is an AIO write request in flight which allocates
+ * the same cluster. In this case we need to wait until the previous
+ * request has completed and updated the L2 table accordingly.
+ *
+ * Returns:
+ *   0       if there was no dependency. *cur_bytes indicates the number of
+ *           bytes from guest_offset that can be read before the next
+ *           dependency must be processed (or the request is complete)
+ *
+ *   -EAGAIN if we had to wait for another request, previously gathered
+ *           information on cluster allocation may be invalid now. The caller
+ *           must start over anyway, so consider *cur_bytes undefined.
+ */
+static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
+    uint64_t *cur_bytes, QCowL2Meta **m)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowL2Meta *old_alloc;
+    uint64_t bytes = *cur_bytes;
+
+    QLIST_FOREACH(old_alloc, &s->cluster_allocs, next_in_flight) {
+
+        uint64_t start = guest_offset;
+        uint64_t end = start + bytes;
+        uint64_t old_start = l2meta_cow_start(old_alloc);
+        uint64_t old_end = l2meta_cow_end(old_alloc);
+
+        if (end <= old_start || start >= old_end) {
+            /* No intersection */
+        } else {
+            if (start < old_start) {
+                /* Stop at the start of a running allocation */
+                bytes = old_start - start;
+            } else {
+                bytes = 0;
+            }
+
+            /* Stop if already an l2meta exists. After yielding, it wouldn't
+             * be valid any more, so we'd have to clean up the old L2Metas
+             * and deal with requests depending on them before starting to
+             * gather new ones. Not worth the trouble. */
+            if (bytes == 0 && *m) {
+                *cur_bytes = 0;
+                return 0;
+            }
+
+            if (bytes == 0) {
+                /* Wait for the dependency to complete. We need to recheck
+                 * the free/allocated clusters when we continue. */
+                qemu_co_mutex_unlock(&s->lock);
+                qemu_co_queue_wait(&old_alloc->dependent_requests);
+                qemu_co_mutex_lock(&s->lock);
+                return -EAGAIN;
+            }
+        }
+    }
+
+    /* Make sure that existing clusters and new allocations are only used up to
+     * the next dependency if we shortened the request above */
+    *cur_bytes = bytes;
+
+    return 0;
+}
+
+/*
+ * Checks how many already allocated clusters that don't require a copy on
+ * write there are at the given guest_offset (up to *bytes). If
+ * *host_offset is not zero, only physically contiguous clusters beginning at
+ * this host offset are counted.
+ *
+ * Note that guest_offset may not be cluster aligned. In this case, the
+ * returned *host_offset points to exact byte referenced by guest_offset and
+ * therefore isn't cluster aligned as well.
+ *
+ * Returns:
+ *   0:     if no allocated clusters are available at the given offset.
+ *          *bytes is normally unchanged. It is set to 0 if the cluster
+ *          is allocated and doesn't need COW, but doesn't have the right
+ *          physical offset.
+ *
+ *   1:     if allocated clusters that don't require a COW are available at
+ *          the requested offset. *bytes may have decreased and describes
+ *          the length of the area that can be written to.
+ *
+ *  -errno: in error cases
+ */
+static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
+    uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
+{
+    BDRVQcowState *s = bs->opaque;
+    int l2_index;
+    uint64_t cluster_offset;
+    uint64_t *l2_table;
+    unsigned int nb_clusters;
+    unsigned int keep_clusters;
+    int ret, pret;
+
+    trace_qcow2_handle_copied(qemu_coroutine_self(), guest_offset, *host_offset,
+                              *bytes);
+
+    assert(*host_offset == 0 ||    offset_into_cluster(s, guest_offset)
+                                == offset_into_cluster(s, *host_offset));
+
+    /*
+     * Calculate the number of clusters to look for. We stop at L2 table
+     * boundaries to keep things simple.
+     */
+    nb_clusters =
+        size_to_clusters(s, offset_into_cluster(s, guest_offset) + *bytes);
+
+    l2_index = offset_to_l2_index(s, guest_offset);
+    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
+
+    /* Find L2 entry for the first involved cluster */
+    ret = get_cluster_table(bs, guest_offset, &l2_table, &l2_index);
+    if (ret < 0) {
+        return ret;
+    }
+
+    cluster_offset = be64_to_cpu(l2_table[l2_index]);
+
+    /* Check how many clusters are already allocated and don't need COW */
+    if (qcow2_get_cluster_type(cluster_offset) == QCOW2_CLUSTER_NORMAL
+        && (cluster_offset & QCOW_OFLAG_COPIED))
+    {
+        /* If a specific host_offset is required, check it */
+        bool offset_matches =
+            (cluster_offset & L2E_OFFSET_MASK) == *host_offset;
+
+        if (*host_offset != 0 && !offset_matches) {
+            *bytes = 0;
+            ret = 0;
+            goto out;
+        }
+
+        /* We keep all QCOW_OFLAG_COPIED clusters */
+        keep_clusters =
+            count_contiguous_clusters(nb_clusters, s->cluster_size,
+                                      &l2_table[l2_index], 0,
+                                      QCOW_OFLAG_COPIED | QCOW_OFLAG_ZERO);
+        assert(keep_clusters <= nb_clusters);
+
+        *bytes = MIN(*bytes,
+                 keep_clusters * s->cluster_size
+                 - offset_into_cluster(s, guest_offset));
+
+        ret = 1;
+    } else {
+        ret = 0;
+    }
+
+    /* Cleanup */
+out:
+    pret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    if (pret < 0) {
+        return pret;
+    }
+
+    /* Only return a host offset if we actually made progress. Otherwise we
+     * would make requirements for handle_alloc() that it can't fulfill */
+    if (ret) {
+        *host_offset = (cluster_offset & L2E_OFFSET_MASK)
+                     + offset_into_cluster(s, guest_offset);
+    }
+
+    return ret;
+}
+
+/*
+ * Allocates new clusters for the given guest_offset.
+ *
+ * At most *nb_clusters are allocated, and on return *nb_clusters is updated to
+ * contain the number of clusters that have been allocated and are contiguous
+ * in the image file.
+ *
+ * If *host_offset is non-zero, it specifies the offset in the image file at
+ * which the new clusters must start. *nb_clusters can be 0 on return in this
+ * case if the cluster at host_offset is already in use. If *host_offset is
+ * zero, the clusters can be allocated anywhere in the image file.
+ *
+ * *host_offset is updated to contain the offset into the image file at which
+ * the first allocated cluster starts.
+ *
+ * Return 0 on success and -errno in error cases. -EAGAIN means that the
+ * function has been waiting for another request and the allocation must be
+ * restarted, but the whole request should not be failed.
+ */
+static int do_alloc_cluster_offset(BlockDriverState *bs, uint64_t guest_offset,
+    uint64_t *host_offset, unsigned int *nb_clusters)
+{
+    BDRVQcowState *s = bs->opaque;
+
+    trace_qcow2_do_alloc_clusters_offset(qemu_coroutine_self(), guest_offset,
+                                         *host_offset, *nb_clusters);
+
+    /* Allocate new clusters */
+    trace_qcow2_cluster_alloc_phys(qemu_coroutine_self());
+    if (*host_offset == 0) {
+        int64_t cluster_offset =
+            qcow2_alloc_clusters(bs, *nb_clusters * s->cluster_size);
+        if (cluster_offset < 0) {
+            return cluster_offset;
+        }
+        *host_offset = cluster_offset;
+        return 0;
+    } else {
+        int ret = qcow2_alloc_clusters_at(bs, *host_offset, *nb_clusters);
+        if (ret < 0) {
+            return ret;
+        }
+        *nb_clusters = ret;
+        return 0;
+    }
+}
+
+/*
+ * Allocates new clusters for an area that either is yet unallocated or needs a
+ * copy on write. If *host_offset is non-zero, clusters are only allocated if
+ * the new allocation can match the specified host offset.
+ *
+ * Note that guest_offset may not be cluster aligned. In this case, the
+ * returned *host_offset points to exact byte referenced by guest_offset and
+ * therefore isn't cluster aligned as well.
+ *
+ * Returns:
+ *   0:     if no clusters could be allocated. *bytes is set to 0,
+ *          *host_offset is left unchanged.
+ *
+ *   1:     if new clusters were allocated. *bytes may be decreased if the
+ *          new allocation doesn't cover all of the requested area.
+ *          *host_offset is updated to contain the host offset of the first
+ *          newly allocated cluster.
+ *
+ *  -errno: in error cases
+ */
+static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
+    uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
+{
+    BDRVQcowState *s = bs->opaque;
+    int l2_index;
+    uint64_t *l2_table;
+    uint64_t entry;
+    unsigned int nb_clusters;
+    int ret;
+
+    uint64_t alloc_cluster_offset;
+
+    trace_qcow2_handle_alloc(qemu_coroutine_self(), guest_offset, *host_offset,
+                             *bytes);
+    assert(*bytes > 0);
+
+    /*
+     * Calculate the number of clusters to look for. We stop at L2 table
+     * boundaries to keep things simple.
+     */
+    nb_clusters =
+        size_to_clusters(s, offset_into_cluster(s, guest_offset) + *bytes);
+
+    l2_index = offset_to_l2_index(s, guest_offset);
+    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
+
+    /* Find L2 entry for the first involved cluster */
+    ret = get_cluster_table(bs, guest_offset, &l2_table, &l2_index);
+    if (ret < 0) {
+        return ret;
+    }
+
+    entry = be64_to_cpu(l2_table[l2_index]);
+
+    /* For the moment, overwrite compressed clusters one by one */
+    if (entry & QCOW_OFLAG_COMPRESSED) {
+        nb_clusters = 1;
+    } else {
+        nb_clusters = count_cow_clusters(s, nb_clusters, l2_table, l2_index);
+    }
+
+    /* This function is only called when there were no non-COW clusters, so if
+     * we can't find any unallocated or COW clusters either, something is
+     * wrong with our code. */
+    assert(nb_clusters > 0);
+
+    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Allocate, if necessary at a given offset in the image file */
+    alloc_cluster_offset = start_of_cluster(s, *host_offset);
+    ret = do_alloc_cluster_offset(bs, guest_offset, &alloc_cluster_offset,
+                                  &nb_clusters);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* Can't extend contiguous allocation */
+    if (nb_clusters == 0) {
+        *bytes = 0;
+        return 0;
+    }
+
+    /*
+     * Save info needed for meta data update.
+     *
+     * requested_sectors: Number of sectors from the start of the first
+     * newly allocated cluster to the end of the (possibly shortened
+     * before) write request.
+     *
+     * avail_sectors: Number of sectors from the start of the first
+     * newly allocated to the end of the last newly allocated cluster.
+     *
+     * nb_sectors: The number of sectors from the start of the first
+     * newly allocated cluster to the end of the area that the write
+     * request actually writes to (excluding COW at the end)
+     */
+    int requested_sectors =
+        (*bytes + offset_into_cluster(s, guest_offset))
+        >> BDRV_SECTOR_BITS;
+    int avail_sectors = nb_clusters
+                        << (s->cluster_bits - BDRV_SECTOR_BITS);
+    int alloc_n_start = offset_into_cluster(s, guest_offset)
+                        >> BDRV_SECTOR_BITS;
+    int nb_sectors = MIN(requested_sectors, avail_sectors);
+    QCowL2Meta *old_m = *m;
+
+    *m = g_malloc0(sizeof(**m));
+
+    **m = (QCowL2Meta) {
+        .next           = old_m,
+
+        .alloc_offset   = alloc_cluster_offset,
+        .offset         = start_of_cluster(s, guest_offset),
+        .nb_clusters    = nb_clusters,
+        .nb_available   = nb_sectors,
+
+        .cow_start = {
+            .offset     = 0,
+            .nb_sectors = alloc_n_start,
+        },
+        .cow_end = {
+            .offset     = nb_sectors * BDRV_SECTOR_SIZE,
+            .nb_sectors = avail_sectors - nb_sectors,
+        },
+    };
+    qemu_co_queue_init(&(*m)->dependent_requests);
+    QLIST_INSERT_HEAD(&s->cluster_allocs, *m, next_in_flight);
+
+    *host_offset = alloc_cluster_offset + offset_into_cluster(s, guest_offset);
+    *bytes = MIN(*bytes, (nb_sectors * BDRV_SECTOR_SIZE)
+                         - offset_into_cluster(s, guest_offset));
+    assert(*bytes != 0);
+
+    return 1;
+
+fail:
+    if (*m && (*m)->nb_clusters > 0) {
+        QLIST_REMOVE(*m, next_in_flight);
+    }
+    return ret;
+}
+
+/*
+ * alloc_cluster_offset
+ *
+ * For a given offset on the virtual disk, find the cluster offset in qcow2
+ * file. If the offset is not found, allocate a new cluster.
+ *
+ * If the cluster was already allocated, m->nb_clusters is set to 0 and
+ * other fields in m are meaningless.
+ *
+ * If the cluster is newly allocated, m->nb_clusters is set to the number of
+ * contiguous clusters that have been allocated. In this case, the other
+ * fields of m are valid and contain information about the first allocated
+ * cluster.
+ *
+ * If the request conflicts with another write request in flight, the coroutine
+ * is queued and will be reentered when the dependency has completed.
+ *
+ * Return 0 on success and -errno in error cases
+ */
+int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
+    int n_start, int n_end, int *num, uint64_t *host_offset, QCowL2Meta **m)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t start, remaining;
+    uint64_t cluster_offset;
+    uint64_t cur_bytes;
+    int ret;
+
+    trace_qcow2_alloc_clusters_offset(qemu_coroutine_self(), offset,
+                                      n_start, n_end);
+
+    assert(n_start * BDRV_SECTOR_SIZE == offset_into_cluster(s, offset));
+    offset = start_of_cluster(s, offset);
+
+again:
+    start = offset + (n_start << BDRV_SECTOR_BITS);
+    remaining = (n_end - n_start) << BDRV_SECTOR_BITS;
+    cluster_offset = 0;
+    *host_offset = 0;
+    cur_bytes = 0;
+    *m = NULL;
+
+    while (true) {
+
+        if (!*host_offset) {
+            *host_offset = start_of_cluster(s, cluster_offset);
+        }
+
+        assert(remaining >= cur_bytes);
+
+        start           += cur_bytes;
+        remaining       -= cur_bytes;
+        cluster_offset  += cur_bytes;
+
+        if (remaining == 0) {
+            break;
+        }
+
+        cur_bytes = remaining;
+
+        /*
+         * Now start gathering as many contiguous clusters as possible:
+         *
+         * 1. Check for overlaps with in-flight allocations
+         *
+         *      a) Overlap not in the first cluster -> shorten this request and
+         *         let the caller handle the rest in its next loop iteration.
+         *
+         *      b) Real overlaps of two requests. Yield and restart the search
+         *         for contiguous clusters (the situation could have changed
+         *         while we were sleeping)
+         *
+         *      c) TODO: Request starts in the same cluster as the in-flight
+         *         allocation ends. Shorten the COW of the in-fight allocation,
+         *         set cluster_offset to write to the same cluster and set up
+         *         the right synchronisation between the in-flight request and
+         *         the new one.
+         */
+        ret = handle_dependencies(bs, start, &cur_bytes, m);
+        if (ret == -EAGAIN) {
+            /* Currently handle_dependencies() doesn't yield if we already had
+             * an allocation. If it did, we would have to clean up the L2Meta
+             * structs before starting over. */
+            assert(*m == NULL);
+            goto again;
+        } else if (ret < 0) {
+            return ret;
+        } else if (cur_bytes == 0) {
+            break;
+        } else {
+            /* handle_dependencies() may have decreased cur_bytes (shortened
+             * the allocations below) so that the next dependency is processed
+             * correctly during the next loop iteration. */
+        }
+
+        /*
+         * 2. Count contiguous COPIED clusters.
+         */
+        ret = handle_copied(bs, start, &cluster_offset, &cur_bytes, m);
+        if (ret < 0) {
+            return ret;
+        } else if (ret) {
+            continue;
+        } else if (cur_bytes == 0) {
+            break;
+        }
+
+        /*
+         * 3. If the request still hasn't completed, allocate new clusters,
+         *    considering any cluster_offset of steps 1c or 2.
+         */
+        ret = handle_alloc(bs, start, &cluster_offset, &cur_bytes, m);
+        if (ret < 0) {
+            return ret;
+        } else if (ret) {
+            continue;
+        } else {
+            assert(cur_bytes == 0);
+            break;
+        }
+    }
+
+    *num = (n_end - n_start) - (remaining >> BDRV_SECTOR_BITS);
+    assert(*num > 0);
+    assert(*host_offset != 0);
+
+    return 0;
+}
+
+static int decompress_buffer(uint8_t *out_buf, int out_buf_size,
+                             const uint8_t *buf, int buf_size)
+{
+    z_stream strm1, *strm = &strm1;
+    int ret, out_len;
+
+    memset(strm, 0, sizeof(*strm));
+
+    strm->next_in = (uint8_t *)buf;
+    strm->avail_in = buf_size;
+    strm->next_out = out_buf;
+    strm->avail_out = out_buf_size;
+
+    ret = inflateInit2(strm, -12);
+    if (ret != Z_OK)
+        return -1;
+    ret = inflate(strm, Z_FINISH);
+    out_len = strm->next_out - out_buf;
+    if ((ret != Z_STREAM_END && ret != Z_BUF_ERROR) ||
+        out_len != out_buf_size) {
+        inflateEnd(strm);
+        return -1;
+    }
+    inflateEnd(strm);
+    return 0;
+}
+
+int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret, csize, nb_csectors, sector_offset;
+    uint64_t coffset;
+
+    coffset = cluster_offset & s->cluster_offset_mask;
+    if (s->cluster_cache_offset != coffset) {
+        nb_csectors = ((cluster_offset >> s->csize_shift) & s->csize_mask) + 1;
+        sector_offset = coffset & 511;
+        csize = nb_csectors * 512 - sector_offset;
+        BLKDBG_EVENT(bs->file, BLKDBG_READ_COMPRESSED);
+        ret = bdrv_read(bs->file, coffset >> 9, s->cluster_data, nb_csectors);
+        if (ret < 0) {
+            return ret;
+        }
+        if (decompress_buffer(s->cluster_cache, s->cluster_size,
+                              s->cluster_data + sector_offset, csize) < 0) {
+            return -EIO;
+        }
+        s->cluster_cache_offset = coffset;
+    }
+    return 0;
+}
+
+/*
+ * This discards as many clusters of nb_clusters as possible at once (i.e.
+ * all clusters in the same L2 table) and returns the number of discarded
+ * clusters.
+ */
+static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
+    unsigned int nb_clusters)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t *l2_table;
+    int l2_index;
+    int ret;
+    int i;
+
+    ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Limit nb_clusters to one L2 table */
+    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
+
+    for (i = 0; i < nb_clusters; i++) {
+        uint64_t old_offset;
+
+        old_offset = be64_to_cpu(l2_table[l2_index + i]);
+        if ((old_offset & L2E_OFFSET_MASK) == 0) {
+            continue;
+        }
+
+        /* First remove L2 entries */
+        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+        l2_table[l2_index + i] = cpu_to_be64(0);
+
+        /* Then decrease the refcount */
+        qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
+    }
+
+    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    if (ret < 0) {
+        return ret;
+    }
+
+    return nb_clusters;
+}
+
+int qcow2_discard_clusters(BlockDriverState *bs, uint64_t offset,
+    int nb_sectors)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t end_offset;
+    unsigned int nb_clusters;
+    int ret;
+
+    end_offset = offset + (nb_sectors << BDRV_SECTOR_BITS);
+
+    /* Round start up and end down */
+    offset = align_offset(offset, s->cluster_size);
+    end_offset &= ~(s->cluster_size - 1);
+
+    if (offset > end_offset) {
+        return 0;
+    }
+
+    nb_clusters = size_to_clusters(s, end_offset - offset);
+
+    s->cache_discards = true;
+
+    /* Each L2 table is handled by its own loop iteration */
+    while (nb_clusters > 0) {
+        ret = discard_single_l2(bs, offset, nb_clusters);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        nb_clusters -= ret;
+        offset += (ret * s->cluster_size);
+    }
+
+    ret = 0;
+fail:
+    s->cache_discards = false;
+    qcow2_process_discards(bs, ret);
+
+    return ret;
+}
+
+/*
+ * This zeroes as many clusters of nb_clusters as possible at once (i.e.
+ * all clusters in the same L2 table) and returns the number of zeroed
+ * clusters.
+ */
+static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
+    unsigned int nb_clusters)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t *l2_table;
+    int l2_index;
+    int ret;
+    int i;
+
+    ret = get_cluster_table(bs, offset, &l2_table, &l2_index);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Limit nb_clusters to one L2 table */
+    nb_clusters = MIN(nb_clusters, s->l2_size - l2_index);
+
+    for (i = 0; i < nb_clusters; i++) {
+        uint64_t old_offset;
+
+        old_offset = be64_to_cpu(l2_table[l2_index + i]);
+
+        /* Update L2 entries */
+        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+        if (old_offset & QCOW_OFLAG_COMPRESSED) {
+            l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
+            qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
+        } else {
+            l2_table[l2_index + i] |= cpu_to_be64(QCOW_OFLAG_ZERO);
+        }
+    }
+
+    ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    if (ret < 0) {
+        return ret;
+    }
+
+    return nb_clusters;
+}
+
+int qcow2_zero_clusters(BlockDriverState *bs, uint64_t offset, int nb_sectors)
+{
+    BDRVQcowState *s = bs->opaque;
+    unsigned int nb_clusters;
+    int ret;
+
+    /* The zero flag is only supported by version 3 and newer */
+    if (s->qcow_version < 3) {
+        return -ENOTSUP;
+    }
+
+    /* Each L2 table is handled by its own loop iteration */
+    nb_clusters = size_to_clusters(s, nb_sectors << BDRV_SECTOR_BITS);
+
+    s->cache_discards = true;
+
+    while (nb_clusters > 0) {
+        ret = zero_single_l2(bs, offset, nb_clusters);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        nb_clusters -= ret;
+        offset += (ret * s->cluster_size);
+    }
+
+    ret = 0;
+fail:
+    s->cache_discards = false;
+    qcow2_process_discards(bs, ret);
+
+    return ret;
+}
diff --git a/contrib/qemu/block/qcow2-refcount.c b/contrib/qemu/block/qcow2-refcount.c
new file mode 100644
index 000000000..1244693f3
--- /dev/null
+++ b/contrib/qemu/block/qcow2-refcount.c
@@ -0,0 +1,1374 @@
+/*
+ * Block driver for the QCOW version 2 format
+ *
+ * Copyright (c) 2004-2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu-common.h"
+#include "block/block_int.h"
+#include "block/qcow2.h"
+
+static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size);
+static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
+                            int64_t offset, int64_t length,
+                            int addend, enum qcow2_discard_type type);
+
+
+/*********************************************************/
+/* refcount handling */
+
+int qcow2_refcount_init(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret, refcount_table_size2, i;
+
+    refcount_table_size2 = s->refcount_table_size * sizeof(uint64_t);
+    s->refcount_table = g_malloc(refcount_table_size2);
+    if (s->refcount_table_size > 0) {
+        BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD);
+        ret = bdrv_pread(bs->file, s->refcount_table_offset,
+                         s->refcount_table, refcount_table_size2);
+        if (ret != refcount_table_size2)
+            goto fail;
+        for(i = 0; i < s->refcount_table_size; i++)
+            be64_to_cpus(&s->refcount_table[i]);
+    }
+    return 0;
+ fail:
+    return -ENOMEM;
+}
+
+void qcow2_refcount_close(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    g_free(s->refcount_table);
+}
+
+
+static int load_refcount_block(BlockDriverState *bs,
+                               int64_t refcount_block_offset,
+                               void **refcount_block)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_LOAD);
+    ret = qcow2_cache_get(bs, s->refcount_block_cache, refcount_block_offset,
+        refcount_block);
+
+    return ret;
+}
+
+/*
+ * Returns the refcount of the cluster given by its index. Any non-negative
+ * return value is the refcount of the cluster, negative values are -errno
+ * and indicate an error.
+ */
+static int get_refcount(BlockDriverState *bs, int64_t cluster_index)
+{
+    BDRVQcowState *s = bs->opaque;
+    int refcount_table_index, block_index;
+    int64_t refcount_block_offset;
+    int ret;
+    uint16_t *refcount_block;
+    uint16_t refcount;
+
+    refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
+    if (refcount_table_index >= s->refcount_table_size)
+        return 0;
+    refcount_block_offset = s->refcount_table[refcount_table_index];
+    if (!refcount_block_offset)
+        return 0;
+
+    ret = qcow2_cache_get(bs, s->refcount_block_cache, refcount_block_offset,
+        (void**) &refcount_block);
+    if (ret < 0) {
+        return ret;
+    }
+
+    block_index = cluster_index &
+        ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1);
+    refcount = be16_to_cpu(refcount_block[block_index]);
+
+    ret = qcow2_cache_put(bs, s->refcount_block_cache,
+        (void**) &refcount_block);
+    if (ret < 0) {
+        return ret;
+    }
+
+    return refcount;
+}
+
+/*
+ * Rounds the refcount table size up to avoid growing the table for each single
+ * refcount block that is allocated.
+ */
+static unsigned int next_refcount_table_size(BDRVQcowState *s,
+    unsigned int min_size)
+{
+    unsigned int min_clusters = (min_size >> (s->cluster_bits - 3)) + 1;
+    unsigned int refcount_table_clusters =
+        MAX(1, s->refcount_table_size >> (s->cluster_bits - 3));
+
+    while (min_clusters > refcount_table_clusters) {
+        refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2;
+    }
+
+    return refcount_table_clusters << (s->cluster_bits - 3);
+}
+
+
+/* Checks if two offsets are described by the same refcount block */
+static int in_same_refcount_block(BDRVQcowState *s, uint64_t offset_a,
+    uint64_t offset_b)
+{
+    uint64_t block_a = offset_a >> (2 * s->cluster_bits - REFCOUNT_SHIFT);
+    uint64_t block_b = offset_b >> (2 * s->cluster_bits - REFCOUNT_SHIFT);
+
+    return (block_a == block_b);
+}
+
+/*
+ * Loads a refcount block. If it doesn't exist yet, it is allocated first
+ * (including growing the refcount table if needed).
+ *
+ * Returns 0 on success or -errno in error case
+ */
+static int alloc_refcount_block(BlockDriverState *bs,
+    int64_t cluster_index, uint16_t **refcount_block)
+{
+    BDRVQcowState *s = bs->opaque;
+    unsigned int refcount_table_index;
+    int ret;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC);
+
+    /* Find the refcount block for the given cluster */
+    refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
+
+    if (refcount_table_index < s->refcount_table_size) {
+
+        uint64_t refcount_block_offset =
+            s->refcount_table[refcount_table_index] & REFT_OFFSET_MASK;
+
+        /* If it's already there, we're done */
+        if (refcount_block_offset) {
+             return load_refcount_block(bs, refcount_block_offset,
+                 (void**) refcount_block);
+        }
+    }
+
+    /*
+     * If we came here, we need to allocate something. Something is at least
+     * a cluster for the new refcount block. It may also include a new refcount
+     * table if the old refcount table is too small.
+     *
+     * Note that allocating clusters here needs some special care:
+     *
+     * - We can't use the normal qcow2_alloc_clusters(), it would try to
+     *   increase the refcount and very likely we would end up with an endless
+     *   recursion. Instead we must place the refcount blocks in a way that
+     *   they can describe them themselves.
+     *
+     * - We need to consider that at this point we are inside update_refcounts
+     *   and doing the initial refcount increase. This means that some clusters
+     *   have already been allocated by the caller, but their refcount isn't
+     *   accurate yet. free_cluster_index tells us where this allocation ends
+     *   as long as we don't overwrite it by freeing clusters.
+     *
+     * - alloc_clusters_noref and qcow2_free_clusters may load a different
+     *   refcount block into the cache
+     */
+
+    *refcount_block = NULL;
+
+    /* We write to the refcount table, so we might depend on L2 tables */
+    ret = qcow2_cache_flush(bs, s->l2_table_cache);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Allocate the refcount block itself and mark it as used */
+    int64_t new_block = alloc_clusters_noref(bs, s->cluster_size);
+    if (new_block < 0) {
+        return new_block;
+    }
+
+#ifdef DEBUG_ALLOC2
+    fprintf(stderr, "qcow2: Allocate refcount block %d for %" PRIx64
+        " at %" PRIx64 "\n",
+        refcount_table_index, cluster_index << s->cluster_bits, new_block);
+#endif
+
+    if (in_same_refcount_block(s, new_block, cluster_index << s->cluster_bits)) {
+        /* Zero the new refcount block before updating it */
+        ret = qcow2_cache_get_empty(bs, s->refcount_block_cache, new_block,
+            (void**) refcount_block);
+        if (ret < 0) {
+            goto fail_block;
+        }
+
+        memset(*refcount_block, 0, s->cluster_size);
+
+        /* The block describes itself, need to update the cache */
+        int block_index = (new_block >> s->cluster_bits) &
+            ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1);
+        (*refcount_block)[block_index] = cpu_to_be16(1);
+    } else {
+        /* Described somewhere else. This can recurse at most twice before we
+         * arrive at a block that describes itself. */
+        ret = update_refcount(bs, new_block, s->cluster_size, 1,
+                              QCOW2_DISCARD_NEVER);
+        if (ret < 0) {
+            goto fail_block;
+        }
+
+        ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+        if (ret < 0) {
+            goto fail_block;
+        }
+
+        /* Initialize the new refcount block only after updating its refcount,
+         * update_refcount uses the refcount cache itself */
+        ret = qcow2_cache_get_empty(bs, s->refcount_block_cache, new_block,
+            (void**) refcount_block);
+        if (ret < 0) {
+            goto fail_block;
+        }
+
+        memset(*refcount_block, 0, s->cluster_size);
+    }
+
+    /* Now the new refcount block needs to be written to disk */
+    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE);
+    qcow2_cache_entry_mark_dirty(s->refcount_block_cache, *refcount_block);
+    ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+    if (ret < 0) {
+        goto fail_block;
+    }
+
+    /* If the refcount table is big enough, just hook the block up there */
+    if (refcount_table_index < s->refcount_table_size) {
+        uint64_t data64 = cpu_to_be64(new_block);
+        BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_HOOKUP);
+        ret = bdrv_pwrite_sync(bs->file,
+            s->refcount_table_offset + refcount_table_index * sizeof(uint64_t),
+            &data64, sizeof(data64));
+        if (ret < 0) {
+            goto fail_block;
+        }
+
+        s->refcount_table[refcount_table_index] = new_block;
+        return 0;
+    }
+
+    ret = qcow2_cache_put(bs, s->refcount_block_cache, (void**) refcount_block);
+    if (ret < 0) {
+        goto fail_block;
+    }
+
+    /*
+     * If we come here, we need to grow the refcount table. Again, a new
+     * refcount table needs some space and we can't simply allocate to avoid
+     * endless recursion.
+     *
+     * Therefore let's grab new refcount blocks at the end of the image, which
+     * will describe themselves and the new refcount table. This way we can
+     * reference them only in the new table and do the switch to the new
+     * refcount table at once without producing an inconsistent state in
+     * between.
+     */
+    BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_GROW);
+
+    /* Calculate the number of refcount blocks needed so far */
+    uint64_t refcount_block_clusters = 1 << (s->cluster_bits - REFCOUNT_SHIFT);
+    uint64_t blocks_used = (s->free_cluster_index +
+        refcount_block_clusters - 1) / refcount_block_clusters;
+
+    /* And now we need at least one block more for the new metadata */
+    uint64_t table_size = next_refcount_table_size(s, blocks_used + 1);
+    uint64_t last_table_size;
+    uint64_t blocks_clusters;
+    do {
+        uint64_t table_clusters =
+            size_to_clusters(s, table_size * sizeof(uint64_t));
+        blocks_clusters = 1 +
+            ((table_clusters + refcount_block_clusters - 1)
+            / refcount_block_clusters);
+        uint64_t meta_clusters = table_clusters + blocks_clusters;
+
+        last_table_size = table_size;
+        table_size = next_refcount_table_size(s, blocks_used +
+            ((meta_clusters + refcount_block_clusters - 1)
+            / refcount_block_clusters));
+
+    } while (last_table_size != table_size);
+
+#ifdef DEBUG_ALLOC2
+    fprintf(stderr, "qcow2: Grow refcount table %" PRId32 " => %" PRId64 "\n",
+        s->refcount_table_size, table_size);
+#endif
+
+    /* Create the new refcount table and blocks */
+    uint64_t meta_offset = (blocks_used * refcount_block_clusters) *
+        s->cluster_size;
+    uint64_t table_offset = meta_offset + blocks_clusters * s->cluster_size;
+    uint16_t *new_blocks = g_malloc0(blocks_clusters * s->cluster_size);
+    uint64_t *new_table = g_malloc0(table_size * sizeof(uint64_t));
+
+    assert(meta_offset >= (s->free_cluster_index * s->cluster_size));
+
+    /* Fill the new refcount table */
+    memcpy(new_table, s->refcount_table,
+        s->refcount_table_size * sizeof(uint64_t));
+    new_table[refcount_table_index] = new_block;
+
+    int i;
+    for (i = 0; i < blocks_clusters; i++) {
+        new_table[blocks_used + i] = meta_offset + (i * s->cluster_size);
+    }
+
+    /* Fill the refcount blocks */
+    uint64_t table_clusters = size_to_clusters(s, table_size * sizeof(uint64_t));
+    int block = 0;
+    for (i = 0; i < table_clusters + blocks_clusters; i++) {
+        new_blocks[block++] = cpu_to_be16(1);
+    }
+
+    /* Write refcount blocks to disk */
+    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_BLOCKS);
+    ret = bdrv_pwrite_sync(bs->file, meta_offset, new_blocks,
+        blocks_clusters * s->cluster_size);
+    g_free(new_blocks);
+    if (ret < 0) {
+        goto fail_table;
+    }
+
+    /* Write refcount table to disk */
+    for(i = 0; i < table_size; i++) {
+        cpu_to_be64s(&new_table[i]);
+    }
+
+    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_TABLE);
+    ret = bdrv_pwrite_sync(bs->file, table_offset, new_table,
+        table_size * sizeof(uint64_t));
+    if (ret < 0) {
+        goto fail_table;
+    }
+
+    for(i = 0; i < table_size; i++) {
+        be64_to_cpus(&new_table[i]);
+    }
+
+    /* Hook up the new refcount table in the qcow2 header */
+    uint8_t data[12];
+    cpu_to_be64w((uint64_t*)data, table_offset);
+    cpu_to_be32w((uint32_t*)(data + 8), table_clusters);
+    BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_SWITCH_TABLE);
+    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, refcount_table_offset),
+        data, sizeof(data));
+    if (ret < 0) {
+        goto fail_table;
+    }
+
+    /* And switch it in memory */
+    uint64_t old_table_offset = s->refcount_table_offset;
+    uint64_t old_table_size = s->refcount_table_size;
+
+    g_free(s->refcount_table);
+    s->refcount_table = new_table;
+    s->refcount_table_size = table_size;
+    s->refcount_table_offset = table_offset;
+
+    /* Free old table. Remember, we must not change free_cluster_index */
+    uint64_t old_free_cluster_index = s->free_cluster_index;
+    qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t),
+                        QCOW2_DISCARD_OTHER);
+    s->free_cluster_index = old_free_cluster_index;
+
+    ret = load_refcount_block(bs, new_block, (void**) refcount_block);
+    if (ret < 0) {
+        return ret;
+    }
+
+    return 0;
+
+fail_table:
+    g_free(new_table);
+fail_block:
+    if (*refcount_block != NULL) {
+        qcow2_cache_put(bs, s->refcount_block_cache, (void**) refcount_block);
+    }
+    return ret;
+}
+
+void qcow2_process_discards(BlockDriverState *bs, int ret)
+{
+    BDRVQcowState *s = bs->opaque;
+    Qcow2DiscardRegion *d, *next;
+
+    QTAILQ_FOREACH_SAFE(d, &s->discards, next, next) {
+        QTAILQ_REMOVE(&s->discards, d, next);
+
+        /* Discard is optional, ignore the return value */
+        if (ret >= 0) {
+            bdrv_discard(bs->file,
+                         d->offset >> BDRV_SECTOR_BITS,
+                         d->bytes >> BDRV_SECTOR_BITS);
+        }
+
+        g_free(d);
+    }
+}
+
+static void update_refcount_discard(BlockDriverState *bs,
+                                    uint64_t offset, uint64_t length)
+{
+    BDRVQcowState *s = bs->opaque;
+    Qcow2DiscardRegion *d, *p, *next;
+
+    QTAILQ_FOREACH(d, &s->discards, next) {
+        uint64_t new_start = MIN(offset, d->offset);
+        uint64_t new_end = MAX(offset + length, d->offset + d->bytes);
+
+        if (new_end - new_start <= length + d->bytes) {
+            /* There can't be any overlap, areas ending up here have no
+             * references any more and therefore shouldn't get freed another
+             * time. */
+            assert(d->bytes + length == new_end - new_start);
+            d->offset = new_start;
+            d->bytes = new_end - new_start;
+            goto found;
+        }
+    }
+
+    d = g_malloc(sizeof(*d));
+    *d = (Qcow2DiscardRegion) {
+        .bs     = bs,
+        .offset = offset,
+        .bytes  = length,
+    };
+    QTAILQ_INSERT_TAIL(&s->discards, d, next);
+
+found:
+    /* Merge discard requests if they are adjacent now */
+    QTAILQ_FOREACH_SAFE(p, &s->discards, next, next) {
+        if (p == d
+            || p->offset > d->offset + d->bytes
+            || d->offset > p->offset + p->bytes)
+        {
+            continue;
+        }
+
+        /* Still no overlap possible */
+        assert(p->offset == d->offset + d->bytes
+            || d->offset == p->offset + p->bytes);
+
+        QTAILQ_REMOVE(&s->discards, p, next);
+        d->offset = MIN(d->offset, p->offset);
+        d->bytes += p->bytes;
+    }
+}
+
+/* XXX: cache several refcount block clusters ? */
+static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
+    int64_t offset, int64_t length, int addend, enum qcow2_discard_type type)
+{
+    BDRVQcowState *s = bs->opaque;
+    int64_t start, last, cluster_offset;
+    uint16_t *refcount_block = NULL;
+    int64_t old_table_index = -1;
+    int ret;
+
+#ifdef DEBUG_ALLOC2
+    fprintf(stderr, "update_refcount: offset=%" PRId64 " size=%" PRId64 " addend=%d\n",
+           offset, length, addend);
+#endif
+    if (length < 0) {
+        return -EINVAL;
+    } else if (length == 0) {
+        return 0;
+    }
+
+    if (addend < 0) {
+        qcow2_cache_set_dependency(bs, s->refcount_block_cache,
+            s->l2_table_cache);
+    }
+
+    start = offset & ~(s->cluster_size - 1);
+    last = (offset + length - 1) & ~(s->cluster_size - 1);
+    for(cluster_offset = start; cluster_offset <= last;
+        cluster_offset += s->cluster_size)
+    {
+        int block_index, refcount;
+        int64_t cluster_index = cluster_offset >> s->cluster_bits;
+        int64_t table_index =
+            cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
+
+        /* Load the refcount block and allocate it if needed */
+        if (table_index != old_table_index) {
+            if (refcount_block) {
+                ret = qcow2_cache_put(bs, s->refcount_block_cache,
+                    (void**) &refcount_block);
+                if (ret < 0) {
+                    goto fail;
+                }
+            }
+
+            ret = alloc_refcount_block(bs, cluster_index, &refcount_block);
+            if (ret < 0) {
+                goto fail;
+            }
+        }
+        old_table_index = table_index;
+
+        qcow2_cache_entry_mark_dirty(s->refcount_block_cache, refcount_block);
+
+        /* we can update the count and save it */
+        block_index = cluster_index &
+            ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1);
+
+        refcount = be16_to_cpu(refcount_block[block_index]);
+        refcount += addend;
+        if (refcount < 0 || refcount > 0xffff) {
+            ret = -EINVAL;
+            goto fail;
+        }
+        if (refcount == 0 && cluster_index < s->free_cluster_index) {
+            s->free_cluster_index = cluster_index;
+        }
+        refcount_block[block_index] = cpu_to_be16(refcount);
+
+        if (refcount == 0 && s->discard_passthrough[type]) {
+            update_refcount_discard(bs, cluster_offset, s->cluster_size);
+        }
+    }
+
+    ret = 0;
+fail:
+    if (!s->cache_discards) {
+        qcow2_process_discards(bs, ret);
+    }
+
+    /* Write last changed block to disk */
+    if (refcount_block) {
+        int wret;
+        wret = qcow2_cache_put(bs, s->refcount_block_cache,
+            (void**) &refcount_block);
+        if (wret < 0) {
+            return ret < 0 ? ret : wret;
+        }
+    }
+
+    /*
+     * Try do undo any updates if an error is returned (This may succeed in
+     * some cases like ENOSPC for allocating a new refcount block)
+     */
+    if (ret < 0) {
+        int dummy;
+        dummy = update_refcount(bs, offset, cluster_offset - offset, -addend,
+                                QCOW2_DISCARD_NEVER);
+        (void)dummy;
+    }
+
+    return ret;
+}
+
+/*
+ * Increases or decreases the refcount of a given cluster by one.
+ * addend must be 1 or -1.
+ *
+ * If the return value is non-negative, it is the new refcount of the cluster.
+ * If it is negative, it is -errno and indicates an error.
+ */
+static int update_cluster_refcount(BlockDriverState *bs,
+                                   int64_t cluster_index,
+                                   int addend,
+                                   enum qcow2_discard_type type)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret;
+
+    ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend,
+                          type);
+    if (ret < 0) {
+        return ret;
+    }
+
+    return get_refcount(bs, cluster_index);
+}
+
+
+
+/*********************************************************/
+/* cluster allocation functions */
+
+
+
+/* return < 0 if error */
+static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size)
+{
+    BDRVQcowState *s = bs->opaque;
+    int i, nb_clusters, refcount;
+
+    nb_clusters = size_to_clusters(s, size);
+retry:
+    for(i = 0; i < nb_clusters; i++) {
+        int64_t next_cluster_index = s->free_cluster_index++;
+        refcount = get_refcount(bs, next_cluster_index);
+
+        if (refcount < 0) {
+            return refcount;
+        } else if (refcount != 0) {
+            goto retry;
+        }
+    }
+#ifdef DEBUG_ALLOC2
+    fprintf(stderr, "alloc_clusters: size=%" PRId64 " -> %" PRId64 "\n",
+            size,
+            (s->free_cluster_index - nb_clusters) << s->cluster_bits);
+#endif
+    return (s->free_cluster_index - nb_clusters) << s->cluster_bits;
+}
+
+int64_t qcow2_alloc_clusters(BlockDriverState *bs, int64_t size)
+{
+    int64_t offset;
+    int ret;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC);
+    offset = alloc_clusters_noref(bs, size);
+    if (offset < 0) {
+        return offset;
+    }
+
+    ret = update_refcount(bs, offset, size, 1, QCOW2_DISCARD_NEVER);
+    if (ret < 0) {
+        return ret;
+    }
+
+    return offset;
+}
+
+int qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
+    int nb_clusters)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t cluster_index;
+    uint64_t old_free_cluster_index;
+    int i, refcount, ret;
+
+    /* Check how many clusters there are free */
+    cluster_index = offset >> s->cluster_bits;
+    for(i = 0; i < nb_clusters; i++) {
+        refcount = get_refcount(bs, cluster_index++);
+
+        if (refcount < 0) {
+            return refcount;
+        } else if (refcount != 0) {
+            break;
+        }
+    }
+
+    /* And then allocate them */
+    old_free_cluster_index = s->free_cluster_index;
+    s->free_cluster_index = cluster_index + i;
+
+    ret = update_refcount(bs, offset, i << s->cluster_bits, 1,
+                          QCOW2_DISCARD_NEVER);
+    if (ret < 0) {
+        return ret;
+    }
+
+    s->free_cluster_index = old_free_cluster_index;
+
+    return i;
+}
+
+/* only used to allocate compressed sectors. We try to allocate
+   contiguous sectors. size must be <= cluster_size */
+int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
+{
+    BDRVQcowState *s = bs->opaque;
+    int64_t offset, cluster_offset;
+    int free_in_cluster;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC_BYTES);
+    assert(size > 0 && size <= s->cluster_size);
+    if (s->free_byte_offset == 0) {
+        offset = qcow2_alloc_clusters(bs, s->cluster_size);
+        if (offset < 0) {
+            return offset;
+        }
+        s->free_byte_offset = offset;
+    }
+ redo:
+    free_in_cluster = s->cluster_size -
+        (s->free_byte_offset & (s->cluster_size - 1));
+    if (size <= free_in_cluster) {
+        /* enough space in current cluster */
+        offset = s->free_byte_offset;
+        s->free_byte_offset += size;
+        free_in_cluster -= size;
+        if (free_in_cluster == 0)
+            s->free_byte_offset = 0;
+        if ((offset & (s->cluster_size - 1)) != 0)
+            update_cluster_refcount(bs, offset >> s->cluster_bits, 1,
+                                    QCOW2_DISCARD_NEVER);
+    } else {
+        offset = qcow2_alloc_clusters(bs, s->cluster_size);
+        if (offset < 0) {
+            return offset;
+        }
+        cluster_offset = s->free_byte_offset & ~(s->cluster_size - 1);
+        if ((cluster_offset + s->cluster_size) == offset) {
+            /* we are lucky: contiguous data */
+            offset = s->free_byte_offset;
+            update_cluster_refcount(bs, offset >> s->cluster_bits, 1,
+                                    QCOW2_DISCARD_NEVER);
+            s->free_byte_offset += size;
+        } else {
+            s->free_byte_offset = offset;
+            goto redo;
+        }
+    }
+
+    /* The cluster refcount was incremented, either by qcow2_alloc_clusters()
+     * or explicitly by update_cluster_refcount().  Refcount blocks must be
+     * flushed before the caller's L2 table updates.
+     */
+    qcow2_cache_set_dependency(bs, s->l2_table_cache, s->refcount_block_cache);
+    return offset;
+}
+
+void qcow2_free_clusters(BlockDriverState *bs,
+                          int64_t offset, int64_t size,
+                          enum qcow2_discard_type type)
+{
+    int ret;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_FREE);
+    ret = update_refcount(bs, offset, size, -1, type);
+    if (ret < 0) {
+        fprintf(stderr, "qcow2_free_clusters failed: %s\n", strerror(-ret));
+        /* TODO Remember the clusters to free them later and avoid leaking */
+    }
+}
+
+/*
+ * Free a cluster using its L2 entry (handles clusters of all types, e.g.
+ * normal cluster, compressed cluster, etc.)
+ */
+void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
+                             int nb_clusters, enum qcow2_discard_type type)
+{
+    BDRVQcowState *s = bs->opaque;
+
+    switch (qcow2_get_cluster_type(l2_entry)) {
+    case QCOW2_CLUSTER_COMPRESSED:
+        {
+            int nb_csectors;
+            nb_csectors = ((l2_entry >> s->csize_shift) &
+                           s->csize_mask) + 1;
+            qcow2_free_clusters(bs,
+                (l2_entry & s->cluster_offset_mask) & ~511,
+                nb_csectors * 512, type);
+        }
+        break;
+    case QCOW2_CLUSTER_NORMAL:
+        qcow2_free_clusters(bs, l2_entry & L2E_OFFSET_MASK,
+                            nb_clusters << s->cluster_bits, type);
+        break;
+    case QCOW2_CLUSTER_UNALLOCATED:
+    case QCOW2_CLUSTER_ZERO:
+        break;
+    default:
+        abort();
+    }
+}
+
+
+
+/*********************************************************/
+/* snapshots and image creation */
+
+
+
+/* update the refcounts of snapshots and the copied flag */
+int qcow2_update_snapshot_refcount(BlockDriverState *bs,
+    int64_t l1_table_offset, int l1_size, int addend)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t *l1_table, *l2_table, l2_offset, offset, l1_size2, l1_allocated;
+    int64_t old_offset, old_l2_offset;
+    int i, j, l1_modified = 0, nb_csectors, refcount;
+    int ret;
+
+    l2_table = NULL;
+    l1_table = NULL;
+    l1_size2 = l1_size * sizeof(uint64_t);
+
+    s->cache_discards = true;
+
+    /* WARNING: qcow2_snapshot_goto relies on this function not using the
+     * l1_table_offset when it is the current s->l1_table_offset! Be careful
+     * when changing this! */
+    if (l1_table_offset != s->l1_table_offset) {
+        l1_table = g_malloc0(align_offset(l1_size2, 512));
+        l1_allocated = 1;
+
+        ret = bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        for(i = 0;i < l1_size; i++)
+            be64_to_cpus(&l1_table[i]);
+    } else {
+        assert(l1_size == s->l1_size);
+        l1_table = s->l1_table;
+        l1_allocated = 0;
+    }
+
+    for(i = 0; i < l1_size; i++) {
+        l2_offset = l1_table[i];
+        if (l2_offset) {
+            old_l2_offset = l2_offset;
+            l2_offset &= L1E_OFFSET_MASK;
+
+            ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset,
+                (void**) &l2_table);
+            if (ret < 0) {
+                goto fail;
+            }
+
+            for(j = 0; j < s->l2_size; j++) {
+                offset = be64_to_cpu(l2_table[j]);
+                if (offset != 0) {
+                    old_offset = offset;
+                    offset &= ~QCOW_OFLAG_COPIED;
+                    if (offset & QCOW_OFLAG_COMPRESSED) {
+                        nb_csectors = ((offset >> s->csize_shift) &
+                                       s->csize_mask) + 1;
+                        if (addend != 0) {
+                            int ret;
+                            ret = update_refcount(bs,
+                                (offset & s->cluster_offset_mask) & ~511,
+                                nb_csectors * 512, addend,
+                                QCOW2_DISCARD_SNAPSHOT);
+                            if (ret < 0) {
+                                goto fail;
+                            }
+                        }
+                        /* compressed clusters are never modified */
+                        refcount = 2;
+                    } else {
+                        uint64_t cluster_index = (offset & L2E_OFFSET_MASK) >> s->cluster_bits;
+                        if (addend != 0) {
+                            refcount = update_cluster_refcount(bs, cluster_index, addend,
+                                                               QCOW2_DISCARD_SNAPSHOT);
+                        } else {
+                            refcount = get_refcount(bs, cluster_index);
+                        }
+
+                        if (refcount < 0) {
+                            ret = refcount;
+                            goto fail;
+                        }
+                    }
+
+                    if (refcount == 1) {
+                        offset |= QCOW_OFLAG_COPIED;
+                    }
+                    if (offset != old_offset) {
+                        if (addend > 0) {
+                            qcow2_cache_set_dependency(bs, s->l2_table_cache,
+                                s->refcount_block_cache);
+                        }
+                        l2_table[j] = cpu_to_be64(offset);
+                        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+                    }
+                }
+            }
+
+            ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+            if (ret < 0) {
+                goto fail;
+            }
+
+
+            if (addend != 0) {
+                refcount = update_cluster_refcount(bs, l2_offset >> s->cluster_bits, addend,
+                                                   QCOW2_DISCARD_SNAPSHOT);
+            } else {
+                refcount = get_refcount(bs, l2_offset >> s->cluster_bits);
+            }
+            if (refcount < 0) {
+                ret = refcount;
+                goto fail;
+            } else if (refcount == 1) {
+                l2_offset |= QCOW_OFLAG_COPIED;
+            }
+            if (l2_offset != old_l2_offset) {
+                l1_table[i] = l2_offset;
+                l1_modified = 1;
+            }
+        }
+    }
+
+    ret = bdrv_flush(bs);
+fail:
+    if (l2_table) {
+        qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
+    }
+
+    s->cache_discards = false;
+    qcow2_process_discards(bs, ret);
+
+    /* Update L1 only if it isn't deleted anyway (addend = -1) */
+    if (ret == 0 && addend >= 0 && l1_modified) {
+        for (i = 0; i < l1_size; i++) {
+            cpu_to_be64s(&l1_table[i]);
+        }
+
+        ret = bdrv_pwrite_sync(bs->file, l1_table_offset, l1_table, l1_size2);
+
+        for (i = 0; i < l1_size; i++) {
+            be64_to_cpus(&l1_table[i]);
+        }
+    }
+    if (l1_allocated)
+        g_free(l1_table);
+    return ret;
+}
+
+
+
+
+/*********************************************************/
+/* refcount checking functions */
+
+
+
+/*
+ * Increases the refcount for a range of clusters in a given refcount table.
+ * This is used to construct a temporary refcount table out of L1 and L2 tables
+ * which can be compared the the refcount table saved in the image.
+ *
+ * Modifies the number of errors in res.
+ */
+static void inc_refcounts(BlockDriverState *bs,
+                          BdrvCheckResult *res,
+                          uint16_t *refcount_table,
+                          int refcount_table_size,
+                          int64_t offset, int64_t size)
+{
+    BDRVQcowState *s = bs->opaque;
+    int64_t start, last, cluster_offset;
+    int k;
+
+    if (size <= 0)
+        return;
+
+    start = offset & ~(s->cluster_size - 1);
+    last = (offset + size - 1) & ~(s->cluster_size - 1);
+    for(cluster_offset = start; cluster_offset <= last;
+        cluster_offset += s->cluster_size) {
+        k = cluster_offset >> s->cluster_bits;
+        if (k < 0) {
+            fprintf(stderr, "ERROR: invalid cluster offset=0x%" PRIx64 "\n",
+                cluster_offset);
+            res->corruptions++;
+        } else if (k >= refcount_table_size) {
+            fprintf(stderr, "Warning: cluster offset=0x%" PRIx64 " is after "
+                "the end of the image file, can't properly check refcounts.\n",
+                cluster_offset);
+            res->check_errors++;
+        } else {
+            if (++refcount_table[k] == 0) {
+                fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64
+                    "\n", cluster_offset);
+                res->corruptions++;
+            }
+        }
+    }
+}
+
+/* Flags for check_refcounts_l1() and check_refcounts_l2() */
+enum {
+    CHECK_OFLAG_COPIED = 0x1,   /* check QCOW_OFLAG_COPIED matches refcount */
+    CHECK_FRAG_INFO = 0x2,      /* update BlockFragInfo counters */
+};
+
+/*
+ * Increases the refcount in the given refcount table for the all clusters
+ * referenced in the L2 table. While doing so, performs some checks on L2
+ * entries.
+ *
+ * Returns the number of errors found by the checks or -errno if an internal
+ * error occurred.
+ */
+static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
+    uint16_t *refcount_table, int refcount_table_size, int64_t l2_offset,
+    int flags)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t *l2_table, l2_entry;
+    uint64_t next_contiguous_offset = 0;
+    int i, l2_size, nb_csectors, refcount;
+
+    /* Read L2 table from disk */
+    l2_size = s->l2_size * sizeof(uint64_t);
+    l2_table = g_malloc(l2_size);
+
+    if (bdrv_pread(bs->file, l2_offset, l2_table, l2_size) != l2_size)
+        goto fail;
+
+    /* Do the actual checks */
+    for(i = 0; i < s->l2_size; i++) {
+        l2_entry = be64_to_cpu(l2_table[i]);
+
+        switch (qcow2_get_cluster_type(l2_entry)) {
+        case QCOW2_CLUSTER_COMPRESSED:
+            /* Compressed clusters don't have QCOW_OFLAG_COPIED */
+            if (l2_entry & QCOW_OFLAG_COPIED) {
+                fprintf(stderr, "ERROR: cluster %" PRId64 ": "
+                    "copied flag must never be set for compressed "
+                    "clusters\n", l2_entry >> s->cluster_bits);
+                l2_entry &= ~QCOW_OFLAG_COPIED;
+                res->corruptions++;
+            }
+
+            /* Mark cluster as used */
+            nb_csectors = ((l2_entry >> s->csize_shift) &
+                           s->csize_mask) + 1;
+            l2_entry &= s->cluster_offset_mask;
+            inc_refcounts(bs, res, refcount_table, refcount_table_size,
+                l2_entry & ~511, nb_csectors * 512);
+
+            if (flags & CHECK_FRAG_INFO) {
+                res->bfi.allocated_clusters++;
+                res->bfi.compressed_clusters++;
+
+                /* Compressed clusters are fragmented by nature.  Since they
+                 * take up sub-sector space but we only have sector granularity
+                 * I/O we need to re-read the same sectors even for adjacent
+                 * compressed clusters.
+                 */
+                res->bfi.fragmented_clusters++;
+            }
+            break;
+
+        case QCOW2_CLUSTER_ZERO:
+            if ((l2_entry & L2E_OFFSET_MASK) == 0) {
+                break;
+            }
+            /* fall through */
+
+        case QCOW2_CLUSTER_NORMAL:
+        {
+            /* QCOW_OFLAG_COPIED must be set iff refcount == 1 */
+            uint64_t offset = l2_entry & L2E_OFFSET_MASK;
+
+            if (flags & CHECK_OFLAG_COPIED) {
+                refcount = get_refcount(bs, offset >> s->cluster_bits);
+                if (refcount < 0) {
+                    fprintf(stderr, "Can't get refcount for offset %"
+                        PRIx64 ": %s\n", l2_entry, strerror(-refcount));
+                    goto fail;
+                }
+                if ((refcount == 1) != ((l2_entry & QCOW_OFLAG_COPIED) != 0)) {
+                    fprintf(stderr, "ERROR OFLAG_COPIED: offset=%"
+                        PRIx64 " refcount=%d\n", l2_entry, refcount);
+                    res->corruptions++;
+                }
+            }
+
+            if (flags & CHECK_FRAG_INFO) {
+                res->bfi.allocated_clusters++;
+                if (next_contiguous_offset &&
+                    offset != next_contiguous_offset) {
+                    res->bfi.fragmented_clusters++;
+                }
+                next_contiguous_offset = offset + s->cluster_size;
+            }
+
+            /* Mark cluster as used */
+            inc_refcounts(bs, res, refcount_table,refcount_table_size,
+                offset, s->cluster_size);
+
+            /* Correct offsets are cluster aligned */
+            if (offset & (s->cluster_size - 1)) {
+                fprintf(stderr, "ERROR offset=%" PRIx64 ": Cluster is not "
+                    "properly aligned; L2 entry corrupted.\n", offset);
+                res->corruptions++;
+            }
+            break;
+        }
+
+        case QCOW2_CLUSTER_UNALLOCATED:
+            break;
+
+        default:
+            abort();
+        }
+    }
+
+    g_free(l2_table);
+    return 0;
+
+fail:
+    fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n");
+    g_free(l2_table);
+    return -EIO;
+}
+
+/*
+ * Increases the refcount for the L1 table, its L2 tables and all referenced
+ * clusters in the given refcount table. While doing so, performs some checks
+ * on L1 and L2 entries.
+ *
+ * Returns the number of errors found by the checks or -errno if an internal
+ * error occurred.
+ */
+static int check_refcounts_l1(BlockDriverState *bs,
+                              BdrvCheckResult *res,
+                              uint16_t *refcount_table,
+                              int refcount_table_size,
+                              int64_t l1_table_offset, int l1_size,
+                              int flags)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t *l1_table, l2_offset, l1_size2;
+    int i, refcount, ret;
+
+    l1_size2 = l1_size * sizeof(uint64_t);
+
+    /* Mark L1 table as used */
+    inc_refcounts(bs, res, refcount_table, refcount_table_size,
+        l1_table_offset, l1_size2);
+
+    /* Read L1 table entries from disk */
+    if (l1_size2 == 0) {
+        l1_table = NULL;
+    } else {
+        l1_table = g_malloc(l1_size2);
+        if (bdrv_pread(bs->file, l1_table_offset,
+                       l1_table, l1_size2) != l1_size2)
+            goto fail;
+        for(i = 0;i < l1_size; i++)
+            be64_to_cpus(&l1_table[i]);
+    }
+
+    /* Do the actual checks */
+    for(i = 0; i < l1_size; i++) {
+        l2_offset = l1_table[i];
+        if (l2_offset) {
+            /* QCOW_OFLAG_COPIED must be set iff refcount == 1 */
+            if (flags & CHECK_OFLAG_COPIED) {
+                refcount = get_refcount(bs, (l2_offset & ~QCOW_OFLAG_COPIED)
+                    >> s->cluster_bits);
+                if (refcount < 0) {
+                    fprintf(stderr, "Can't get refcount for l2_offset %"
+                        PRIx64 ": %s\n", l2_offset, strerror(-refcount));
+                    goto fail;
+                }
+                if ((refcount == 1) != ((l2_offset & QCOW_OFLAG_COPIED) != 0)) {
+                    fprintf(stderr, "ERROR OFLAG_COPIED: l2_offset=%" PRIx64
+                        " refcount=%d\n", l2_offset, refcount);
+                    res->corruptions++;
+                }
+            }
+
+            /* Mark L2 table as used */
+            l2_offset &= L1E_OFFSET_MASK;
+            inc_refcounts(bs, res, refcount_table, refcount_table_size,
+                l2_offset, s->cluster_size);
+
+            /* L2 tables are cluster aligned */
+            if (l2_offset & (s->cluster_size - 1)) {
+                fprintf(stderr, "ERROR l2_offset=%" PRIx64 ": Table is not "
+                    "cluster aligned; L1 entry corrupted\n", l2_offset);
+                res->corruptions++;
+            }
+
+            /* Process and check L2 entries */
+            ret = check_refcounts_l2(bs, res, refcount_table,
+                                     refcount_table_size, l2_offset, flags);
+            if (ret < 0) {
+                goto fail;
+            }
+        }
+    }
+    g_free(l1_table);
+    return 0;
+
+fail:
+    fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
+    res->check_errors++;
+    g_free(l1_table);
+    return -EIO;
+}
+
+/*
+ * Checks an image for refcount consistency.
+ *
+ * Returns 0 if no errors are found, the number of errors in case the image is
+ * detected as corrupted, and -errno when an internal error occurred.
+ */
+int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
+                          BdrvCheckMode fix)
+{
+    BDRVQcowState *s = bs->opaque;
+    int64_t size, i, highest_cluster;
+    int nb_clusters, refcount1, refcount2;
+    QCowSnapshot *sn;
+    uint16_t *refcount_table;
+    int ret;
+
+    size = bdrv_getlength(bs->file);
+    nb_clusters = size_to_clusters(s, size);
+    refcount_table = g_malloc0(nb_clusters * sizeof(uint16_t));
+
+    res->bfi.total_clusters =
+        size_to_clusters(s, bs->total_sectors * BDRV_SECTOR_SIZE);
+
+    /* header */
+    inc_refcounts(bs, res, refcount_table, nb_clusters,
+        0, s->cluster_size);
+
+    /* current L1 table */
+    ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
+                             s->l1_table_offset, s->l1_size,
+                             CHECK_OFLAG_COPIED | CHECK_FRAG_INFO);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* snapshots */
+    for(i = 0; i < s->nb_snapshots; i++) {
+        sn = s->snapshots + i;
+        ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters,
+            sn->l1_table_offset, sn->l1_size, 0);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+    inc_refcounts(bs, res, refcount_table, nb_clusters,
+        s->snapshots_offset, s->snapshots_size);
+
+    /* refcount data */
+    inc_refcounts(bs, res, refcount_table, nb_clusters,
+        s->refcount_table_offset,
+        s->refcount_table_size * sizeof(uint64_t));
+
+    for(i = 0; i < s->refcount_table_size; i++) {
+        uint64_t offset, cluster;
+        offset = s->refcount_table[i];
+        cluster = offset >> s->cluster_bits;
+
+        /* Refcount blocks are cluster aligned */
+        if (offset & (s->cluster_size - 1)) {
+            fprintf(stderr, "ERROR refcount block %" PRId64 " is not "
+                "cluster aligned; refcount table entry corrupted\n", i);
+            res->corruptions++;
+            continue;
+        }
+
+        if (cluster >= nb_clusters) {
+            fprintf(stderr, "ERROR refcount block %" PRId64
+                    " is outside image\n", i);
+            res->corruptions++;
+            continue;
+        }
+
+        if (offset != 0) {
+            inc_refcounts(bs, res, refcount_table, nb_clusters,
+                offset, s->cluster_size);
+            if (refcount_table[cluster] != 1) {
+                fprintf(stderr, "ERROR refcount block %" PRId64
+                    " refcount=%d\n",
+                    i, refcount_table[cluster]);
+                res->corruptions++;
+            }
+        }
+    }
+
+    /* compare ref counts */
+    for (i = 0, highest_cluster = 0; i < nb_clusters; i++) {
+        refcount1 = get_refcount(bs, i);
+        if (refcount1 < 0) {
+            fprintf(stderr, "Can't get refcount for cluster %" PRId64 ": %s\n",
+                i, strerror(-refcount1));
+            res->check_errors++;
+            continue;
+        }
+
+        refcount2 = refcount_table[i];
+
+        if (refcount1 > 0 || refcount2 > 0) {
+            highest_cluster = i;
+        }
+
+        if (refcount1 != refcount2) {
+
+            /* Check if we're allowed to fix the mismatch */
+            int *num_fixed = NULL;
+            if (refcount1 > refcount2 && (fix & BDRV_FIX_LEAKS)) {
+                num_fixed = &res->leaks_fixed;
+            } else if (refcount1 < refcount2 && (fix & BDRV_FIX_ERRORS)) {
+                num_fixed = &res->corruptions_fixed;
+            }
+
+            fprintf(stderr, "%s cluster %" PRId64 " refcount=%d reference=%d\n",
+                   num_fixed != NULL     ? "Repairing" :
+                   refcount1 < refcount2 ? "ERROR" :
+                                           "Leaked",
+                   i, refcount1, refcount2);
+
+            if (num_fixed) {
+                ret = update_refcount(bs, i << s->cluster_bits, 1,
+                                      refcount2 - refcount1,
+                                      QCOW2_DISCARD_ALWAYS);
+                if (ret >= 0) {
+                    (*num_fixed)++;
+                    continue;
+                }
+            }
+
+            /* And if we couldn't, print an error */
+            if (refcount1 < refcount2) {
+                res->corruptions++;
+            } else {
+                res->leaks++;
+            }
+        }
+    }
+
+    res->image_end_offset = (highest_cluster + 1) * s->cluster_size;
+    ret = 0;
+
+fail:
+    g_free(refcount_table);
+
+    return ret;
+}
+
diff --git a/contrib/qemu/block/qcow2-snapshot.c b/contrib/qemu/block/qcow2-snapshot.c
new file mode 100644
index 000000000..0caac9055
--- /dev/null
+++ b/contrib/qemu/block/qcow2-snapshot.c
@@ -0,0 +1,660 @@
+/*
+ * Block driver for the QCOW version 2 format
+ *
+ * Copyright (c) 2004-2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu-common.h"
+#include "block/block_int.h"
+#include "block/qcow2.h"
+
+typedef struct QEMU_PACKED QCowSnapshotHeader {
+    /* header is 8 byte aligned */
+    uint64_t l1_table_offset;
+
+    uint32_t l1_size;
+    uint16_t id_str_size;
+    uint16_t name_size;
+
+    uint32_t date_sec;
+    uint32_t date_nsec;
+
+    uint64_t vm_clock_nsec;
+
+    uint32_t vm_state_size;
+    uint32_t extra_data_size; /* for extension */
+    /* extra data follows */
+    /* id_str follows */
+    /* name follows  */
+} QCowSnapshotHeader;
+
+typedef struct QEMU_PACKED QCowSnapshotExtraData {
+    uint64_t vm_state_size_large;
+    uint64_t disk_size;
+} QCowSnapshotExtraData;
+
+void qcow2_free_snapshots(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    int i;
+
+    for(i = 0; i < s->nb_snapshots; i++) {
+        g_free(s->snapshots[i].name);
+        g_free(s->snapshots[i].id_str);
+    }
+    g_free(s->snapshots);
+    s->snapshots = NULL;
+    s->nb_snapshots = 0;
+}
+
+int qcow2_read_snapshots(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshotHeader h;
+    QCowSnapshotExtraData extra;
+    QCowSnapshot *sn;
+    int i, id_str_size, name_size;
+    int64_t offset;
+    uint32_t extra_data_size;
+    int ret;
+
+    if (!s->nb_snapshots) {
+        s->snapshots = NULL;
+        s->snapshots_size = 0;
+        return 0;
+    }
+
+    offset = s->snapshots_offset;
+    s->snapshots = g_malloc0(s->nb_snapshots * sizeof(QCowSnapshot));
+
+    for(i = 0; i < s->nb_snapshots; i++) {
+        /* Read statically sized part of the snapshot header */
+        offset = align_offset(offset, 8);
+        ret = bdrv_pread(bs->file, offset, &h, sizeof(h));
+        if (ret < 0) {
+            goto fail;
+        }
+
+        offset += sizeof(h);
+        sn = s->snapshots + i;
+        sn->l1_table_offset = be64_to_cpu(h.l1_table_offset);
+        sn->l1_size = be32_to_cpu(h.l1_size);
+        sn->vm_state_size = be32_to_cpu(h.vm_state_size);
+        sn->date_sec = be32_to_cpu(h.date_sec);
+        sn->date_nsec = be32_to_cpu(h.date_nsec);
+        sn->vm_clock_nsec = be64_to_cpu(h.vm_clock_nsec);
+        extra_data_size = be32_to_cpu(h.extra_data_size);
+
+        id_str_size = be16_to_cpu(h.id_str_size);
+        name_size = be16_to_cpu(h.name_size);
+
+        /* Read extra data */
+        ret = bdrv_pread(bs->file, offset, &extra,
+                         MIN(sizeof(extra), extra_data_size));
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += extra_data_size;
+
+        if (extra_data_size >= 8) {
+            sn->vm_state_size = be64_to_cpu(extra.vm_state_size_large);
+        }
+
+        if (extra_data_size >= 16) {
+            sn->disk_size = be64_to_cpu(extra.disk_size);
+        } else {
+            sn->disk_size = bs->total_sectors * BDRV_SECTOR_SIZE;
+        }
+
+        /* Read snapshot ID */
+        sn->id_str = g_malloc(id_str_size + 1);
+        ret = bdrv_pread(bs->file, offset, sn->id_str, id_str_size);
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += id_str_size;
+        sn->id_str[id_str_size] = '\0';
+
+        /* Read snapshot name */
+        sn->name = g_malloc(name_size + 1);
+        ret = bdrv_pread(bs->file, offset, sn->name, name_size);
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += name_size;
+        sn->name[name_size] = '\0';
+    }
+
+    s->snapshots_size = offset - s->snapshots_offset;
+    return 0;
+
+fail:
+    qcow2_free_snapshots(bs);
+    return ret;
+}
+
+/* add at the end of the file a new list of snapshots */
+static int qcow2_write_snapshots(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshot *sn;
+    QCowSnapshotHeader h;
+    QCowSnapshotExtraData extra;
+    int i, name_size, id_str_size, snapshots_size;
+    struct {
+        uint32_t nb_snapshots;
+        uint64_t snapshots_offset;
+    } QEMU_PACKED header_data;
+    int64_t offset, snapshots_offset;
+    int ret;
+
+    /* compute the size of the snapshots */
+    offset = 0;
+    for(i = 0; i < s->nb_snapshots; i++) {
+        sn = s->snapshots + i;
+        offset = align_offset(offset, 8);
+        offset += sizeof(h);
+        offset += sizeof(extra);
+        offset += strlen(sn->id_str);
+        offset += strlen(sn->name);
+    }
+    snapshots_size = offset;
+
+    /* Allocate space for the new snapshot list */
+    snapshots_offset = qcow2_alloc_clusters(bs, snapshots_size);
+    offset = snapshots_offset;
+    if (offset < 0) {
+        return offset;
+    }
+    ret = bdrv_flush(bs);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Write all snapshots to the new list */
+    for(i = 0; i < s->nb_snapshots; i++) {
+        sn = s->snapshots + i;
+        memset(&h, 0, sizeof(h));
+        h.l1_table_offset = cpu_to_be64(sn->l1_table_offset);
+        h.l1_size = cpu_to_be32(sn->l1_size);
+        /* If it doesn't fit in 32 bit, older implementations should treat it
+         * as a disk-only snapshot rather than truncate the VM state */
+        if (sn->vm_state_size <= 0xffffffff) {
+            h.vm_state_size = cpu_to_be32(sn->vm_state_size);
+        }
+        h.date_sec = cpu_to_be32(sn->date_sec);
+        h.date_nsec = cpu_to_be32(sn->date_nsec);
+        h.vm_clock_nsec = cpu_to_be64(sn->vm_clock_nsec);
+        h.extra_data_size = cpu_to_be32(sizeof(extra));
+
+        memset(&extra, 0, sizeof(extra));
+        extra.vm_state_size_large = cpu_to_be64(sn->vm_state_size);
+        extra.disk_size = cpu_to_be64(sn->disk_size);
+
+        id_str_size = strlen(sn->id_str);
+        name_size = strlen(sn->name);
+        h.id_str_size = cpu_to_be16(id_str_size);
+        h.name_size = cpu_to_be16(name_size);
+        offset = align_offset(offset, 8);
+
+        ret = bdrv_pwrite(bs->file, offset, &h, sizeof(h));
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += sizeof(h);
+
+        ret = bdrv_pwrite(bs->file, offset, &extra, sizeof(extra));
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += sizeof(extra);
+
+        ret = bdrv_pwrite(bs->file, offset, sn->id_str, id_str_size);
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += id_str_size;
+
+        ret = bdrv_pwrite(bs->file, offset, sn->name, name_size);
+        if (ret < 0) {
+            goto fail;
+        }
+        offset += name_size;
+    }
+
+    /*
+     * Update the header to point to the new snapshot table. This requires the
+     * new table and its refcounts to be stable on disk.
+     */
+    ret = bdrv_flush(bs);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    QEMU_BUILD_BUG_ON(offsetof(QCowHeader, snapshots_offset) !=
+        offsetof(QCowHeader, nb_snapshots) + sizeof(header_data.nb_snapshots));
+
+    header_data.nb_snapshots        = cpu_to_be32(s->nb_snapshots);
+    header_data.snapshots_offset    = cpu_to_be64(snapshots_offset);
+
+    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, nb_snapshots),
+                           &header_data, sizeof(header_data));
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* free the old snapshot table */
+    qcow2_free_clusters(bs, s->snapshots_offset, s->snapshots_size,
+                        QCOW2_DISCARD_SNAPSHOT);
+    s->snapshots_offset = snapshots_offset;
+    s->snapshots_size = snapshots_size;
+    return 0;
+
+fail:
+    return ret;
+}
+
+static void find_new_snapshot_id(BlockDriverState *bs,
+                                 char *id_str, int id_str_size)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshot *sn;
+    int i, id, id_max = 0;
+
+    for(i = 0; i < s->nb_snapshots; i++) {
+        sn = s->snapshots + i;
+        id = strtoul(sn->id_str, NULL, 10);
+        if (id > id_max)
+            id_max = id;
+    }
+    snprintf(id_str, id_str_size, "%d", id_max + 1);
+}
+
+static int find_snapshot_by_id(BlockDriverState *bs, const char *id_str)
+{
+    BDRVQcowState *s = bs->opaque;
+    int i;
+
+    for(i = 0; i < s->nb_snapshots; i++) {
+        if (!strcmp(s->snapshots[i].id_str, id_str))
+            return i;
+    }
+    return -1;
+}
+
+static int find_snapshot_by_id_or_name(BlockDriverState *bs, const char *name)
+{
+    BDRVQcowState *s = bs->opaque;
+    int i, ret;
+
+    ret = find_snapshot_by_id(bs, name);
+    if (ret >= 0)
+        return ret;
+    for(i = 0; i < s->nb_snapshots; i++) {
+        if (!strcmp(s->snapshots[i].name, name))
+            return i;
+    }
+    return -1;
+}
+
+/* if no id is provided, a new one is constructed */
+int qcow2_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshot *new_snapshot_list = NULL;
+    QCowSnapshot *old_snapshot_list = NULL;
+    QCowSnapshot sn1, *sn = &sn1;
+    int i, ret;
+    uint64_t *l1_table = NULL;
+    int64_t l1_table_offset;
+
+    memset(sn, 0, sizeof(*sn));
+
+    /* Generate an ID if it wasn't passed */
+    if (sn_info->id_str[0] == '\0') {
+        find_new_snapshot_id(bs, sn_info->id_str, sizeof(sn_info->id_str));
+    }
+
+    /* Check that the ID is unique */
+    if (find_snapshot_by_id(bs, sn_info->id_str) >= 0) {
+        return -EEXIST;
+    }
+
+    /* Populate sn with passed data */
+    sn->id_str = g_strdup(sn_info->id_str);
+    sn->name = g_strdup(sn_info->name);
+
+    sn->disk_size = bs->total_sectors * BDRV_SECTOR_SIZE;
+    sn->vm_state_size = sn_info->vm_state_size;
+    sn->date_sec = sn_info->date_sec;
+    sn->date_nsec = sn_info->date_nsec;
+    sn->vm_clock_nsec = sn_info->vm_clock_nsec;
+
+    /* Allocate the L1 table of the snapshot and copy the current one there. */
+    l1_table_offset = qcow2_alloc_clusters(bs, s->l1_size * sizeof(uint64_t));
+    if (l1_table_offset < 0) {
+        ret = l1_table_offset;
+        goto fail;
+    }
+
+    sn->l1_table_offset = l1_table_offset;
+    sn->l1_size = s->l1_size;
+
+    l1_table = g_malloc(s->l1_size * sizeof(uint64_t));
+    for(i = 0; i < s->l1_size; i++) {
+        l1_table[i] = cpu_to_be64(s->l1_table[i]);
+    }
+
+    ret = bdrv_pwrite(bs->file, sn->l1_table_offset, l1_table,
+                      s->l1_size * sizeof(uint64_t));
+    if (ret < 0) {
+        goto fail;
+    }
+
+    g_free(l1_table);
+    l1_table = NULL;
+
+    /*
+     * Increase the refcounts of all clusters and make sure everything is
+     * stable on disk before updating the snapshot table to contain a pointer
+     * to the new L1 table.
+     */
+    ret = qcow2_update_snapshot_refcount(bs, s->l1_table_offset, s->l1_size, 1);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* Append the new snapshot to the snapshot list */
+    new_snapshot_list = g_malloc((s->nb_snapshots + 1) * sizeof(QCowSnapshot));
+    if (s->snapshots) {
+        memcpy(new_snapshot_list, s->snapshots,
+               s->nb_snapshots * sizeof(QCowSnapshot));
+        old_snapshot_list = s->snapshots;
+    }
+    s->snapshots = new_snapshot_list;
+    s->snapshots[s->nb_snapshots++] = *sn;
+
+    ret = qcow2_write_snapshots(bs);
+    if (ret < 0) {
+        g_free(s->snapshots);
+        s->snapshots = old_snapshot_list;
+        goto fail;
+    }
+
+    g_free(old_snapshot_list);
+
+#ifdef DEBUG_ALLOC
+    {
+      BdrvCheckResult result = {0};
+      qcow2_check_refcounts(bs, &result, 0);
+    }
+#endif
+    return 0;
+
+fail:
+    g_free(sn->id_str);
+    g_free(sn->name);
+    g_free(l1_table);
+
+    return ret;
+}
+
+/* copy the snapshot 'snapshot_name' into the current disk image */
+int qcow2_snapshot_goto(BlockDriverState *bs, const char *snapshot_id)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshot *sn;
+    int i, snapshot_index;
+    int cur_l1_bytes, sn_l1_bytes;
+    int ret;
+    uint64_t *sn_l1_table = NULL;
+
+    /* Search the snapshot */
+    snapshot_index = find_snapshot_by_id_or_name(bs, snapshot_id);
+    if (snapshot_index < 0) {
+        return -ENOENT;
+    }
+    sn = &s->snapshots[snapshot_index];
+
+    if (sn->disk_size != bs->total_sectors * BDRV_SECTOR_SIZE) {
+        error_report("qcow2: Loading snapshots with different disk "
+            "size is not implemented");
+        ret = -ENOTSUP;
+        goto fail;
+    }
+
+    /*
+     * Make sure that the current L1 table is big enough to contain the whole
+     * L1 table of the snapshot. If the snapshot L1 table is smaller, the
+     * current one must be padded with zeros.
+     */
+    ret = qcow2_grow_l1_table(bs, sn->l1_size, true);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    cur_l1_bytes = s->l1_size * sizeof(uint64_t);
+    sn_l1_bytes = sn->l1_size * sizeof(uint64_t);
+
+    /*
+     * Copy the snapshot L1 table to the current L1 table.
+     *
+     * Before overwriting the old current L1 table on disk, make sure to
+     * increase all refcounts for the clusters referenced by the new one.
+     * Decrease the refcount referenced by the old one only when the L1
+     * table is overwritten.
+     */
+    sn_l1_table = g_malloc0(cur_l1_bytes);
+
+    ret = bdrv_pread(bs->file, sn->l1_table_offset, sn_l1_table, sn_l1_bytes);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    ret = qcow2_update_snapshot_refcount(bs, sn->l1_table_offset,
+                                         sn->l1_size, 1);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    ret = bdrv_pwrite_sync(bs->file, s->l1_table_offset, sn_l1_table,
+                           cur_l1_bytes);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /*
+     * Decrease refcount of clusters of current L1 table.
+     *
+     * At this point, the in-memory s->l1_table points to the old L1 table,
+     * whereas on disk we already have the new one.
+     *
+     * qcow2_update_snapshot_refcount special cases the current L1 table to use
+     * the in-memory data instead of really using the offset to load a new one,
+     * which is why this works.
+     */
+    ret = qcow2_update_snapshot_refcount(bs, s->l1_table_offset,
+                                         s->l1_size, -1);
+
+    /*
+     * Now update the in-memory L1 table to be in sync with the on-disk one. We
+     * need to do this even if updating refcounts failed.
+     */
+    for(i = 0;i < s->l1_size; i++) {
+        s->l1_table[i] = be64_to_cpu(sn_l1_table[i]);
+    }
+
+    if (ret < 0) {
+        goto fail;
+    }
+
+    g_free(sn_l1_table);
+    sn_l1_table = NULL;
+
+    /*
+     * Update QCOW_OFLAG_COPIED in the active L1 table (it may have changed
+     * when we decreased the refcount of the old snapshot.
+     */
+    ret = qcow2_update_snapshot_refcount(bs, s->l1_table_offset, s->l1_size, 0);
+    if (ret < 0) {
+        goto fail;
+    }
+
+#ifdef DEBUG_ALLOC
+    {
+        BdrvCheckResult result = {0};
+        qcow2_check_refcounts(bs, &result, 0);
+    }
+#endif
+    return 0;
+
+fail:
+    g_free(sn_l1_table);
+    return ret;
+}
+
+int qcow2_snapshot_delete(BlockDriverState *bs, const char *snapshot_id)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshot sn;
+    int snapshot_index, ret;
+
+    /* Search the snapshot */
+    snapshot_index = find_snapshot_by_id_or_name(bs, snapshot_id);
+    if (snapshot_index < 0) {
+        return -ENOENT;
+    }
+    sn = s->snapshots[snapshot_index];
+
+    /* Remove it from the snapshot list */
+    memmove(s->snapshots + snapshot_index,
+            s->snapshots + snapshot_index + 1,
+            (s->nb_snapshots - snapshot_index - 1) * sizeof(sn));
+    s->nb_snapshots--;
+    ret = qcow2_write_snapshots(bs);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /*
+     * The snapshot is now unused, clean up. If we fail after this point, we
+     * won't recover but just leak clusters.
+     */
+    g_free(sn.id_str);
+    g_free(sn.name);
+
+    /*
+     * Now decrease the refcounts of clusters referenced by the snapshot and
+     * free the L1 table.
+     */
+    ret = qcow2_update_snapshot_refcount(bs, sn.l1_table_offset,
+                                         sn.l1_size, -1);
+    if (ret < 0) {
+        return ret;
+    }
+    qcow2_free_clusters(bs, sn.l1_table_offset, sn.l1_size * sizeof(uint64_t),
+                        QCOW2_DISCARD_SNAPSHOT);
+
+    /* must update the copied flag on the current cluster offsets */
+    ret = qcow2_update_snapshot_refcount(bs, s->l1_table_offset, s->l1_size, 0);
+    if (ret < 0) {
+        return ret;
+    }
+
+#ifdef DEBUG_ALLOC
+    {
+        BdrvCheckResult result = {0};
+        qcow2_check_refcounts(bs, &result, 0);
+    }
+#endif
+    return 0;
+}
+
+int qcow2_snapshot_list(BlockDriverState *bs, QEMUSnapshotInfo **psn_tab)
+{
+    BDRVQcowState *s = bs->opaque;
+    QEMUSnapshotInfo *sn_tab, *sn_info;
+    QCowSnapshot *sn;
+    int i;
+
+    if (!s->nb_snapshots) {
+        *psn_tab = NULL;
+        return s->nb_snapshots;
+    }
+
+    sn_tab = g_malloc0(s->nb_snapshots * sizeof(QEMUSnapshotInfo));
+    for(i = 0; i < s->nb_snapshots; i++) {
+        sn_info = sn_tab + i;
+        sn = s->snapshots + i;
+        pstrcpy(sn_info->id_str, sizeof(sn_info->id_str),
+                sn->id_str);
+        pstrcpy(sn_info->name, sizeof(sn_info->name),
+                sn->name);
+        sn_info->vm_state_size = sn->vm_state_size;
+        sn_info->date_sec = sn->date_sec;
+        sn_info->date_nsec = sn->date_nsec;
+        sn_info->vm_clock_nsec = sn->vm_clock_nsec;
+    }
+    *psn_tab = sn_tab;
+    return s->nb_snapshots;
+}
+
+int qcow2_snapshot_load_tmp(BlockDriverState *bs, const char *snapshot_name)
+{
+    int i, snapshot_index;
+    BDRVQcowState *s = bs->opaque;
+    QCowSnapshot *sn;
+    uint64_t *new_l1_table;
+    int new_l1_bytes;
+    int ret;
+
+    assert(bs->read_only);
+
+    /* Search the snapshot */
+    snapshot_index = find_snapshot_by_id_or_name(bs, snapshot_name);
+    if (snapshot_index < 0) {
+        return -ENOENT;
+    }
+    sn = &s->snapshots[snapshot_index];
+
+    /* Allocate and read in the snapshot's L1 table */
+    new_l1_bytes = s->l1_size * sizeof(uint64_t);
+    new_l1_table = g_malloc0(align_offset(new_l1_bytes, 512));
+
+    ret = bdrv_pread(bs->file, sn->l1_table_offset, new_l1_table, new_l1_bytes);
+    if (ret < 0) {
+        g_free(new_l1_table);
+        return ret;
+    }
+
+    /* Switch the L1 table */
+    g_free(s->l1_table);
+
+    s->l1_size = sn->l1_size;
+    s->l1_table_offset = sn->l1_table_offset;
+    s->l1_table = new_l1_table;
+
+    for(i = 0;i < s->l1_size; i++) {
+        be64_to_cpus(&s->l1_table[i]);
+    }
+
+    return 0;
+}
diff --git a/contrib/qemu/block/qcow2.c b/contrib/qemu/block/qcow2.c
new file mode 100644
index 000000000..0eceefe2c
--- /dev/null
+++ b/contrib/qemu/block/qcow2.c
@@ -0,0 +1,1825 @@
+/*
+ * Block driver for the QCOW version 2 format
+ *
+ * Copyright (c) 2004-2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "qemu-common.h"
+#include "block/block_int.h"
+#include "qemu/module.h"
+#include <zlib.h>
+#include "qemu/aes.h"
+#include "block/qcow2.h"
+#include "qemu/error-report.h"
+#include "qapi/qmp/qerror.h"
+#include "qapi/qmp/qbool.h"
+#include "trace.h"
+
+/*
+  Differences with QCOW:
+
+  - Support for multiple incremental snapshots.
+  - Memory management by reference counts.
+  - Clusters which have a reference count of one have the bit
+    QCOW_OFLAG_COPIED to optimize write performance.
+  - Size of compressed clusters is stored in sectors to reduce bit usage
+    in the cluster offsets.
+  - Support for storing additional data (such as the VM state) in the
+    snapshots.
+  - If a backing store is used, the cluster size is not constrained
+    (could be backported to QCOW).
+  - L2 tables have always a size of one cluster.
+*/
+
+
+typedef struct {
+    uint32_t magic;
+    uint32_t len;
+} QCowExtension;
+
+#define  QCOW2_EXT_MAGIC_END 0
+#define  QCOW2_EXT_MAGIC_BACKING_FORMAT 0xE2792ACA
+#define  QCOW2_EXT_MAGIC_FEATURE_TABLE 0x6803f857
+
+static int qcow2_probe(const uint8_t *buf, int buf_size, const char *filename)
+{
+    const QCowHeader *cow_header = (const void *)buf;
+
+    if (buf_size >= sizeof(QCowHeader) &&
+        be32_to_cpu(cow_header->magic) == QCOW_MAGIC &&
+        be32_to_cpu(cow_header->version) >= 2)
+        return 100;
+    else
+        return 0;
+}
+
+
+/* 
+ * read qcow2 extension and fill bs
+ * start reading from start_offset
+ * finish reading upon magic of value 0 or when end_offset reached
+ * unknown magic is skipped (future extension this version knows nothing about)
+ * return 0 upon success, non-0 otherwise
+ */
+static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
+                                 uint64_t end_offset, void **p_feature_table)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowExtension ext;
+    uint64_t offset;
+    int ret;
+
+#ifdef DEBUG_EXT
+    printf("qcow2_read_extensions: start=%ld end=%ld\n", start_offset, end_offset);
+#endif
+    offset = start_offset;
+    while (offset < end_offset) {
+
+#ifdef DEBUG_EXT
+        /* Sanity check */
+        if (offset > s->cluster_size)
+            printf("qcow2_read_extension: suspicious offset %lu\n", offset);
+
+        printf("attempting to read extended header in offset %lu\n", offset);
+#endif
+
+        if (bdrv_pread(bs->file, offset, &ext, sizeof(ext)) != sizeof(ext)) {
+            fprintf(stderr, "qcow2_read_extension: ERROR: "
+                    "pread fail from offset %" PRIu64 "\n",
+                    offset);
+            return 1;
+        }
+        be32_to_cpus(&ext.magic);
+        be32_to_cpus(&ext.len);
+        offset += sizeof(ext);
+#ifdef DEBUG_EXT
+        printf("ext.magic = 0x%x\n", ext.magic);
+#endif
+        if (ext.len > end_offset - offset) {
+            error_report("Header extension too large");
+            return -EINVAL;
+        }
+
+        switch (ext.magic) {
+        case QCOW2_EXT_MAGIC_END:
+            return 0;
+
+        case QCOW2_EXT_MAGIC_BACKING_FORMAT:
+            if (ext.len >= sizeof(bs->backing_format)) {
+                fprintf(stderr, "ERROR: ext_backing_format: len=%u too large"
+                        " (>=%zu)\n",
+                        ext.len, sizeof(bs->backing_format));
+                return 2;
+            }
+            if (bdrv_pread(bs->file, offset , bs->backing_format,
+                           ext.len) != ext.len)
+                return 3;
+            bs->backing_format[ext.len] = '\0';
+#ifdef DEBUG_EXT
+            printf("Qcow2: Got format extension %s\n", bs->backing_format);
+#endif
+            break;
+
+        case QCOW2_EXT_MAGIC_FEATURE_TABLE:
+            if (p_feature_table != NULL) {
+                void* feature_table = g_malloc0(ext.len + 2 * sizeof(Qcow2Feature));
+                ret = bdrv_pread(bs->file, offset , feature_table, ext.len);
+                if (ret < 0) {
+                    return ret;
+                }
+
+                *p_feature_table = feature_table;
+            }
+            break;
+
+        default:
+            /* unknown magic - save it in case we need to rewrite the header */
+            {
+                Qcow2UnknownHeaderExtension *uext;
+
+                uext = g_malloc0(sizeof(*uext)  + ext.len);
+                uext->magic = ext.magic;
+                uext->len = ext.len;
+                QLIST_INSERT_HEAD(&s->unknown_header_ext, uext, next);
+
+                ret = bdrv_pread(bs->file, offset , uext->data, uext->len);
+                if (ret < 0) {
+                    return ret;
+                }
+            }
+            break;
+        }
+
+        offset += ((ext.len + 7) & ~7);
+    }
+
+    return 0;
+}
+
+static void cleanup_unknown_header_ext(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    Qcow2UnknownHeaderExtension *uext, *next;
+
+    QLIST_FOREACH_SAFE(uext, &s->unknown_header_ext, next, next) {
+        QLIST_REMOVE(uext, next);
+        g_free(uext);
+    }
+}
+
+static void GCC_FMT_ATTR(2, 3) report_unsupported(BlockDriverState *bs,
+    const char *fmt, ...)
+{
+    char msg[64];
+    va_list ap;
+
+    va_start(ap, fmt);
+    vsnprintf(msg, sizeof(msg), fmt, ap);
+    va_end(ap);
+
+    qerror_report(QERR_UNKNOWN_BLOCK_FORMAT_FEATURE,
+        bs->device_name, "qcow2", msg);
+}
+
+static void report_unsupported_feature(BlockDriverState *bs,
+    Qcow2Feature *table, uint64_t mask)
+{
+    while (table && table->name[0] != '\0') {
+        if (table->type == QCOW2_FEAT_TYPE_INCOMPATIBLE) {
+            if (mask & (1 << table->bit)) {
+                report_unsupported(bs, "%.46s",table->name);
+                mask &= ~(1 << table->bit);
+            }
+        }
+        table++;
+    }
+
+    if (mask) {
+        report_unsupported(bs, "Unknown incompatible feature: %" PRIx64, mask);
+    }
+}
+
+/*
+ * Sets the dirty bit and flushes afterwards if necessary.
+ *
+ * The incompatible_features bit is only set if the image file header was
+ * updated successfully.  Therefore it is not required to check the return
+ * value of this function.
+ */
+int qcow2_mark_dirty(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t val;
+    int ret;
+
+    assert(s->qcow_version >= 3);
+
+    if (s->incompatible_features & QCOW2_INCOMPAT_DIRTY) {
+        return 0; /* already dirty */
+    }
+
+    val = cpu_to_be64(s->incompatible_features | QCOW2_INCOMPAT_DIRTY);
+    ret = bdrv_pwrite(bs->file, offsetof(QCowHeader, incompatible_features),
+                      &val, sizeof(val));
+    if (ret < 0) {
+        return ret;
+    }
+    ret = bdrv_flush(bs->file);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Only treat image as dirty if the header was updated successfully */
+    s->incompatible_features |= QCOW2_INCOMPAT_DIRTY;
+    return 0;
+}
+
+/*
+ * Clears the dirty bit and flushes before if necessary.  Only call this
+ * function when there are no pending requests, it does not guard against
+ * concurrent requests dirtying the image.
+ */
+static int qcow2_mark_clean(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+
+    if (s->incompatible_features & QCOW2_INCOMPAT_DIRTY) {
+        int ret = bdrv_flush(bs);
+        if (ret < 0) {
+            return ret;
+        }
+
+        s->incompatible_features &= ~QCOW2_INCOMPAT_DIRTY;
+        return qcow2_update_header(bs);
+    }
+    return 0;
+}
+
+static int qcow2_check(BlockDriverState *bs, BdrvCheckResult *result,
+                       BdrvCheckMode fix)
+{
+    int ret = qcow2_check_refcounts(bs, result, fix);
+    if (ret < 0) {
+        return ret;
+    }
+
+    if (fix && result->check_errors == 0 && result->corruptions == 0) {
+        return qcow2_mark_clean(bs);
+    }
+    return ret;
+}
+
+static QemuOptsList qcow2_runtime_opts = {
+    .name = "qcow2",
+    .head = QTAILQ_HEAD_INITIALIZER(qcow2_runtime_opts.head),
+    .desc = {
+        {
+            .name = "lazy_refcounts",
+            .type = QEMU_OPT_BOOL,
+            .help = "Postpone refcount updates",
+        },
+        {
+            .name = QCOW2_OPT_DISCARD_REQUEST,
+            .type = QEMU_OPT_BOOL,
+            .help = "Pass guest discard requests to the layer below",
+        },
+        {
+            .name = QCOW2_OPT_DISCARD_SNAPSHOT,
+            .type = QEMU_OPT_BOOL,
+            .help = "Generate discard requests when snapshot related space "
+                    "is freed",
+        },
+        {
+            .name = QCOW2_OPT_DISCARD_OTHER,
+            .type = QEMU_OPT_BOOL,
+            .help = "Generate discard requests when other clusters are freed",
+        },
+        { /* end of list */ }
+    },
+};
+
+static int qcow2_open(BlockDriverState *bs, QDict *options, int flags)
+{
+    BDRVQcowState *s = bs->opaque;
+    int len, i, ret = 0;
+    QCowHeader header;
+    QemuOpts *opts;
+    Error *local_err = NULL;
+    uint64_t ext_end;
+    uint64_t l1_vm_state_index;
+
+    ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
+    if (ret < 0) {
+        goto fail;
+    }
+    be32_to_cpus(&header.magic);
+    be32_to_cpus(&header.version);
+    be64_to_cpus(&header.backing_file_offset);
+    be32_to_cpus(&header.backing_file_size);
+    be64_to_cpus(&header.size);
+    be32_to_cpus(&header.cluster_bits);
+    be32_to_cpus(&header.crypt_method);
+    be64_to_cpus(&header.l1_table_offset);
+    be32_to_cpus(&header.l1_size);
+    be64_to_cpus(&header.refcount_table_offset);
+    be32_to_cpus(&header.refcount_table_clusters);
+    be64_to_cpus(&header.snapshots_offset);
+    be32_to_cpus(&header.nb_snapshots);
+
+    if (header.magic != QCOW_MAGIC) {
+        ret = -EMEDIUMTYPE;
+        goto fail;
+    }
+    if (header.version < 2 || header.version > 3) {
+        report_unsupported(bs, "QCOW version %d", header.version);
+        ret = -ENOTSUP;
+        goto fail;
+    }
+
+    s->qcow_version = header.version;
+
+    /* Initialise version 3 header fields */
+    if (header.version == 2) {
+        header.incompatible_features    = 0;
+        header.compatible_features      = 0;
+        header.autoclear_features       = 0;
+        header.refcount_order           = 4;
+        header.header_length            = 72;
+    } else {
+        be64_to_cpus(&header.incompatible_features);
+        be64_to_cpus(&header.compatible_features);
+        be64_to_cpus(&header.autoclear_features);
+        be32_to_cpus(&header.refcount_order);
+        be32_to_cpus(&header.header_length);
+    }
+
+    if (header.header_length > sizeof(header)) {
+        s->unknown_header_fields_size = header.header_length - sizeof(header);
+        s->unknown_header_fields = g_malloc(s->unknown_header_fields_size);
+        ret = bdrv_pread(bs->file, sizeof(header), s->unknown_header_fields,
+                         s->unknown_header_fields_size);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    if (header.backing_file_offset) {
+        ext_end = header.backing_file_offset;
+    } else {
+        ext_end = 1 << header.cluster_bits;
+    }
+
+    /* Handle feature bits */
+    s->incompatible_features    = header.incompatible_features;
+    s->compatible_features      = header.compatible_features;
+    s->autoclear_features       = header.autoclear_features;
+
+    if (s->incompatible_features & ~QCOW2_INCOMPAT_MASK) {
+        void *feature_table = NULL;
+        qcow2_read_extensions(bs, header.header_length, ext_end,
+                              &feature_table);
+        report_unsupported_feature(bs, feature_table,
+                                   s->incompatible_features &
+                                   ~QCOW2_INCOMPAT_MASK);
+        ret = -ENOTSUP;
+        goto fail;
+    }
+
+    /* Check support for various header values */
+    if (header.refcount_order != 4) {
+        report_unsupported(bs, "%d bit reference counts",
+                           1 << header.refcount_order);
+        ret = -ENOTSUP;
+        goto fail;
+    }
+
+    if (header.cluster_bits < MIN_CLUSTER_BITS ||
+        header.cluster_bits > MAX_CLUSTER_BITS) {
+        ret = -EINVAL;
+        goto fail;
+    }
+    if (header.crypt_method > QCOW_CRYPT_AES) {
+        ret = -EINVAL;
+        goto fail;
+    }
+    s->crypt_method_header = header.crypt_method;
+    if (s->crypt_method_header) {
+        bs->encrypted = 1;
+    }
+    s->cluster_bits = header.cluster_bits;
+    s->cluster_size = 1 << s->cluster_bits;
+    s->cluster_sectors = 1 << (s->cluster_bits - 9);
+    s->l2_bits = s->cluster_bits - 3; /* L2 is always one cluster */
+    s->l2_size = 1 << s->l2_bits;
+    bs->total_sectors = header.size / 512;
+    s->csize_shift = (62 - (s->cluster_bits - 8));
+    s->csize_mask = (1 << (s->cluster_bits - 8)) - 1;
+    s->cluster_offset_mask = (1LL << s->csize_shift) - 1;
+    s->refcount_table_offset = header.refcount_table_offset;
+    s->refcount_table_size =
+        header.refcount_table_clusters << (s->cluster_bits - 3);
+
+    s->snapshots_offset = header.snapshots_offset;
+    s->nb_snapshots = header.nb_snapshots;
+
+    /* read the level 1 table */
+    s->l1_size = header.l1_size;
+
+    l1_vm_state_index = size_to_l1(s, header.size);
+    if (l1_vm_state_index > INT_MAX) {
+        ret = -EFBIG;
+        goto fail;
+    }
+    s->l1_vm_state_index = l1_vm_state_index;
+
+    /* the L1 table must contain at least enough entries to put
+       header.size bytes */
+    if (s->l1_size < s->l1_vm_state_index) {
+        ret = -EINVAL;
+        goto fail;
+    }
+    s->l1_table_offset = header.l1_table_offset;
+    if (s->l1_size > 0) {
+        s->l1_table = g_malloc0(
+            align_offset(s->l1_size * sizeof(uint64_t), 512));
+        ret = bdrv_pread(bs->file, s->l1_table_offset, s->l1_table,
+                         s->l1_size * sizeof(uint64_t));
+        if (ret < 0) {
+            goto fail;
+        }
+        for(i = 0;i < s->l1_size; i++) {
+            be64_to_cpus(&s->l1_table[i]);
+        }
+    }
+
+    /* alloc L2 table/refcount block cache */
+    s->l2_table_cache = qcow2_cache_create(bs, L2_CACHE_SIZE);
+    s->refcount_block_cache = qcow2_cache_create(bs, REFCOUNT_CACHE_SIZE);
+
+    s->cluster_cache = g_malloc(s->cluster_size);
+    /* one more sector for decompressed data alignment */
+    s->cluster_data = qemu_blockalign(bs, QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size
+                                  + 512);
+    s->cluster_cache_offset = -1;
+    s->flags = flags;
+
+    ret = qcow2_refcount_init(bs);
+    if (ret != 0) {
+        goto fail;
+    }
+
+    QLIST_INIT(&s->cluster_allocs);
+    QTAILQ_INIT(&s->discards);
+
+    /* read qcow2 extensions */
+    if (qcow2_read_extensions(bs, header.header_length, ext_end, NULL)) {
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    /* read the backing file name */
+    if (header.backing_file_offset != 0) {
+        len = header.backing_file_size;
+        if (len > 1023) {
+            len = 1023;
+        }
+        ret = bdrv_pread(bs->file, header.backing_file_offset,
+                         bs->backing_file, len);
+        if (ret < 0) {
+            goto fail;
+        }
+        bs->backing_file[len] = '\0';
+    }
+
+    ret = qcow2_read_snapshots(bs);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* Clear unknown autoclear feature bits */
+    if (!bs->read_only && s->autoclear_features != 0) {
+        s->autoclear_features = 0;
+        ret = qcow2_update_header(bs);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    /* Initialise locks */
+    qemu_co_mutex_init(&s->lock);
+
+    /* Repair image if dirty */
+    if (!(flags & BDRV_O_CHECK) && !bs->read_only &&
+        (s->incompatible_features & QCOW2_INCOMPAT_DIRTY)) {
+        BdrvCheckResult result = {0};
+
+        ret = qcow2_check(bs, &result, BDRV_FIX_ERRORS);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    /* Enable lazy_refcounts according to image and command line options */
+    opts = qemu_opts_create_nofail(&qcow2_runtime_opts);
+    qemu_opts_absorb_qdict(opts, options, &local_err);
+    if (error_is_set(&local_err)) {
+        qerror_report_err(local_err);
+        error_free(local_err);
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    s->use_lazy_refcounts = qemu_opt_get_bool(opts, QCOW2_OPT_LAZY_REFCOUNTS,
+        (s->compatible_features & QCOW2_COMPAT_LAZY_REFCOUNTS));
+
+    s->discard_passthrough[QCOW2_DISCARD_NEVER] = false;
+    s->discard_passthrough[QCOW2_DISCARD_ALWAYS] = true;
+    s->discard_passthrough[QCOW2_DISCARD_REQUEST] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_REQUEST,
+                          flags & BDRV_O_UNMAP);
+    s->discard_passthrough[QCOW2_DISCARD_SNAPSHOT] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_SNAPSHOT, true);
+    s->discard_passthrough[QCOW2_DISCARD_OTHER] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_OTHER, false);
+
+    qemu_opts_del(opts);
+
+    if (s->use_lazy_refcounts && s->qcow_version < 3) {
+        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Lazy refcounts require "
+            "a qcow2 image with at least qemu 1.1 compatibility level");
+        ret = -EINVAL;
+        goto fail;
+    }
+
+#ifdef DEBUG_ALLOC
+    {
+        BdrvCheckResult result = {0};
+        qcow2_check_refcounts(bs, &result, 0);
+    }
+#endif
+    return ret;
+
+ fail:
+    g_free(s->unknown_header_fields);
+    cleanup_unknown_header_ext(bs);
+    qcow2_free_snapshots(bs);
+    qcow2_refcount_close(bs);
+    g_free(s->l1_table);
+    if (s->l2_table_cache) {
+        qcow2_cache_destroy(bs, s->l2_table_cache);
+    }
+    g_free(s->cluster_cache);
+    qemu_vfree(s->cluster_data);
+    return ret;
+}
+
+static int qcow2_set_key(BlockDriverState *bs, const char *key)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint8_t keybuf[16];
+    int len, i;
+
+    memset(keybuf, 0, 16);
+    len = strlen(key);
+    if (len > 16)
+        len = 16;
+    /* XXX: we could compress the chars to 7 bits to increase
+       entropy */
+    for(i = 0;i < len;i++) {
+        keybuf[i] = key[i];
+    }
+    s->crypt_method = s->crypt_method_header;
+
+    if (AES_set_encrypt_key(keybuf, 128, &s->aes_encrypt_key) != 0)
+        return -1;
+    if (AES_set_decrypt_key(keybuf, 128, &s->aes_decrypt_key) != 0)
+        return -1;
+#if 0
+    /* test */
+    {
+        uint8_t in[16];
+        uint8_t out[16];
+        uint8_t tmp[16];
+        for(i=0;i<16;i++)
+            in[i] = i;
+        AES_encrypt(in, tmp, &s->aes_encrypt_key);
+        AES_decrypt(tmp, out, &s->aes_decrypt_key);
+        for(i = 0; i < 16; i++)
+            printf(" %02x", tmp[i]);
+        printf("\n");
+        for(i = 0; i < 16; i++)
+            printf(" %02x", out[i]);
+        printf("\n");
+    }
+#endif
+    return 0;
+}
+
+/* We have nothing to do for QCOW2 reopen, stubs just return
+ * success */
+static int qcow2_reopen_prepare(BDRVReopenState *state,
+                                BlockReopenQueue *queue, Error **errp)
+{
+    return 0;
+}
+
+static int coroutine_fn qcow2_co_is_allocated(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors, int *pnum)
+{
+    BDRVQcowState *s = bs->opaque;
+    uint64_t cluster_offset;
+    int ret;
+
+    *pnum = nb_sectors;
+    /* FIXME We can get errors here, but the bdrv_co_is_allocated interface
+     * can't pass them on today */
+    qemu_co_mutex_lock(&s->lock);
+    ret = qcow2_get_cluster_offset(bs, sector_num << 9, pnum, &cluster_offset);
+    qemu_co_mutex_unlock(&s->lock);
+    if (ret < 0) {
+        *pnum = 0;
+    }
+
+    return (cluster_offset != 0) || (ret == QCOW2_CLUSTER_ZERO);
+}
+
+/* handle reading after the end of the backing file */
+int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov,
+                  int64_t sector_num, int nb_sectors)
+{
+    int n1;
+    if ((sector_num + nb_sectors) <= bs->total_sectors)
+        return nb_sectors;
+    if (sector_num >= bs->total_sectors)
+        n1 = 0;
+    else
+        n1 = bs->total_sectors - sector_num;
+
+    qemu_iovec_memset(qiov, 512 * n1, 0, 512 * (nb_sectors - n1));
+
+    return n1;
+}
+
+static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
+                          int remaining_sectors, QEMUIOVector *qiov)
+{
+    BDRVQcowState *s = bs->opaque;
+    int index_in_cluster, n1;
+    int ret;
+    int cur_nr_sectors; /* number of sectors in current iteration */
+    uint64_t cluster_offset = 0;
+    uint64_t bytes_done = 0;
+    QEMUIOVector hd_qiov;
+    uint8_t *cluster_data = NULL;
+
+    qemu_iovec_init(&hd_qiov, qiov->niov);
+
+    qemu_co_mutex_lock(&s->lock);
+
+    while (remaining_sectors != 0) {
+
+        /* prepare next request */
+        cur_nr_sectors = remaining_sectors;
+        if (s->crypt_method) {
+            cur_nr_sectors = MIN(cur_nr_sectors,
+                QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors);
+        }
+
+        ret = qcow2_get_cluster_offset(bs, sector_num << 9,
+            &cur_nr_sectors, &cluster_offset);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        index_in_cluster = sector_num & (s->cluster_sectors - 1);
+
+        qemu_iovec_reset(&hd_qiov);
+        qemu_iovec_concat(&hd_qiov, qiov, bytes_done,
+            cur_nr_sectors * 512);
+
+        switch (ret) {
+        case QCOW2_CLUSTER_UNALLOCATED:
+
+            if (bs->backing_hd) {
+                /* read from the base image */
+                n1 = qcow2_backing_read1(bs->backing_hd, &hd_qiov,
+                    sector_num, cur_nr_sectors);
+                if (n1 > 0) {
+                    BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO);
+                    qemu_co_mutex_unlock(&s->lock);
+                    ret = bdrv_co_readv(bs->backing_hd, sector_num,
+                                        n1, &hd_qiov);
+                    qemu_co_mutex_lock(&s->lock);
+                    if (ret < 0) {
+                        goto fail;
+                    }
+                }
+            } else {
+                /* Note: in this case, no need to wait */
+                qemu_iovec_memset(&hd_qiov, 0, 0, 512 * cur_nr_sectors);
+            }
+            break;
+
+        case QCOW2_CLUSTER_ZERO:
+            qemu_iovec_memset(&hd_qiov, 0, 0, 512 * cur_nr_sectors);
+            break;
+
+        case QCOW2_CLUSTER_COMPRESSED:
+            /* add AIO support for compressed blocks ? */
+            ret = qcow2_decompress_cluster(bs, cluster_offset);
+            if (ret < 0) {
+                goto fail;
+            }
+
+            qemu_iovec_from_buf(&hd_qiov, 0,
+                s->cluster_cache + index_in_cluster * 512,
+                512 * cur_nr_sectors);
+            break;
+
+        case QCOW2_CLUSTER_NORMAL:
+            if ((cluster_offset & 511) != 0) {
+                ret = -EIO;
+                goto fail;
+            }
+
+            if (s->crypt_method) {
+                /*
+                 * For encrypted images, read everything into a temporary
+                 * contiguous buffer on which the AES functions can work.
+                 */
+                if (!cluster_data) {
+                    cluster_data =
+                        qemu_blockalign(bs, QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size);
+                }
+
+                assert(cur_nr_sectors <=
+                    QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors);
+                qemu_iovec_reset(&hd_qiov);
+                qemu_iovec_add(&hd_qiov, cluster_data,
+                    512 * cur_nr_sectors);
+            }
+
+            BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
+            qemu_co_mutex_unlock(&s->lock);
+            ret = bdrv_co_readv(bs->file,
+                                (cluster_offset >> 9) + index_in_cluster,
+                                cur_nr_sectors, &hd_qiov);
+            qemu_co_mutex_lock(&s->lock);
+            if (ret < 0) {
+                goto fail;
+            }
+            if (s->crypt_method) {
+                qcow2_encrypt_sectors(s, sector_num,  cluster_data,
+                    cluster_data, cur_nr_sectors, 0, &s->aes_decrypt_key);
+                qemu_iovec_from_buf(qiov, bytes_done,
+                    cluster_data, 512 * cur_nr_sectors);
+            }
+            break;
+
+        default:
+            g_assert_not_reached();
+            ret = -EIO;
+            goto fail;
+        }
+
+        remaining_sectors -= cur_nr_sectors;
+        sector_num += cur_nr_sectors;
+        bytes_done += cur_nr_sectors * 512;
+    }
+    ret = 0;
+
+fail:
+    qemu_co_mutex_unlock(&s->lock);
+
+    qemu_iovec_destroy(&hd_qiov);
+    qemu_vfree(cluster_data);
+
+    return ret;
+}
+
+static coroutine_fn int qcow2_co_writev(BlockDriverState *bs,
+                           int64_t sector_num,
+                           int remaining_sectors,
+                           QEMUIOVector *qiov)
+{
+    BDRVQcowState *s = bs->opaque;
+    int index_in_cluster;
+    int n_end;
+    int ret;
+    int cur_nr_sectors; /* number of sectors in current iteration */
+    uint64_t cluster_offset;
+    QEMUIOVector hd_qiov;
+    uint64_t bytes_done = 0;
+    uint8_t *cluster_data = NULL;
+    QCowL2Meta *l2meta = NULL;
+
+    trace_qcow2_writev_start_req(qemu_coroutine_self(), sector_num,
+                                 remaining_sectors);
+
+    qemu_iovec_init(&hd_qiov, qiov->niov);
+
+    s->cluster_cache_offset = -1; /* disable compressed cache */
+
+    qemu_co_mutex_lock(&s->lock);
+
+    while (remaining_sectors != 0) {
+
+        l2meta = NULL;
+
+        trace_qcow2_writev_start_part(qemu_coroutine_self());
+        index_in_cluster = sector_num & (s->cluster_sectors - 1);
+        n_end = index_in_cluster + remaining_sectors;
+        if (s->crypt_method &&
+            n_end > QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors) {
+            n_end = QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors;
+        }
+
+        ret = qcow2_alloc_cluster_offset(bs, sector_num << 9,
+            index_in_cluster, n_end, &cur_nr_sectors, &cluster_offset, &l2meta);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        assert((cluster_offset & 511) == 0);
+
+        qemu_iovec_reset(&hd_qiov);
+        qemu_iovec_concat(&hd_qiov, qiov, bytes_done,
+            cur_nr_sectors * 512);
+
+        if (s->crypt_method) {
+            if (!cluster_data) {
+                cluster_data = qemu_blockalign(bs, QCOW_MAX_CRYPT_CLUSTERS *
+                                                 s->cluster_size);
+            }
+
+            assert(hd_qiov.size <=
+                   QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size);
+            qemu_iovec_to_buf(&hd_qiov, 0, cluster_data, hd_qiov.size);
+
+            qcow2_encrypt_sectors(s, sector_num, cluster_data,
+                cluster_data, cur_nr_sectors, 1, &s->aes_encrypt_key);
+
+            qemu_iovec_reset(&hd_qiov);
+            qemu_iovec_add(&hd_qiov, cluster_data,
+                cur_nr_sectors * 512);
+        }
+
+        qemu_co_mutex_unlock(&s->lock);
+        BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO);
+        trace_qcow2_writev_data(qemu_coroutine_self(),
+                                (cluster_offset >> 9) + index_in_cluster);
+        ret = bdrv_co_writev(bs->file,
+                             (cluster_offset >> 9) + index_in_cluster,
+                             cur_nr_sectors, &hd_qiov);
+        qemu_co_mutex_lock(&s->lock);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        while (l2meta != NULL) {
+            QCowL2Meta *next;
+
+            ret = qcow2_alloc_cluster_link_l2(bs, l2meta);
+            if (ret < 0) {
+                goto fail;
+            }
+
+            /* Take the request off the list of running requests */
+            if (l2meta->nb_clusters != 0) {
+                QLIST_REMOVE(l2meta, next_in_flight);
+            }
+
+            qemu_co_queue_restart_all(&l2meta->dependent_requests);
+
+            next = l2meta->next;
+            g_free(l2meta);
+            l2meta = next;
+        }
+
+        remaining_sectors -= cur_nr_sectors;
+        sector_num += cur_nr_sectors;
+        bytes_done += cur_nr_sectors * 512;
+        trace_qcow2_writev_done_part(qemu_coroutine_self(), cur_nr_sectors);
+    }
+    ret = 0;
+
+fail:
+    qemu_co_mutex_unlock(&s->lock);
+
+    while (l2meta != NULL) {
+        QCowL2Meta *next;
+
+        if (l2meta->nb_clusters != 0) {
+            QLIST_REMOVE(l2meta, next_in_flight);
+        }
+        qemu_co_queue_restart_all(&l2meta->dependent_requests);
+
+        next = l2meta->next;
+        g_free(l2meta);
+        l2meta = next;
+    }
+
+    qemu_iovec_destroy(&hd_qiov);
+    qemu_vfree(cluster_data);
+    trace_qcow2_writev_done_req(qemu_coroutine_self(), ret);
+
+    return ret;
+}
+
+static void qcow2_close(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    g_free(s->l1_table);
+
+    qcow2_cache_flush(bs, s->l2_table_cache);
+    qcow2_cache_flush(bs, s->refcount_block_cache);
+
+    qcow2_mark_clean(bs);
+
+    qcow2_cache_destroy(bs, s->l2_table_cache);
+    qcow2_cache_destroy(bs, s->refcount_block_cache);
+
+    g_free(s->unknown_header_fields);
+    cleanup_unknown_header_ext(bs);
+
+    g_free(s->cluster_cache);
+    qemu_vfree(s->cluster_data);
+    qcow2_refcount_close(bs);
+    qcow2_free_snapshots(bs);
+}
+
+static void qcow2_invalidate_cache(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    int flags = s->flags;
+    AES_KEY aes_encrypt_key;
+    AES_KEY aes_decrypt_key;
+    uint32_t crypt_method = 0;
+    QDict *options;
+
+    /*
+     * Backing files are read-only which makes all of their metadata immutable,
+     * that means we don't have to worry about reopening them here.
+     */
+
+    if (s->crypt_method) {
+        crypt_method = s->crypt_method;
+        memcpy(&aes_encrypt_key, &s->aes_encrypt_key, sizeof(aes_encrypt_key));
+        memcpy(&aes_decrypt_key, &s->aes_decrypt_key, sizeof(aes_decrypt_key));
+    }
+
+    qcow2_close(bs);
+
+    options = qdict_new();
+    qdict_put(options, QCOW2_OPT_LAZY_REFCOUNTS,
+              qbool_from_int(s->use_lazy_refcounts));
+
+    memset(s, 0, sizeof(BDRVQcowState));
+    qcow2_open(bs, options, flags);
+
+    QDECREF(options);
+
+    if (crypt_method) {
+        s->crypt_method = crypt_method;
+        memcpy(&s->aes_encrypt_key, &aes_encrypt_key, sizeof(aes_encrypt_key));
+        memcpy(&s->aes_decrypt_key, &aes_decrypt_key, sizeof(aes_decrypt_key));
+    }
+}
+
+static size_t header_ext_add(char *buf, uint32_t magic, const void *s,
+    size_t len, size_t buflen)
+{
+    QCowExtension *ext_backing_fmt = (QCowExtension*) buf;
+    size_t ext_len = sizeof(QCowExtension) + ((len + 7) & ~7);
+
+    if (buflen < ext_len) {
+        return -ENOSPC;
+    }
+
+    *ext_backing_fmt = (QCowExtension) {
+        .magic  = cpu_to_be32(magic),
+        .len    = cpu_to_be32(len),
+    };
+    memcpy(buf + sizeof(QCowExtension), s, len);
+
+    return ext_len;
+}
+
+/*
+ * Updates the qcow2 header, including the variable length parts of it, i.e.
+ * the backing file name and all extensions. qcow2 was not designed to allow
+ * such changes, so if we run out of space (we can only use the first cluster)
+ * this function may fail.
+ *
+ * Returns 0 on success, -errno in error cases.
+ */
+int qcow2_update_header(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    QCowHeader *header;
+    char *buf;
+    size_t buflen = s->cluster_size;
+    int ret;
+    uint64_t total_size;
+    uint32_t refcount_table_clusters;
+    size_t header_length;
+    Qcow2UnknownHeaderExtension *uext;
+
+    buf = qemu_blockalign(bs, buflen);
+
+    /* Header structure */
+    header = (QCowHeader*) buf;
+
+    if (buflen < sizeof(*header)) {
+        ret = -ENOSPC;
+        goto fail;
+    }
+
+    header_length = sizeof(*header) + s->unknown_header_fields_size;
+    total_size = bs->total_sectors * BDRV_SECTOR_SIZE;
+    refcount_table_clusters = s->refcount_table_size >> (s->cluster_bits - 3);
+
+    *header = (QCowHeader) {
+        /* Version 2 fields */
+        .magic                  = cpu_to_be32(QCOW_MAGIC),
+        .version                = cpu_to_be32(s->qcow_version),
+        .backing_file_offset    = 0,
+        .backing_file_size      = 0,
+        .cluster_bits           = cpu_to_be32(s->cluster_bits),
+        .size                   = cpu_to_be64(total_size),
+        .crypt_method           = cpu_to_be32(s->crypt_method_header),
+        .l1_size                = cpu_to_be32(s->l1_size),
+        .l1_table_offset        = cpu_to_be64(s->l1_table_offset),
+        .refcount_table_offset  = cpu_to_be64(s->refcount_table_offset),
+        .refcount_table_clusters = cpu_to_be32(refcount_table_clusters),
+        .nb_snapshots           = cpu_to_be32(s->nb_snapshots),
+        .snapshots_offset       = cpu_to_be64(s->snapshots_offset),
+
+        /* Version 3 fields */
+        .incompatible_features  = cpu_to_be64(s->incompatible_features),
+        .compatible_features    = cpu_to_be64(s->compatible_features),
+        .autoclear_features     = cpu_to_be64(s->autoclear_features),
+        .refcount_order         = cpu_to_be32(3 + REFCOUNT_SHIFT),
+        .header_length          = cpu_to_be32(header_length),
+    };
+
+    /* For older versions, write a shorter header */
+    switch (s->qcow_version) {
+    case 2:
+        ret = offsetof(QCowHeader, incompatible_features);
+        break;
+    case 3:
+        ret = sizeof(*header);
+        break;
+    default:
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    buf += ret;
+    buflen -= ret;
+    memset(buf, 0, buflen);
+
+    /* Preserve any unknown field in the header */
+    if (s->unknown_header_fields_size) {
+        if (buflen < s->unknown_header_fields_size) {
+            ret = -ENOSPC;
+            goto fail;
+        }
+
+        memcpy(buf, s->unknown_header_fields, s->unknown_header_fields_size);
+        buf += s->unknown_header_fields_size;
+        buflen -= s->unknown_header_fields_size;
+    }
+
+    /* Backing file format header extension */
+    if (*bs->backing_format) {
+        ret = header_ext_add(buf, QCOW2_EXT_MAGIC_BACKING_FORMAT,
+                             bs->backing_format, strlen(bs->backing_format),
+                             buflen);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        buf += ret;
+        buflen -= ret;
+    }
+
+    /* Feature table */
+    Qcow2Feature features[] = {
+        {
+            .type = QCOW2_FEAT_TYPE_INCOMPATIBLE,
+            .bit  = QCOW2_INCOMPAT_DIRTY_BITNR,
+            .name = "dirty bit",
+        },
+        {
+            .type = QCOW2_FEAT_TYPE_COMPATIBLE,
+            .bit  = QCOW2_COMPAT_LAZY_REFCOUNTS_BITNR,
+            .name = "lazy refcounts",
+        },
+    };
+
+    ret = header_ext_add(buf, QCOW2_EXT_MAGIC_FEATURE_TABLE,
+                         features, sizeof(features), buflen);
+    if (ret < 0) {
+        goto fail;
+    }
+    buf += ret;
+    buflen -= ret;
+
+    /* Keep unknown header extensions */
+    QLIST_FOREACH(uext, &s->unknown_header_ext, next) {
+        ret = header_ext_add(buf, uext->magic, uext->data, uext->len, buflen);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        buf += ret;
+        buflen -= ret;
+    }
+
+    /* End of header extensions */
+    ret = header_ext_add(buf, QCOW2_EXT_MAGIC_END, NULL, 0, buflen);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    buf += ret;
+    buflen -= ret;
+
+    /* Backing file name */
+    if (*bs->backing_file) {
+        size_t backing_file_len = strlen(bs->backing_file);
+
+        if (buflen < backing_file_len) {
+            ret = -ENOSPC;
+            goto fail;
+        }
+
+        /* Using strncpy is ok here, since buf is not NUL-terminated. */
+        strncpy(buf, bs->backing_file, buflen);
+
+        header->backing_file_offset = cpu_to_be64(buf - ((char*) header));
+        header->backing_file_size   = cpu_to_be32(backing_file_len);
+    }
+
+    /* Write the new header */
+    ret = bdrv_pwrite(bs->file, 0, header, s->cluster_size);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    ret = 0;
+fail:
+    qemu_vfree(header);
+    return ret;
+}
+
+static int qcow2_change_backing_file(BlockDriverState *bs,
+    const char *backing_file, const char *backing_fmt)
+{
+    pstrcpy(bs->backing_file, sizeof(bs->backing_file), backing_file ?: "");
+    pstrcpy(bs->backing_format, sizeof(bs->backing_format), backing_fmt ?: "");
+
+    return qcow2_update_header(bs);
+}
+
+static int preallocate(BlockDriverState *bs)
+{
+    uint64_t nb_sectors;
+    uint64_t offset;
+    uint64_t host_offset = 0;
+    int num;
+    int ret;
+    QCowL2Meta *meta;
+
+    nb_sectors = bdrv_getlength(bs) >> 9;
+    offset = 0;
+
+    while (nb_sectors) {
+        num = MIN(nb_sectors, INT_MAX >> 9);
+        ret = qcow2_alloc_cluster_offset(bs, offset, 0, num, &num,
+                                         &host_offset, &meta);
+        if (ret < 0) {
+            return ret;
+        }
+
+        ret = qcow2_alloc_cluster_link_l2(bs, meta);
+        if (ret < 0) {
+            qcow2_free_any_clusters(bs, meta->alloc_offset, meta->nb_clusters,
+                                    QCOW2_DISCARD_NEVER);
+            return ret;
+        }
+
+        /* There are no dependent requests, but we need to remove our request
+         * from the list of in-flight requests */
+        if (meta != NULL) {
+            QLIST_REMOVE(meta, next_in_flight);
+        }
+
+        /* TODO Preallocate data if requested */
+
+        nb_sectors -= num;
+        offset += num << 9;
+    }
+
+    /*
+     * It is expected that the image file is large enough to actually contain
+     * all of the allocated clusters (otherwise we get failing reads after
+     * EOF). Extend the image to the last allocated sector.
+     */
+    if (host_offset != 0) {
+        uint8_t buf[512];
+        memset(buf, 0, 512);
+        ret = bdrv_write(bs->file, (host_offset >> 9) + num - 1, buf, 1);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    return 0;
+}
+
+static int qcow2_create2(const char *filename, int64_t total_size,
+                         const char *backing_file, const char *backing_format,
+                         int flags, size_t cluster_size, int prealloc,
+                         QEMUOptionParameter *options, int version)
+{
+    /* Calculate cluster_bits */
+    int cluster_bits;
+    cluster_bits = ffs(cluster_size) - 1;
+    if (cluster_bits < MIN_CLUSTER_BITS || cluster_bits > MAX_CLUSTER_BITS ||
+        (1 << cluster_bits) != cluster_size)
+    {
+        error_report(
+            "Cluster size must be a power of two between %d and %dk",
+            1 << MIN_CLUSTER_BITS, 1 << (MAX_CLUSTER_BITS - 10));
+        return -EINVAL;
+    }
+
+    /*
+     * Open the image file and write a minimal qcow2 header.
+     *
+     * We keep things simple and start with a zero-sized image. We also
+     * do without refcount blocks or a L1 table for now. We'll fix the
+     * inconsistency later.
+     *
+     * We do need a refcount table because growing the refcount table means
+     * allocating two new refcount blocks - the seconds of which would be at
+     * 2 GB for 64k clusters, and we don't want to have a 2 GB initial file
+     * size for any qcow2 image.
+     */
+    BlockDriverState* bs;
+    QCowHeader header;
+    uint8_t* refcount_table;
+    int ret;
+
+    ret = bdrv_create_file(filename, options);
+    if (ret < 0) {
+        return ret;
+    }
+
+    ret = bdrv_file_open(&bs, filename, NULL, BDRV_O_RDWR);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* Write the header */
+    memset(&header, 0, sizeof(header));
+    header.magic = cpu_to_be32(QCOW_MAGIC);
+    header.version = cpu_to_be32(version);
+    header.cluster_bits = cpu_to_be32(cluster_bits);
+    header.size = cpu_to_be64(0);
+    header.l1_table_offset = cpu_to_be64(0);
+    header.l1_size = cpu_to_be32(0);
+    header.refcount_table_offset = cpu_to_be64(cluster_size);
+    header.refcount_table_clusters = cpu_to_be32(1);
+    header.refcount_order = cpu_to_be32(3 + REFCOUNT_SHIFT);
+    header.header_length = cpu_to_be32(sizeof(header));
+
+    if (flags & BLOCK_FLAG_ENCRYPT) {
+        header.crypt_method = cpu_to_be32(QCOW_CRYPT_AES);
+    } else {
+        header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
+    }
+
+    if (flags & BLOCK_FLAG_LAZY_REFCOUNTS) {
+        header.compatible_features |=
+            cpu_to_be64(QCOW2_COMPAT_LAZY_REFCOUNTS);
+    }
+
+    ret = bdrv_pwrite(bs, 0, &header, sizeof(header));
+    if (ret < 0) {
+        goto out;
+    }
+
+    /* Write an empty refcount table */
+    refcount_table = g_malloc0(cluster_size);
+    ret = bdrv_pwrite(bs, cluster_size, refcount_table, cluster_size);
+    g_free(refcount_table);
+
+    if (ret < 0) {
+        goto out;
+    }
+
+    bdrv_close(bs);
+
+    /*
+     * And now open the image and make it consistent first (i.e. increase the
+     * refcount of the cluster that is occupied by the header and the refcount
+     * table)
+     */
+    BlockDriver* drv = bdrv_find_format("qcow2");
+    assert(drv != NULL);
+    ret = bdrv_open(bs, filename, NULL,
+        BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_NO_FLUSH, drv);
+    if (ret < 0) {
+        goto out;
+    }
+
+    ret = qcow2_alloc_clusters(bs, 2 * cluster_size);
+    if (ret < 0) {
+        goto out;
+
+    } else if (ret != 0) {
+        error_report("Huh, first cluster in empty image is already in use?");
+        abort();
+    }
+
+    /* Okay, now that we have a valid image, let's give it the right size */
+    ret = bdrv_truncate(bs, total_size * BDRV_SECTOR_SIZE);
+    if (ret < 0) {
+        goto out;
+    }
+
+    /* Want a backing file? There you go.*/
+    if (backing_file) {
+        ret = bdrv_change_backing_file(bs, backing_file, backing_format);
+        if (ret < 0) {
+            goto out;
+        }
+    }
+
+    /* And if we're supposed to preallocate metadata, do that now */
+    if (prealloc) {
+        BDRVQcowState *s = bs->opaque;
+        qemu_co_mutex_lock(&s->lock);
+        ret = preallocate(bs);
+        qemu_co_mutex_unlock(&s->lock);
+        if (ret < 0) {
+            goto out;
+        }
+    }
+
+    ret = 0;
+out:
+    bdrv_delete(bs);
+    return ret;
+}
+
+static int qcow2_create(const char *filename, QEMUOptionParameter *options)
+{
+    const char *backing_file = NULL;
+    const char *backing_fmt = NULL;
+    uint64_t sectors = 0;
+    int flags = 0;
+    size_t cluster_size = DEFAULT_CLUSTER_SIZE;
+    int prealloc = 0;
+    int version = 2;
+
+    /* Read out options */
+    while (options && options->name) {
+        if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
+            sectors = options->value.n / 512;
+        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
+            backing_file = options->value.s;
+        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FMT)) {
+            backing_fmt = options->value.s;
+        } else if (!strcmp(options->name, BLOCK_OPT_ENCRYPT)) {
+            flags |= options->value.n ? BLOCK_FLAG_ENCRYPT : 0;
+        } else if (!strcmp(options->name, BLOCK_OPT_CLUSTER_SIZE)) {
+            if (options->value.n) {
+                cluster_size = options->value.n;
+            }
+        } else if (!strcmp(options->name, BLOCK_OPT_PREALLOC)) {
+            if (!options->value.s || !strcmp(options->value.s, "off")) {
+                prealloc = 0;
+            } else if (!strcmp(options->value.s, "metadata")) {
+                prealloc = 1;
+            } else {
+                fprintf(stderr, "Invalid preallocation mode: '%s'\n",
+                    options->value.s);
+                return -EINVAL;
+            }
+        } else if (!strcmp(options->name, BLOCK_OPT_COMPAT_LEVEL)) {
+            if (!options->value.s || !strcmp(options->value.s, "0.10")) {
+                version = 2;
+            } else if (!strcmp(options->value.s, "1.1")) {
+                version = 3;
+            } else {
+                fprintf(stderr, "Invalid compatibility level: '%s'\n",
+                    options->value.s);
+                return -EINVAL;
+            }
+        } else if (!strcmp(options->name, BLOCK_OPT_LAZY_REFCOUNTS)) {
+            flags |= options->value.n ? BLOCK_FLAG_LAZY_REFCOUNTS : 0;
+        }
+        options++;
+    }
+
+    if (backing_file && prealloc) {
+        fprintf(stderr, "Backing file and preallocation cannot be used at "
+            "the same time\n");
+        return -EINVAL;
+    }
+
+    if (version < 3 && (flags & BLOCK_FLAG_LAZY_REFCOUNTS)) {
+        fprintf(stderr, "Lazy refcounts only supported with compatibility "
+                "level 1.1 and above (use compat=1.1 or greater)\n");
+        return -EINVAL;
+    }
+
+    return qcow2_create2(filename, sectors, backing_file, backing_fmt, flags,
+                         cluster_size, prealloc, options, version);
+}
+
+static int qcow2_make_empty(BlockDriverState *bs)
+{
+#if 0
+    /* XXX: not correct */
+    BDRVQcowState *s = bs->opaque;
+    uint32_t l1_length = s->l1_size * sizeof(uint64_t);
+    int ret;
+
+    memset(s->l1_table, 0, l1_length);
+    if (bdrv_pwrite(bs->file, s->l1_table_offset, s->l1_table, l1_length) < 0)
+        return -1;
+    ret = bdrv_truncate(bs->file, s->l1_table_offset + l1_length);
+    if (ret < 0)
+        return ret;
+
+    l2_cache_reset(bs);
+#endif
+    return 0;
+}
+
+static coroutine_fn int qcow2_co_write_zeroes(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors)
+{
+    int ret;
+    BDRVQcowState *s = bs->opaque;
+
+    /* Emulate misaligned zero writes */
+    if (sector_num % s->cluster_sectors || nb_sectors % s->cluster_sectors) {
+        return -ENOTSUP;
+    }
+
+    /* Whatever is left can use real zero clusters */
+    qemu_co_mutex_lock(&s->lock);
+    ret = qcow2_zero_clusters(bs, sector_num << BDRV_SECTOR_BITS,
+        nb_sectors);
+    qemu_co_mutex_unlock(&s->lock);
+
+    return ret;
+}
+
+static coroutine_fn int qcow2_co_discard(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors)
+{
+    int ret;
+    BDRVQcowState *s = bs->opaque;
+
+    qemu_co_mutex_lock(&s->lock);
+    ret = qcow2_discard_clusters(bs, sector_num << BDRV_SECTOR_BITS,
+        nb_sectors);
+    qemu_co_mutex_unlock(&s->lock);
+    return ret;
+}
+
+static int qcow2_truncate(BlockDriverState *bs, int64_t offset)
+{
+    BDRVQcowState *s = bs->opaque;
+    int64_t new_l1_size;
+    int ret;
+
+    if (offset & 511) {
+        error_report("The new size must be a multiple of 512");
+        return -EINVAL;
+    }
+
+    /* cannot proceed if image has snapshots */
+    if (s->nb_snapshots) {
+        error_report("Can't resize an image which has snapshots");
+        return -ENOTSUP;
+    }
+
+    /* shrinking is currently not supported */
+    if (offset < bs->total_sectors * 512) {
+        error_report("qcow2 doesn't support shrinking images yet");
+        return -ENOTSUP;
+    }
+
+    new_l1_size = size_to_l1(s, offset);
+    ret = qcow2_grow_l1_table(bs, new_l1_size, true);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* write updated header.size */
+    offset = cpu_to_be64(offset);
+    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, size),
+                           &offset, sizeof(uint64_t));
+    if (ret < 0) {
+        return ret;
+    }
+
+    s->l1_vm_state_index = new_l1_size;
+    return 0;
+}
+
+/* XXX: put compressed sectors first, then all the cluster aligned
+   tables to avoid losing bytes in alignment */
+static int qcow2_write_compressed(BlockDriverState *bs, int64_t sector_num,
+                                  const uint8_t *buf, int nb_sectors)
+{
+    BDRVQcowState *s = bs->opaque;
+    z_stream strm;
+    int ret, out_len;
+    uint8_t *out_buf;
+    uint64_t cluster_offset;
+
+    if (nb_sectors == 0) {
+        /* align end of file to a sector boundary to ease reading with
+           sector based I/Os */
+        cluster_offset = bdrv_getlength(bs->file);
+        cluster_offset = (cluster_offset + 511) & ~511;
+        bdrv_truncate(bs->file, cluster_offset);
+        return 0;
+    }
+
+    if (nb_sectors != s->cluster_sectors) {
+        ret = -EINVAL;
+
+        /* Zero-pad last write if image size is not cluster aligned */
+        if (sector_num + nb_sectors == bs->total_sectors &&
+            nb_sectors < s->cluster_sectors) {
+            uint8_t *pad_buf = qemu_blockalign(bs, s->cluster_size);
+            memset(pad_buf, 0, s->cluster_size);
+            memcpy(pad_buf, buf, nb_sectors * BDRV_SECTOR_SIZE);
+            ret = qcow2_write_compressed(bs, sector_num,
+                                         pad_buf, s->cluster_sectors);
+            qemu_vfree(pad_buf);
+        }
+        return ret;
+    }
+
+    out_buf = g_malloc(s->cluster_size + (s->cluster_size / 1000) + 128);
+
+    /* best compression, small window, no zlib header */
+    memset(&strm, 0, sizeof(strm));
+    ret = deflateInit2(&strm, Z_DEFAULT_COMPRESSION,
+                       Z_DEFLATED, -12,
+                       9, Z_DEFAULT_STRATEGY);
+    if (ret != 0) {
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    strm.avail_in = s->cluster_size;
+    strm.next_in = (uint8_t *)buf;
+    strm.avail_out = s->cluster_size;
+    strm.next_out = out_buf;
+
+    ret = deflate(&strm, Z_FINISH);
+    if (ret != Z_STREAM_END && ret != Z_OK) {
+        deflateEnd(&strm);
+        ret = -EINVAL;
+        goto fail;
+    }
+    out_len = strm.next_out - out_buf;
+
+    deflateEnd(&strm);
+
+    if (ret != Z_STREAM_END || out_len >= s->cluster_size) {
+        /* could not compress: write normal cluster */
+        ret = bdrv_write(bs, sector_num, buf, s->cluster_sectors);
+        if (ret < 0) {
+            goto fail;
+        }
+    } else {
+        cluster_offset = qcow2_alloc_compressed_cluster_offset(bs,
+            sector_num << 9, out_len);
+        if (!cluster_offset) {
+            ret = -EIO;
+            goto fail;
+        }
+        cluster_offset &= s->cluster_offset_mask;
+        BLKDBG_EVENT(bs->file, BLKDBG_WRITE_COMPRESSED);
+        ret = bdrv_pwrite(bs->file, cluster_offset, out_buf, out_len);
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
+    ret = 0;
+fail:
+    g_free(out_buf);
+    return ret;
+}
+
+static coroutine_fn int qcow2_co_flush_to_os(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    int ret;
+
+    qemu_co_mutex_lock(&s->lock);
+    ret = qcow2_cache_flush(bs, s->l2_table_cache);
+    if (ret < 0) {
+        qemu_co_mutex_unlock(&s->lock);
+        return ret;
+    }
+
+    if (qcow2_need_accurate_refcounts(s)) {
+        ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+        if (ret < 0) {
+            qemu_co_mutex_unlock(&s->lock);
+            return ret;
+        }
+    }
+    qemu_co_mutex_unlock(&s->lock);
+
+    return 0;
+}
+
+static int64_t qcow2_vm_state_offset(BDRVQcowState *s)
+{
+	return (int64_t)s->l1_vm_state_index << (s->cluster_bits + s->l2_bits);
+}
+
+static int qcow2_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
+{
+    BDRVQcowState *s = bs->opaque;
+    bdi->cluster_size = s->cluster_size;
+    bdi->vm_state_offset = qcow2_vm_state_offset(s);
+    return 0;
+}
+
+#if 0
+static void dump_refcounts(BlockDriverState *bs)
+{
+    BDRVQcowState *s = bs->opaque;
+    int64_t nb_clusters, k, k1, size;
+    int refcount;
+
+    size = bdrv_getlength(bs->file);
+    nb_clusters = size_to_clusters(s, size);
+    for(k = 0; k < nb_clusters;) {
+        k1 = k;
+        refcount = get_refcount(bs, k);
+        k++;
+        while (k < nb_clusters && get_refcount(bs, k) == refcount)
+            k++;
+        printf("%" PRId64 ": refcount=%d nb=%" PRId64 "\n", k, refcount,
+               k - k1);
+    }
+}
+#endif
+
+static int qcow2_save_vmstate(BlockDriverState *bs, QEMUIOVector *qiov,
+                              int64_t pos)
+{
+    BDRVQcowState *s = bs->opaque;
+    int growable = bs->growable;
+    int ret;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_VMSTATE_SAVE);
+    bs->growable = 1;
+    ret = bdrv_pwritev(bs, qcow2_vm_state_offset(s) + pos, qiov);
+    bs->growable = growable;
+
+    return ret;
+}
+
+static int qcow2_load_vmstate(BlockDriverState *bs, uint8_t *buf,
+                              int64_t pos, int size)
+{
+    BDRVQcowState *s = bs->opaque;
+    int growable = bs->growable;
+    int ret;
+
+    BLKDBG_EVENT(bs->file, BLKDBG_VMSTATE_LOAD);
+    bs->growable = 1;
+    ret = bdrv_pread(bs, qcow2_vm_state_offset(s) + pos, buf, size);
+    bs->growable = growable;
+
+    return ret;
+}
+
+static QEMUOptionParameter qcow2_create_options[] = {
+    {
+        .name = BLOCK_OPT_SIZE,
+        .type = OPT_SIZE,
+        .help = "Virtual disk size"
+    },
+    {
+        .name = BLOCK_OPT_COMPAT_LEVEL,
+        .type = OPT_STRING,
+        .help = "Compatibility level (0.10 or 1.1)"
+    },
+    {
+        .name = BLOCK_OPT_BACKING_FILE,
+        .type = OPT_STRING,
+        .help = "File name of a base image"
+    },
+    {
+        .name = BLOCK_OPT_BACKING_FMT,
+        .type = OPT_STRING,
+        .help = "Image format of the base image"
+    },
+    {
+        .name = BLOCK_OPT_ENCRYPT,
+        .type = OPT_FLAG,
+        .help = "Encrypt the image"
+    },
+    {
+        .name = BLOCK_OPT_CLUSTER_SIZE,
+        .type = OPT_SIZE,
+        .help = "qcow2 cluster size",
+        .value = { .n = DEFAULT_CLUSTER_SIZE },
+    },
+    {
+        .name = BLOCK_OPT_PREALLOC,
+        .type = OPT_STRING,
+        .help = "Preallocation mode (allowed values: off, metadata)"
+    },
+    {
+        .name = BLOCK_OPT_LAZY_REFCOUNTS,
+        .type = OPT_FLAG,
+        .help = "Postpone refcount updates",
+    },
+    { NULL }
+};
+
+static BlockDriver bdrv_qcow2 = {
+    .format_name        = "qcow2",
+    .instance_size      = sizeof(BDRVQcowState),
+    .bdrv_probe         = qcow2_probe,
+    .bdrv_open          = qcow2_open,
+    .bdrv_close         = qcow2_close,
+    .bdrv_reopen_prepare  = qcow2_reopen_prepare,
+    .bdrv_create        = qcow2_create,
+    .bdrv_has_zero_init = bdrv_has_zero_init_1,
+    .bdrv_co_is_allocated = qcow2_co_is_allocated,
+    .bdrv_set_key       = qcow2_set_key,
+    .bdrv_make_empty    = qcow2_make_empty,
+
+    .bdrv_co_readv          = qcow2_co_readv,
+    .bdrv_co_writev         = qcow2_co_writev,
+    .bdrv_co_flush_to_os    = qcow2_co_flush_to_os,
+
+    .bdrv_co_write_zeroes   = qcow2_co_write_zeroes,
+    .bdrv_co_discard        = qcow2_co_discard,
+    .bdrv_truncate          = qcow2_truncate,
+    .bdrv_write_compressed  = qcow2_write_compressed,
+
+    .bdrv_snapshot_create   = qcow2_snapshot_create,
+    .bdrv_snapshot_goto     = qcow2_snapshot_goto,
+    .bdrv_snapshot_delete   = qcow2_snapshot_delete,
+    .bdrv_snapshot_list     = qcow2_snapshot_list,
+    .bdrv_snapshot_load_tmp     = qcow2_snapshot_load_tmp,
+    .bdrv_get_info      = qcow2_get_info,
+
+    .bdrv_save_vmstate    = qcow2_save_vmstate,
+    .bdrv_load_vmstate    = qcow2_load_vmstate,
+
+    .bdrv_change_backing_file   = qcow2_change_backing_file,
+
+    .bdrv_invalidate_cache      = qcow2_invalidate_cache,
+
+    .create_options = qcow2_create_options,
+    .bdrv_check = qcow2_check,
+};
+
+static void bdrv_qcow2_init(void)
+{
+    bdrv_register(&bdrv_qcow2);
+}
+
+block_init(bdrv_qcow2_init);
diff --git a/contrib/qemu/block/qcow2.h b/contrib/qemu/block/qcow2.h
new file mode 100644
index 000000000..3b2d5cda7
--- /dev/null
+++ b/contrib/qemu/block/qcow2.h
@@ -0,0 +1,437 @@
+/*
+ * Block driver for the QCOW version 2 format
+ *
+ * Copyright (c) 2004-2006 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef BLOCK_QCOW2_H
+#define BLOCK_QCOW2_H
+
+#include "qemu/aes.h"
+#include "block/coroutine.h"
+
+//#define DEBUG_ALLOC
+//#define DEBUG_ALLOC2
+//#define DEBUG_EXT
+
+#define QCOW_MAGIC (('Q' << 24) | ('F' << 16) | ('I' << 8) | 0xfb)
+
+#define QCOW_CRYPT_NONE 0
+#define QCOW_CRYPT_AES  1
+
+#define QCOW_MAX_CRYPT_CLUSTERS 32
+
+/* indicate that the refcount of the referenced cluster is exactly one. */
+#define QCOW_OFLAG_COPIED     (1LL << 63)
+/* indicate that the cluster is compressed (they never have the copied flag) */
+#define QCOW_OFLAG_COMPRESSED (1LL << 62)
+/* The cluster reads as all zeros */
+#define QCOW_OFLAG_ZERO (1LL << 0)
+
+#define REFCOUNT_SHIFT 1 /* refcount size is 2 bytes */
+
+#define MIN_CLUSTER_BITS 9
+#define MAX_CLUSTER_BITS 21
+
+#define L2_CACHE_SIZE 16
+
+/* Must be at least 4 to cover all cases of refcount table growth */
+#define REFCOUNT_CACHE_SIZE 4
+
+#define DEFAULT_CLUSTER_SIZE 65536
+
+
+#define QCOW2_OPT_LAZY_REFCOUNTS "lazy_refcounts"
+#define QCOW2_OPT_DISCARD_REQUEST "pass_discard_request"
+#define QCOW2_OPT_DISCARD_SNAPSHOT "pass_discard_snapshot"
+#define QCOW2_OPT_DISCARD_OTHER "pass_discard_other"
+
+typedef struct QCowHeader {
+    uint32_t magic;
+    uint32_t version;
+    uint64_t backing_file_offset;
+    uint32_t backing_file_size;
+    uint32_t cluster_bits;
+    uint64_t size; /* in bytes */
+    uint32_t crypt_method;
+    uint32_t l1_size; /* XXX: save number of clusters instead ? */
+    uint64_t l1_table_offset;
+    uint64_t refcount_table_offset;
+    uint32_t refcount_table_clusters;
+    uint32_t nb_snapshots;
+    uint64_t snapshots_offset;
+
+    /* The following fields are only valid for version >= 3 */
+    uint64_t incompatible_features;
+    uint64_t compatible_features;
+    uint64_t autoclear_features;
+
+    uint32_t refcount_order;
+    uint32_t header_length;
+} QCowHeader;
+
+typedef struct QCowSnapshot {
+    uint64_t l1_table_offset;
+    uint32_t l1_size;
+    char *id_str;
+    char *name;
+    uint64_t disk_size;
+    uint64_t vm_state_size;
+    uint32_t date_sec;
+    uint32_t date_nsec;
+    uint64_t vm_clock_nsec;
+} QCowSnapshot;
+
+struct Qcow2Cache;
+typedef struct Qcow2Cache Qcow2Cache;
+
+typedef struct Qcow2UnknownHeaderExtension {
+    uint32_t magic;
+    uint32_t len;
+    QLIST_ENTRY(Qcow2UnknownHeaderExtension) next;
+    uint8_t data[];
+} Qcow2UnknownHeaderExtension;
+
+enum {
+    QCOW2_FEAT_TYPE_INCOMPATIBLE    = 0,
+    QCOW2_FEAT_TYPE_COMPATIBLE      = 1,
+    QCOW2_FEAT_TYPE_AUTOCLEAR       = 2,
+};
+
+/* Incompatible feature bits */
+enum {
+    QCOW2_INCOMPAT_DIRTY_BITNR   = 0,
+    QCOW2_INCOMPAT_DIRTY         = 1 << QCOW2_INCOMPAT_DIRTY_BITNR,
+
+    QCOW2_INCOMPAT_MASK          = QCOW2_INCOMPAT_DIRTY,
+};
+
+/* Compatible feature bits */
+enum {
+    QCOW2_COMPAT_LAZY_REFCOUNTS_BITNR = 0,
+    QCOW2_COMPAT_LAZY_REFCOUNTS       = 1 << QCOW2_COMPAT_LAZY_REFCOUNTS_BITNR,
+
+    QCOW2_COMPAT_FEAT_MASK            = QCOW2_COMPAT_LAZY_REFCOUNTS,
+};
+
+enum qcow2_discard_type {
+    QCOW2_DISCARD_NEVER = 0,
+    QCOW2_DISCARD_ALWAYS,
+    QCOW2_DISCARD_REQUEST,
+    QCOW2_DISCARD_SNAPSHOT,
+    QCOW2_DISCARD_OTHER,
+    QCOW2_DISCARD_MAX
+};
+
+typedef struct Qcow2Feature {
+    uint8_t type;
+    uint8_t bit;
+    char    name[46];
+} QEMU_PACKED Qcow2Feature;
+
+typedef struct Qcow2DiscardRegion {
+    BlockDriverState *bs;
+    uint64_t offset;
+    uint64_t bytes;
+    QTAILQ_ENTRY(Qcow2DiscardRegion) next;
+} Qcow2DiscardRegion;
+
+typedef struct BDRVQcowState {
+    int cluster_bits;
+    int cluster_size;
+    int cluster_sectors;
+    int l2_bits;
+    int l2_size;
+    int l1_size;
+    int l1_vm_state_index;
+    int csize_shift;
+    int csize_mask;
+    uint64_t cluster_offset_mask;
+    uint64_t l1_table_offset;
+    uint64_t *l1_table;
+
+    Qcow2Cache* l2_table_cache;
+    Qcow2Cache* refcount_block_cache;
+
+    uint8_t *cluster_cache;
+    uint8_t *cluster_data;
+    uint64_t cluster_cache_offset;
+    QLIST_HEAD(QCowClusterAlloc, QCowL2Meta) cluster_allocs;
+
+    uint64_t *refcount_table;
+    uint64_t refcount_table_offset;
+    uint32_t refcount_table_size;
+    int64_t free_cluster_index;
+    int64_t free_byte_offset;
+
+    CoMutex lock;
+
+    uint32_t crypt_method; /* current crypt method, 0 if no key yet */
+    uint32_t crypt_method_header;
+    AES_KEY aes_encrypt_key;
+    AES_KEY aes_decrypt_key;
+    uint64_t snapshots_offset;
+    int snapshots_size;
+    int nb_snapshots;
+    QCowSnapshot *snapshots;
+
+    int flags;
+    int qcow_version;
+    bool use_lazy_refcounts;
+
+    bool discard_passthrough[QCOW2_DISCARD_MAX];
+
+    uint64_t incompatible_features;
+    uint64_t compatible_features;
+    uint64_t autoclear_features;
+
+    size_t unknown_header_fields_size;
+    void* unknown_header_fields;
+    QLIST_HEAD(, Qcow2UnknownHeaderExtension) unknown_header_ext;
+    QTAILQ_HEAD (, Qcow2DiscardRegion) discards;
+    bool cache_discards;
+} BDRVQcowState;
+
+/* XXX: use std qcow open function ? */
+typedef struct QCowCreateState {
+    int cluster_size;
+    int cluster_bits;
+    uint16_t *refcount_block;
+    uint64_t *refcount_table;
+    int64_t l1_table_offset;
+    int64_t refcount_table_offset;
+    int64_t refcount_block_offset;
+} QCowCreateState;
+
+struct QCowAIOCB;
+
+typedef struct Qcow2COWRegion {
+    /**
+     * Offset of the COW region in bytes from the start of the first cluster
+     * touched by the request.
+     */
+    uint64_t    offset;
+
+    /** Number of sectors to copy */
+    int         nb_sectors;
+} Qcow2COWRegion;
+
+/**
+ * Describes an in-flight (part of a) write request that writes to clusters
+ * that are not referenced in their L2 table yet.
+ */
+typedef struct QCowL2Meta
+{
+    /** Guest offset of the first newly allocated cluster */
+    uint64_t offset;
+
+    /** Host offset of the first newly allocated cluster */
+    uint64_t alloc_offset;
+
+    /**
+     * Number of sectors from the start of the first allocated cluster to
+     * the end of the (possibly shortened) request
+     */
+    int nb_available;
+
+    /** Number of newly allocated clusters */
+    int nb_clusters;
+
+    /**
+     * Requests that overlap with this allocation and wait to be restarted
+     * when the allocating request has completed.
+     */
+    CoQueue dependent_requests;
+
+    /**
+     * The COW Region between the start of the first allocated cluster and the
+     * area the guest actually writes to.
+     */
+    Qcow2COWRegion cow_start;
+
+    /**
+     * The COW Region between the area the guest actually writes to and the
+     * end of the last allocated cluster.
+     */
+    Qcow2COWRegion cow_end;
+
+    /** Pointer to next L2Meta of the same write request */
+    struct QCowL2Meta *next;
+
+    QLIST_ENTRY(QCowL2Meta) next_in_flight;
+} QCowL2Meta;
+
+enum {
+    QCOW2_CLUSTER_UNALLOCATED,
+    QCOW2_CLUSTER_NORMAL,
+    QCOW2_CLUSTER_COMPRESSED,
+    QCOW2_CLUSTER_ZERO
+};
+
+#define L1E_OFFSET_MASK 0x00ffffffffffff00ULL
+#define L2E_OFFSET_MASK 0x00ffffffffffff00ULL
+#define L2E_COMPRESSED_OFFSET_SIZE_MASK 0x3fffffffffffffffULL
+
+#define REFT_OFFSET_MASK 0xffffffffffffff00ULL
+
+static inline int64_t start_of_cluster(BDRVQcowState *s, int64_t offset)
+{
+    return offset & ~(s->cluster_size - 1);
+}
+
+static inline int64_t offset_into_cluster(BDRVQcowState *s, int64_t offset)
+{
+    return offset & (s->cluster_size - 1);
+}
+
+static inline int size_to_clusters(BDRVQcowState *s, int64_t size)
+{
+    return (size + (s->cluster_size - 1)) >> s->cluster_bits;
+}
+
+static inline int64_t size_to_l1(BDRVQcowState *s, int64_t size)
+{
+    int shift = s->cluster_bits + s->l2_bits;
+    return (size + (1ULL << shift) - 1) >> shift;
+}
+
+static inline int offset_to_l2_index(BDRVQcowState *s, int64_t offset)
+{
+    return (offset >> s->cluster_bits) & (s->l2_size - 1);
+}
+
+static inline int64_t align_offset(int64_t offset, int n)
+{
+    offset = (offset + n - 1) & ~(n - 1);
+    return offset;
+}
+
+static inline int qcow2_get_cluster_type(uint64_t l2_entry)
+{
+    if (l2_entry & QCOW_OFLAG_COMPRESSED) {
+        return QCOW2_CLUSTER_COMPRESSED;
+    } else if (l2_entry & QCOW_OFLAG_ZERO) {
+        return QCOW2_CLUSTER_ZERO;
+    } else if (!(l2_entry & L2E_OFFSET_MASK)) {
+        return QCOW2_CLUSTER_UNALLOCATED;
+    } else {
+        return QCOW2_CLUSTER_NORMAL;
+    }
+}
+
+/* Check whether refcounts are eager or lazy */
+static inline bool qcow2_need_accurate_refcounts(BDRVQcowState *s)
+{
+    return !(s->incompatible_features & QCOW2_INCOMPAT_DIRTY);
+}
+
+static inline uint64_t l2meta_cow_start(QCowL2Meta *m)
+{
+    return m->offset + m->cow_start.offset;
+}
+
+static inline uint64_t l2meta_cow_end(QCowL2Meta *m)
+{
+    return m->offset + m->cow_end.offset
+        + (m->cow_end.nb_sectors << BDRV_SECTOR_BITS);
+}
+
+// FIXME Need qcow2_ prefix to global functions
+
+/* qcow2.c functions */
+int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov,
+                  int64_t sector_num, int nb_sectors);
+
+int qcow2_mark_dirty(BlockDriverState *bs);
+int qcow2_update_header(BlockDriverState *bs);
+
+/* qcow2-refcount.c functions */
+int qcow2_refcount_init(BlockDriverState *bs);
+void qcow2_refcount_close(BlockDriverState *bs);
+
+int64_t qcow2_alloc_clusters(BlockDriverState *bs, int64_t size);
+int qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
+    int nb_clusters);
+int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size);
+void qcow2_free_clusters(BlockDriverState *bs,
+                          int64_t offset, int64_t size,
+                          enum qcow2_discard_type type);
+void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
+                             int nb_clusters, enum qcow2_discard_type type);
+
+int qcow2_update_snapshot_refcount(BlockDriverState *bs,
+    int64_t l1_table_offset, int l1_size, int addend);
+
+int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
+                          BdrvCheckMode fix);
+
+void qcow2_process_discards(BlockDriverState *bs, int ret);
+
+/* qcow2-cluster.c functions */
+int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
+                        bool exact_size);
+void qcow2_l2_cache_reset(BlockDriverState *bs);
+int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset);
+void qcow2_encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
+                     uint8_t *out_buf, const uint8_t *in_buf,
+                     int nb_sectors, int enc,
+                     const AES_KEY *key);
+
+int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
+    int *num, uint64_t *cluster_offset);
+int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
+    int n_start, int n_end, int *num, uint64_t *host_offset, QCowL2Meta **m);
+uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
+                                         uint64_t offset,
+                                         int compressed_size);
+
+int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m);
+int qcow2_discard_clusters(BlockDriverState *bs, uint64_t offset,
+    int nb_sectors);
+int qcow2_zero_clusters(BlockDriverState *bs, uint64_t offset, int nb_sectors);
+
+/* qcow2-snapshot.c functions */
+int qcow2_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info);
+int qcow2_snapshot_goto(BlockDriverState *bs, const char *snapshot_id);
+int qcow2_snapshot_delete(BlockDriverState *bs, const char *snapshot_id);
+int qcow2_snapshot_list(BlockDriverState *bs, QEMUSnapshotInfo **psn_tab);
+int qcow2_snapshot_load_tmp(BlockDriverState *bs, const char *snapshot_name);
+
+void qcow2_free_snapshots(BlockDriverState *bs);
+int qcow2_read_snapshots(BlockDriverState *bs);
+
+/* qcow2-cache.c functions */
+Qcow2Cache *qcow2_cache_create(BlockDriverState *bs, int num_tables);
+int qcow2_cache_destroy(BlockDriverState* bs, Qcow2Cache *c);
+
+void qcow2_cache_entry_mark_dirty(Qcow2Cache *c, void *table);
+int qcow2_cache_flush(BlockDriverState *bs, Qcow2Cache *c);
+int qcow2_cache_set_dependency(BlockDriverState *bs, Qcow2Cache *c,
+    Qcow2Cache *dependency);
+void qcow2_cache_depends_on_flush(Qcow2Cache *c);
+
+int qcow2_cache_get(BlockDriverState *bs, Qcow2Cache *c, uint64_t offset,
+    void **table);
+int qcow2_cache_get_empty(BlockDriverState *bs, Qcow2Cache *c, uint64_t offset,
+    void **table);
+int qcow2_cache_put(BlockDriverState *bs, Qcow2Cache *c, void **table);
+
+#endif
diff --git a/contrib/qemu/block/qed-check.c b/contrib/qemu/block/qed-check.c
new file mode 100644
index 000000000..b473dcd61
--- /dev/null
+++ b/contrib/qemu/block/qed-check.c
@@ -0,0 +1,248 @@
+/*
+ * QEMU Enhanced Disk Format Consistency Check
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#include "qed.h"
+
+typedef struct {
+    BDRVQEDState *s;
+    BdrvCheckResult *result;
+    bool fix;                           /* whether to fix invalid offsets */
+
+    uint64_t nclusters;
+    uint32_t *used_clusters;            /* referenced cluster bitmap */
+
+    QEDRequest request;
+} QEDCheck;
+
+static bool qed_test_bit(uint32_t *bitmap, uint64_t n) {
+    return !!(bitmap[n / 32] & (1 << (n % 32)));
+}
+
+static void qed_set_bit(uint32_t *bitmap, uint64_t n) {
+    bitmap[n / 32] |= 1 << (n % 32);
+}
+
+/**
+ * Set bitmap bits for clusters
+ *
+ * @check:          Check structure
+ * @offset:         Starting offset in bytes
+ * @n:              Number of clusters
+ */
+static bool qed_set_used_clusters(QEDCheck *check, uint64_t offset,
+                                  unsigned int n)
+{
+    uint64_t cluster = qed_bytes_to_clusters(check->s, offset);
+    unsigned int corruptions = 0;
+
+    while (n-- != 0) {
+        /* Clusters should only be referenced once */
+        if (qed_test_bit(check->used_clusters, cluster)) {
+            corruptions++;
+        }
+
+        qed_set_bit(check->used_clusters, cluster);
+        cluster++;
+    }
+
+    check->result->corruptions += corruptions;
+    return corruptions == 0;
+}
+
+/**
+ * Check an L2 table
+ *
+ * @ret:            Number of invalid cluster offsets
+ */
+static unsigned int qed_check_l2_table(QEDCheck *check, QEDTable *table)
+{
+    BDRVQEDState *s = check->s;
+    unsigned int i, num_invalid = 0;
+    uint64_t last_offset = 0;
+
+    for (i = 0; i < s->table_nelems; i++) {
+        uint64_t offset = table->offsets[i];
+
+        if (qed_offset_is_unalloc_cluster(offset) ||
+            qed_offset_is_zero_cluster(offset)) {
+            continue;
+        }
+        check->result->bfi.allocated_clusters++;
+        if (last_offset && (last_offset + s->header.cluster_size != offset)) {
+            check->result->bfi.fragmented_clusters++;
+        }
+        last_offset = offset;
+
+        /* Detect invalid cluster offset */
+        if (!qed_check_cluster_offset(s, offset)) {
+            if (check->fix) {
+                table->offsets[i] = 0;
+                check->result->corruptions_fixed++;
+            } else {
+                check->result->corruptions++;
+            }
+
+            num_invalid++;
+            continue;
+        }
+
+        qed_set_used_clusters(check, offset, 1);
+    }
+
+    return num_invalid;
+}
+
+/**
+ * Descend tables and check each cluster is referenced once only
+ */
+static int qed_check_l1_table(QEDCheck *check, QEDTable *table)
+{
+    BDRVQEDState *s = check->s;
+    unsigned int i, num_invalid_l1 = 0;
+    int ret, last_error = 0;
+
+    /* Mark L1 table clusters used */
+    qed_set_used_clusters(check, s->header.l1_table_offset,
+                          s->header.table_size);
+
+    for (i = 0; i < s->table_nelems; i++) {
+        unsigned int num_invalid_l2;
+        uint64_t offset = table->offsets[i];
+
+        if (qed_offset_is_unalloc_cluster(offset)) {
+            continue;
+        }
+
+        /* Detect invalid L2 offset */
+        if (!qed_check_table_offset(s, offset)) {
+            /* Clear invalid offset */
+            if (check->fix) {
+                table->offsets[i] = 0;
+                check->result->corruptions_fixed++;
+            } else {
+                check->result->corruptions++;
+            }
+
+            num_invalid_l1++;
+            continue;
+        }
+
+        if (!qed_set_used_clusters(check, offset, s->header.table_size)) {
+            continue; /* skip an invalid table */
+        }
+
+        ret = qed_read_l2_table_sync(s, &check->request, offset);
+        if (ret) {
+            check->result->check_errors++;
+            last_error = ret;
+            continue;
+        }
+
+        num_invalid_l2 = qed_check_l2_table(check,
+                                            check->request.l2_table->table);
+
+        /* Write out fixed L2 table */
+        if (num_invalid_l2 > 0 && check->fix) {
+            ret = qed_write_l2_table_sync(s, &check->request, 0,
+                                          s->table_nelems, false);
+            if (ret) {
+                check->result->check_errors++;
+                last_error = ret;
+                continue;
+            }
+        }
+    }
+
+    /* Drop reference to final table */
+    qed_unref_l2_cache_entry(check->request.l2_table);
+    check->request.l2_table = NULL;
+
+    /* Write out fixed L1 table */
+    if (num_invalid_l1 > 0 && check->fix) {
+        ret = qed_write_l1_table_sync(s, 0, s->table_nelems);
+        if (ret) {
+            check->result->check_errors++;
+            last_error = ret;
+        }
+    }
+
+    return last_error;
+}
+
+/**
+ * Check for unreferenced (leaked) clusters
+ */
+static void qed_check_for_leaks(QEDCheck *check)
+{
+    BDRVQEDState *s = check->s;
+    uint64_t i;
+
+    for (i = s->header.header_size; i < check->nclusters; i++) {
+        if (!qed_test_bit(check->used_clusters, i)) {
+            check->result->leaks++;
+        }
+    }
+}
+
+/**
+ * Mark an image clean once it passes check or has been repaired
+ */
+static void qed_check_mark_clean(BDRVQEDState *s, BdrvCheckResult *result)
+{
+    /* Skip if there were unfixable corruptions or I/O errors */
+    if (result->corruptions > 0 || result->check_errors > 0) {
+        return;
+    }
+
+    /* Skip if image is already marked clean */
+    if (!(s->header.features & QED_F_NEED_CHECK)) {
+        return;
+    }
+
+    /* Ensure fixes reach storage before clearing check bit */
+    bdrv_flush(s->bs);
+
+    s->header.features &= ~QED_F_NEED_CHECK;
+    qed_write_header_sync(s);
+}
+
+int qed_check(BDRVQEDState *s, BdrvCheckResult *result, bool fix)
+{
+    QEDCheck check = {
+        .s = s,
+        .result = result,
+        .nclusters = qed_bytes_to_clusters(s, s->file_size),
+        .request = { .l2_table = NULL },
+        .fix = fix,
+    };
+    int ret;
+
+    check.used_clusters = g_malloc0(((check.nclusters + 31) / 32) *
+                                       sizeof(check.used_clusters[0]));
+
+    check.result->bfi.total_clusters =
+        (s->header.image_size + s->header.cluster_size - 1) /
+            s->header.cluster_size;
+    ret = qed_check_l1_table(&check, s->l1_table);
+    if (ret == 0) {
+        /* Only check for leaks if entire image was scanned successfully */
+        qed_check_for_leaks(&check);
+
+        if (fix) {
+            qed_check_mark_clean(s, result);
+        }
+    }
+
+    g_free(check.used_clusters);
+    return ret;
+}
diff --git a/contrib/qemu/block/qed-cluster.c b/contrib/qemu/block/qed-cluster.c
new file mode 100644
index 000000000..f64b2af8f
--- /dev/null
+++ b/contrib/qemu/block/qed-cluster.c
@@ -0,0 +1,165 @@
+/*
+ * QEMU Enhanced Disk Format Cluster functions
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#include "qed.h"
+
+/**
+ * Count the number of contiguous data clusters
+ *
+ * @s:              QED state
+ * @table:          L2 table
+ * @index:          First cluster index
+ * @n:              Maximum number of clusters
+ * @offset:         Set to first cluster offset
+ *
+ * This function scans tables for contiguous clusters.  A contiguous run of
+ * clusters may be allocated, unallocated, or zero.
+ */
+static unsigned int qed_count_contiguous_clusters(BDRVQEDState *s,
+                                                  QEDTable *table,
+                                                  unsigned int index,
+                                                  unsigned int n,
+                                                  uint64_t *offset)
+{
+    unsigned int end = MIN(index + n, s->table_nelems);
+    uint64_t last = table->offsets[index];
+    unsigned int i;
+
+    *offset = last;
+
+    for (i = index + 1; i < end; i++) {
+        if (qed_offset_is_unalloc_cluster(last)) {
+            /* Counting unallocated clusters */
+            if (!qed_offset_is_unalloc_cluster(table->offsets[i])) {
+                break;
+            }
+        } else if (qed_offset_is_zero_cluster(last)) {
+            /* Counting zero clusters */
+            if (!qed_offset_is_zero_cluster(table->offsets[i])) {
+                break;
+            }
+        } else {
+            /* Counting allocated clusters */
+            if (table->offsets[i] != last + s->header.cluster_size) {
+                break;
+            }
+            last = table->offsets[i];
+        }
+    }
+    return i - index;
+}
+
+typedef struct {
+    BDRVQEDState *s;
+    uint64_t pos;
+    size_t len;
+
+    QEDRequest *request;
+
+    /* User callback */
+    QEDFindClusterFunc *cb;
+    void *opaque;
+} QEDFindClusterCB;
+
+static void qed_find_cluster_cb(void *opaque, int ret)
+{
+    QEDFindClusterCB *find_cluster_cb = opaque;
+    BDRVQEDState *s = find_cluster_cb->s;
+    QEDRequest *request = find_cluster_cb->request;
+    uint64_t offset = 0;
+    size_t len = 0;
+    unsigned int index;
+    unsigned int n;
+
+    if (ret) {
+        goto out;
+    }
+
+    index = qed_l2_index(s, find_cluster_cb->pos);
+    n = qed_bytes_to_clusters(s,
+                              qed_offset_into_cluster(s, find_cluster_cb->pos) +
+                              find_cluster_cb->len);
+    n = qed_count_contiguous_clusters(s, request->l2_table->table,
+                                      index, n, &offset);
+
+    if (qed_offset_is_unalloc_cluster(offset)) {
+        ret = QED_CLUSTER_L2;
+    } else if (qed_offset_is_zero_cluster(offset)) {
+        ret = QED_CLUSTER_ZERO;
+    } else if (qed_check_cluster_offset(s, offset)) {
+        ret = QED_CLUSTER_FOUND;
+    } else {
+        ret = -EINVAL;
+    }
+
+    len = MIN(find_cluster_cb->len, n * s->header.cluster_size -
+              qed_offset_into_cluster(s, find_cluster_cb->pos));
+
+out:
+    find_cluster_cb->cb(find_cluster_cb->opaque, ret, offset, len);
+    g_free(find_cluster_cb);
+}
+
+/**
+ * Find the offset of a data cluster
+ *
+ * @s:          QED state
+ * @request:    L2 cache entry
+ * @pos:        Byte position in device
+ * @len:        Number of bytes
+ * @cb:         Completion function
+ * @opaque:     User data for completion function
+ *
+ * This function translates a position in the block device to an offset in the
+ * image file.  It invokes the cb completion callback to report back the
+ * translated offset or unallocated range in the image file.
+ *
+ * If the L2 table exists, request->l2_table points to the L2 table cache entry
+ * and the caller must free the reference when they are finished.  The cache
+ * entry is exposed in this way to avoid callers having to read the L2 table
+ * again later during request processing.  If request->l2_table is non-NULL it
+ * will be unreferenced before taking on the new cache entry.
+ */
+void qed_find_cluster(BDRVQEDState *s, QEDRequest *request, uint64_t pos,
+                      size_t len, QEDFindClusterFunc *cb, void *opaque)
+{
+    QEDFindClusterCB *find_cluster_cb;
+    uint64_t l2_offset;
+
+    /* Limit length to L2 boundary.  Requests are broken up at the L2 boundary
+     * so that a request acts on one L2 table at a time.
+     */
+    len = MIN(len, (((pos >> s->l1_shift) + 1) << s->l1_shift) - pos);
+
+    l2_offset = s->l1_table->offsets[qed_l1_index(s, pos)];
+    if (qed_offset_is_unalloc_cluster(l2_offset)) {
+        cb(opaque, QED_CLUSTER_L1, 0, len);
+        return;
+    }
+    if (!qed_check_table_offset(s, l2_offset)) {
+        cb(opaque, -EINVAL, 0, 0);
+        return;
+    }
+
+    find_cluster_cb = g_malloc(sizeof(*find_cluster_cb));
+    find_cluster_cb->s = s;
+    find_cluster_cb->pos = pos;
+    find_cluster_cb->len = len;
+    find_cluster_cb->cb = cb;
+    find_cluster_cb->opaque = opaque;
+    find_cluster_cb->request = request;
+
+    qed_read_l2_table(s, request, l2_offset,
+                      qed_find_cluster_cb, find_cluster_cb);
+}
diff --git a/contrib/qemu/block/qed-gencb.c b/contrib/qemu/block/qed-gencb.c
new file mode 100644
index 000000000..7d7ac1ffc
--- /dev/null
+++ b/contrib/qemu/block/qed-gencb.c
@@ -0,0 +1,32 @@
+/*
+ * QEMU Enhanced Disk Format
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#include "qed.h"
+
+void *gencb_alloc(size_t len, BlockDriverCompletionFunc *cb, void *opaque)
+{
+    GenericCB *gencb = g_malloc(len);
+    gencb->cb = cb;
+    gencb->opaque = opaque;
+    return gencb;
+}
+
+void gencb_complete(void *opaque, int ret)
+{
+    GenericCB *gencb = opaque;
+    BlockDriverCompletionFunc *cb = gencb->cb;
+    void *user_opaque = gencb->opaque;
+
+    g_free(gencb);
+    cb(user_opaque, ret);
+}
diff --git a/contrib/qemu/block/qed-l2-cache.c b/contrib/qemu/block/qed-l2-cache.c
new file mode 100644
index 000000000..e9b2aae44
--- /dev/null
+++ b/contrib/qemu/block/qed-l2-cache.c
@@ -0,0 +1,187 @@
+/*
+ * QEMU Enhanced Disk Format L2 Cache
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+/*
+ * L2 table cache usage is as follows:
+ *
+ * An open image has one L2 table cache that is used to avoid accessing the
+ * image file for recently referenced L2 tables.
+ *
+ * Cluster offset lookup translates the logical offset within the block device
+ * to a cluster offset within the image file.  This is done by indexing into
+ * the L1 and L2 tables which store cluster offsets.  It is here where the L2
+ * table cache serves up recently referenced L2 tables.
+ *
+ * If there is a cache miss, that L2 table is read from the image file and
+ * committed to the cache.  Subsequent accesses to that L2 table will be served
+ * from the cache until the table is evicted from the cache.
+ *
+ * L2 tables are also committed to the cache when new L2 tables are allocated
+ * in the image file.  Since the L2 table cache is write-through, the new L2
+ * table is first written out to the image file and then committed to the
+ * cache.
+ *
+ * Multiple I/O requests may be using an L2 table cache entry at any given
+ * time.  That means an entry may be in use across several requests and
+ * reference counting is needed to free the entry at the correct time.  In
+ * particular, an entry evicted from the cache will only be freed once all
+ * references are dropped.
+ *
+ * An in-flight I/O request will hold a reference to a L2 table cache entry for
+ * the period during which it needs to access the L2 table.  This includes
+ * cluster offset lookup, L2 table allocation, and L2 table update when a new
+ * data cluster has been allocated.
+ *
+ * An interesting case occurs when two requests need to access an L2 table that
+ * is not in the cache.  Since the operation to read the table from the image
+ * file takes some time to complete, both requests may see a cache miss and
+ * start reading the L2 table from the image file.  The first to finish will
+ * commit its L2 table into the cache.  When the second tries to commit its
+ * table will be deleted in favor of the existing cache entry.
+ */
+
+#include "trace.h"
+#include "qed.h"
+
+/* Each L2 holds 2GB so this let's us fully cache a 100GB disk */
+#define MAX_L2_CACHE_SIZE 50
+
+/**
+ * Initialize the L2 cache
+ */
+void qed_init_l2_cache(L2TableCache *l2_cache)
+{
+    QTAILQ_INIT(&l2_cache->entries);
+    l2_cache->n_entries = 0;
+}
+
+/**
+ * Free the L2 cache
+ */
+void qed_free_l2_cache(L2TableCache *l2_cache)
+{
+    CachedL2Table *entry, *next_entry;
+
+    QTAILQ_FOREACH_SAFE(entry, &l2_cache->entries, node, next_entry) {
+        qemu_vfree(entry->table);
+        g_free(entry);
+    }
+}
+
+/**
+ * Allocate an uninitialized entry from the cache
+ *
+ * The returned entry has a reference count of 1 and is owned by the caller.
+ * The caller must allocate the actual table field for this entry and it must
+ * be freeable using qemu_vfree().
+ */
+CachedL2Table *qed_alloc_l2_cache_entry(L2TableCache *l2_cache)
+{
+    CachedL2Table *entry;
+
+    entry = g_malloc0(sizeof(*entry));
+    entry->ref++;
+
+    trace_qed_alloc_l2_cache_entry(l2_cache, entry);
+
+    return entry;
+}
+
+/**
+ * Decrease an entry's reference count and free if necessary when the reference
+ * count drops to zero.
+ */
+void qed_unref_l2_cache_entry(CachedL2Table *entry)
+{
+    if (!entry) {
+        return;
+    }
+
+    entry->ref--;
+    trace_qed_unref_l2_cache_entry(entry, entry->ref);
+    if (entry->ref == 0) {
+        qemu_vfree(entry->table);
+        g_free(entry);
+    }
+}
+
+/**
+ * Find an entry in the L2 cache.  This may return NULL and it's up to the
+ * caller to satisfy the cache miss.
+ *
+ * For a cached entry, this function increases the reference count and returns
+ * the entry.
+ */
+CachedL2Table *qed_find_l2_cache_entry(L2TableCache *l2_cache, uint64_t offset)
+{
+    CachedL2Table *entry;
+
+    QTAILQ_FOREACH(entry, &l2_cache->entries, node) {
+        if (entry->offset == offset) {
+            trace_qed_find_l2_cache_entry(l2_cache, entry, offset, entry->ref);
+            entry->ref++;
+            return entry;
+        }
+    }
+    return NULL;
+}
+
+/**
+ * Commit an L2 cache entry into the cache.  This is meant to be used as part of
+ * the process to satisfy a cache miss.  A caller would allocate an entry which
+ * is not actually in the L2 cache and then once the entry was valid and
+ * present on disk, the entry can be committed into the cache.
+ *
+ * Since the cache is write-through, it's important that this function is not
+ * called until the entry is present on disk and the L1 has been updated to
+ * point to the entry.
+ *
+ * N.B. This function steals a reference to the l2_table from the caller so the
+ * caller must obtain a new reference by issuing a call to
+ * qed_find_l2_cache_entry().
+ */
+void qed_commit_l2_cache_entry(L2TableCache *l2_cache, CachedL2Table *l2_table)
+{
+    CachedL2Table *entry;
+
+    entry = qed_find_l2_cache_entry(l2_cache, l2_table->offset);
+    if (entry) {
+        qed_unref_l2_cache_entry(entry);
+        qed_unref_l2_cache_entry(l2_table);
+        return;
+    }
+
+    /* Evict an unused cache entry so we have space.  If all entries are in use
+     * we can grow the cache temporarily and we try to shrink back down later.
+     */
+    if (l2_cache->n_entries >= MAX_L2_CACHE_SIZE) {
+        CachedL2Table *next;
+        QTAILQ_FOREACH_SAFE(entry, &l2_cache->entries, node, next) {
+            if (entry->ref > 1) {
+                continue;
+            }
+
+            QTAILQ_REMOVE(&l2_cache->entries, entry, node);
+            l2_cache->n_entries--;
+            qed_unref_l2_cache_entry(entry);
+
+            /* Stop evicting when we've shrunk back to max size */
+            if (l2_cache->n_entries < MAX_L2_CACHE_SIZE) {
+                break;
+            }
+        }
+    }
+
+    l2_cache->n_entries++;
+    QTAILQ_INSERT_TAIL(&l2_cache->entries, l2_table, node);
+}
diff --git a/contrib/qemu/block/qed-table.c b/contrib/qemu/block/qed-table.c
new file mode 100644
index 000000000..76d2dcccf
--- /dev/null
+++ b/contrib/qemu/block/qed-table.c
@@ -0,0 +1,296 @@
+/*
+ * QEMU Enhanced Disk Format Table I/O
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#include "trace.h"
+#include "qemu/sockets.h" /* for EINPROGRESS on Windows */
+#include "qed.h"
+
+typedef struct {
+    GenericCB gencb;
+    BDRVQEDState *s;
+    QEDTable *table;
+
+    struct iovec iov;
+    QEMUIOVector qiov;
+} QEDReadTableCB;
+
+static void qed_read_table_cb(void *opaque, int ret)
+{
+    QEDReadTableCB *read_table_cb = opaque;
+    QEDTable *table = read_table_cb->table;
+    int noffsets = read_table_cb->qiov.size / sizeof(uint64_t);
+    int i;
+
+    /* Handle I/O error */
+    if (ret) {
+        goto out;
+    }
+
+    /* Byteswap offsets */
+    for (i = 0; i < noffsets; i++) {
+        table->offsets[i] = le64_to_cpu(table->offsets[i]);
+    }
+
+out:
+    /* Completion */
+    trace_qed_read_table_cb(read_table_cb->s, read_table_cb->table, ret);
+    gencb_complete(&read_table_cb->gencb, ret);
+}
+
+static void qed_read_table(BDRVQEDState *s, uint64_t offset, QEDTable *table,
+                           BlockDriverCompletionFunc *cb, void *opaque)
+{
+    QEDReadTableCB *read_table_cb = gencb_alloc(sizeof(*read_table_cb),
+                                                cb, opaque);
+    QEMUIOVector *qiov = &read_table_cb->qiov;
+
+    trace_qed_read_table(s, offset, table);
+
+    read_table_cb->s = s;
+    read_table_cb->table = table;
+    read_table_cb->iov.iov_base = table->offsets,
+    read_table_cb->iov.iov_len = s->header.cluster_size * s->header.table_size,
+
+    qemu_iovec_init_external(qiov, &read_table_cb->iov, 1);
+    bdrv_aio_readv(s->bs->file, offset / BDRV_SECTOR_SIZE, qiov,
+                   qiov->size / BDRV_SECTOR_SIZE,
+                   qed_read_table_cb, read_table_cb);
+}
+
+typedef struct {
+    GenericCB gencb;
+    BDRVQEDState *s;
+    QEDTable *orig_table;
+    QEDTable *table;
+    bool flush;             /* flush after write? */
+
+    struct iovec iov;
+    QEMUIOVector qiov;
+} QEDWriteTableCB;
+
+static void qed_write_table_cb(void *opaque, int ret)
+{
+    QEDWriteTableCB *write_table_cb = opaque;
+
+    trace_qed_write_table_cb(write_table_cb->s,
+                             write_table_cb->orig_table,
+                             write_table_cb->flush,
+                             ret);
+
+    if (ret) {
+        goto out;
+    }
+
+    if (write_table_cb->flush) {
+        /* We still need to flush first */
+        write_table_cb->flush = false;
+        bdrv_aio_flush(write_table_cb->s->bs, qed_write_table_cb,
+                       write_table_cb);
+        return;
+    }
+
+out:
+    qemu_vfree(write_table_cb->table);
+    gencb_complete(&write_table_cb->gencb, ret);
+}
+
+/**
+ * Write out an updated part or all of a table
+ *
+ * @s:          QED state
+ * @offset:     Offset of table in image file, in bytes
+ * @table:      Table
+ * @index:      Index of first element
+ * @n:          Number of elements
+ * @flush:      Whether or not to sync to disk
+ * @cb:         Completion function
+ * @opaque:     Argument for completion function
+ */
+static void qed_write_table(BDRVQEDState *s, uint64_t offset, QEDTable *table,
+                            unsigned int index, unsigned int n, bool flush,
+                            BlockDriverCompletionFunc *cb, void *opaque)
+{
+    QEDWriteTableCB *write_table_cb;
+    unsigned int sector_mask = BDRV_SECTOR_SIZE / sizeof(uint64_t) - 1;
+    unsigned int start, end, i;
+    size_t len_bytes;
+
+    trace_qed_write_table(s, offset, table, index, n);
+
+    /* Calculate indices of the first and one after last elements */
+    start = index & ~sector_mask;
+    end = (index + n + sector_mask) & ~sector_mask;
+
+    len_bytes = (end - start) * sizeof(uint64_t);
+
+    write_table_cb = gencb_alloc(sizeof(*write_table_cb), cb, opaque);
+    write_table_cb->s = s;
+    write_table_cb->orig_table = table;
+    write_table_cb->flush = flush;
+    write_table_cb->table = qemu_blockalign(s->bs, len_bytes);
+    write_table_cb->iov.iov_base = write_table_cb->table->offsets;
+    write_table_cb->iov.iov_len = len_bytes;
+    qemu_iovec_init_external(&write_table_cb->qiov, &write_table_cb->iov, 1);
+
+    /* Byteswap table */
+    for (i = start; i < end; i++) {
+        uint64_t le_offset = cpu_to_le64(table->offsets[i]);
+        write_table_cb->table->offsets[i - start] = le_offset;
+    }
+
+    /* Adjust for offset into table */
+    offset += start * sizeof(uint64_t);
+
+    bdrv_aio_writev(s->bs->file, offset / BDRV_SECTOR_SIZE,
+                    &write_table_cb->qiov,
+                    write_table_cb->qiov.size / BDRV_SECTOR_SIZE,
+                    qed_write_table_cb, write_table_cb);
+}
+
+/**
+ * Propagate return value from async callback
+ */
+static void qed_sync_cb(void *opaque, int ret)
+{
+    *(int *)opaque = ret;
+}
+
+int qed_read_l1_table_sync(BDRVQEDState *s)
+{
+    int ret = -EINPROGRESS;
+
+    qed_read_table(s, s->header.l1_table_offset,
+                   s->l1_table, qed_sync_cb, &ret);
+    while (ret == -EINPROGRESS) {
+        qemu_aio_wait();
+    }
+
+    return ret;
+}
+
+void qed_write_l1_table(BDRVQEDState *s, unsigned int index, unsigned int n,
+                        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    BLKDBG_EVENT(s->bs->file, BLKDBG_L1_UPDATE);
+    qed_write_table(s, s->header.l1_table_offset,
+                    s->l1_table, index, n, false, cb, opaque);
+}
+
+int qed_write_l1_table_sync(BDRVQEDState *s, unsigned int index,
+                            unsigned int n)
+{
+    int ret = -EINPROGRESS;
+
+    qed_write_l1_table(s, index, n, qed_sync_cb, &ret);
+    while (ret == -EINPROGRESS) {
+        qemu_aio_wait();
+    }
+
+    return ret;
+}
+
+typedef struct {
+    GenericCB gencb;
+    BDRVQEDState *s;
+    uint64_t l2_offset;
+    QEDRequest *request;
+} QEDReadL2TableCB;
+
+static void qed_read_l2_table_cb(void *opaque, int ret)
+{
+    QEDReadL2TableCB *read_l2_table_cb = opaque;
+    QEDRequest *request = read_l2_table_cb->request;
+    BDRVQEDState *s = read_l2_table_cb->s;
+    CachedL2Table *l2_table = request->l2_table;
+    uint64_t l2_offset = read_l2_table_cb->l2_offset;
+
+    if (ret) {
+        /* can't trust loaded L2 table anymore */
+        qed_unref_l2_cache_entry(l2_table);
+        request->l2_table = NULL;
+    } else {
+        l2_table->offset = l2_offset;
+
+        qed_commit_l2_cache_entry(&s->l2_cache, l2_table);
+
+        /* This is guaranteed to succeed because we just committed the entry
+         * to the cache.
+         */
+        request->l2_table = qed_find_l2_cache_entry(&s->l2_cache, l2_offset);
+        assert(request->l2_table != NULL);
+    }
+
+    gencb_complete(&read_l2_table_cb->gencb, ret);
+}
+
+void qed_read_l2_table(BDRVQEDState *s, QEDRequest *request, uint64_t offset,
+                       BlockDriverCompletionFunc *cb, void *opaque)
+{
+    QEDReadL2TableCB *read_l2_table_cb;
+
+    qed_unref_l2_cache_entry(request->l2_table);
+
+    /* Check for cached L2 entry */
+    request->l2_table = qed_find_l2_cache_entry(&s->l2_cache, offset);
+    if (request->l2_table) {
+        cb(opaque, 0);
+        return;
+    }
+
+    request->l2_table = qed_alloc_l2_cache_entry(&s->l2_cache);
+    request->l2_table->table = qed_alloc_table(s);
+
+    read_l2_table_cb = gencb_alloc(sizeof(*read_l2_table_cb), cb, opaque);
+    read_l2_table_cb->s = s;
+    read_l2_table_cb->l2_offset = offset;
+    read_l2_table_cb->request = request;
+
+    BLKDBG_EVENT(s->bs->file, BLKDBG_L2_LOAD);
+    qed_read_table(s, offset, request->l2_table->table,
+                   qed_read_l2_table_cb, read_l2_table_cb);
+}
+
+int qed_read_l2_table_sync(BDRVQEDState *s, QEDRequest *request, uint64_t offset)
+{
+    int ret = -EINPROGRESS;
+
+    qed_read_l2_table(s, request, offset, qed_sync_cb, &ret);
+    while (ret == -EINPROGRESS) {
+        qemu_aio_wait();
+    }
+
+    return ret;
+}
+
+void qed_write_l2_table(BDRVQEDState *s, QEDRequest *request,
+                        unsigned int index, unsigned int n, bool flush,
+                        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    BLKDBG_EVENT(s->bs->file, BLKDBG_L2_UPDATE);
+    qed_write_table(s, request->l2_table->offset,
+                    request->l2_table->table, index, n, flush, cb, opaque);
+}
+
+int qed_write_l2_table_sync(BDRVQEDState *s, QEDRequest *request,
+                            unsigned int index, unsigned int n, bool flush)
+{
+    int ret = -EINPROGRESS;
+
+    qed_write_l2_table(s, request, index, n, flush, qed_sync_cb, &ret);
+    while (ret == -EINPROGRESS) {
+        qemu_aio_wait();
+    }
+
+    return ret;
+}
diff --git a/contrib/qemu/block/qed.c b/contrib/qemu/block/qed.c
new file mode 100644
index 000000000..f767b0528
--- /dev/null
+++ b/contrib/qemu/block/qed.c
@@ -0,0 +1,1596 @@
+/*
+ * QEMU Enhanced Disk Format
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#include "qemu/timer.h"
+#include "trace.h"
+#include "qed.h"
+#include "qapi/qmp/qerror.h"
+#include "migration/migration.h"
+
+static void qed_aio_cancel(BlockDriverAIOCB *blockacb)
+{
+    QEDAIOCB *acb = (QEDAIOCB *)blockacb;
+    bool finished = false;
+
+    /* Wait for the request to finish */
+    acb->finished = &finished;
+    while (!finished) {
+        qemu_aio_wait();
+    }
+}
+
+static const AIOCBInfo qed_aiocb_info = {
+    .aiocb_size         = sizeof(QEDAIOCB),
+    .cancel             = qed_aio_cancel,
+};
+
+static int bdrv_qed_probe(const uint8_t *buf, int buf_size,
+                          const char *filename)
+{
+    const QEDHeader *header = (const QEDHeader *)buf;
+
+    if (buf_size < sizeof(*header)) {
+        return 0;
+    }
+    if (le32_to_cpu(header->magic) != QED_MAGIC) {
+        return 0;
+    }
+    return 100;
+}
+
+/**
+ * Check whether an image format is raw
+ *
+ * @fmt:    Backing file format, may be NULL
+ */
+static bool qed_fmt_is_raw(const char *fmt)
+{
+    return fmt && strcmp(fmt, "raw") == 0;
+}
+
+static void qed_header_le_to_cpu(const QEDHeader *le, QEDHeader *cpu)
+{
+    cpu->magic = le32_to_cpu(le->magic);
+    cpu->cluster_size = le32_to_cpu(le->cluster_size);
+    cpu->table_size = le32_to_cpu(le->table_size);
+    cpu->header_size = le32_to_cpu(le->header_size);
+    cpu->features = le64_to_cpu(le->features);
+    cpu->compat_features = le64_to_cpu(le->compat_features);
+    cpu->autoclear_features = le64_to_cpu(le->autoclear_features);
+    cpu->l1_table_offset = le64_to_cpu(le->l1_table_offset);
+    cpu->image_size = le64_to_cpu(le->image_size);
+    cpu->backing_filename_offset = le32_to_cpu(le->backing_filename_offset);
+    cpu->backing_filename_size = le32_to_cpu(le->backing_filename_size);
+}
+
+static void qed_header_cpu_to_le(const QEDHeader *cpu, QEDHeader *le)
+{
+    le->magic = cpu_to_le32(cpu->magic);
+    le->cluster_size = cpu_to_le32(cpu->cluster_size);
+    le->table_size = cpu_to_le32(cpu->table_size);
+    le->header_size = cpu_to_le32(cpu->header_size);
+    le->features = cpu_to_le64(cpu->features);
+    le->compat_features = cpu_to_le64(cpu->compat_features);
+    le->autoclear_features = cpu_to_le64(cpu->autoclear_features);
+    le->l1_table_offset = cpu_to_le64(cpu->l1_table_offset);
+    le->image_size = cpu_to_le64(cpu->image_size);
+    le->backing_filename_offset = cpu_to_le32(cpu->backing_filename_offset);
+    le->backing_filename_size = cpu_to_le32(cpu->backing_filename_size);
+}
+
+int qed_write_header_sync(BDRVQEDState *s)
+{
+    QEDHeader le;
+    int ret;
+
+    qed_header_cpu_to_le(&s->header, &le);
+    ret = bdrv_pwrite(s->bs->file, 0, &le, sizeof(le));
+    if (ret != sizeof(le)) {
+        return ret;
+    }
+    return 0;
+}
+
+typedef struct {
+    GenericCB gencb;
+    BDRVQEDState *s;
+    struct iovec iov;
+    QEMUIOVector qiov;
+    int nsectors;
+    uint8_t *buf;
+} QEDWriteHeaderCB;
+
+static void qed_write_header_cb(void *opaque, int ret)
+{
+    QEDWriteHeaderCB *write_header_cb = opaque;
+
+    qemu_vfree(write_header_cb->buf);
+    gencb_complete(write_header_cb, ret);
+}
+
+static void qed_write_header_read_cb(void *opaque, int ret)
+{
+    QEDWriteHeaderCB *write_header_cb = opaque;
+    BDRVQEDState *s = write_header_cb->s;
+
+    if (ret) {
+        qed_write_header_cb(write_header_cb, ret);
+        return;
+    }
+
+    /* Update header */
+    qed_header_cpu_to_le(&s->header, (QEDHeader *)write_header_cb->buf);
+
+    bdrv_aio_writev(s->bs->file, 0, &write_header_cb->qiov,
+                    write_header_cb->nsectors, qed_write_header_cb,
+                    write_header_cb);
+}
+
+/**
+ * Update header in-place (does not rewrite backing filename or other strings)
+ *
+ * This function only updates known header fields in-place and does not affect
+ * extra data after the QED header.
+ */
+static void qed_write_header(BDRVQEDState *s, BlockDriverCompletionFunc cb,
+                             void *opaque)
+{
+    /* We must write full sectors for O_DIRECT but cannot necessarily generate
+     * the data following the header if an unrecognized compat feature is
+     * active.  Therefore, first read the sectors containing the header, update
+     * them, and write back.
+     */
+
+    int nsectors = (sizeof(QEDHeader) + BDRV_SECTOR_SIZE - 1) /
+                   BDRV_SECTOR_SIZE;
+    size_t len = nsectors * BDRV_SECTOR_SIZE;
+    QEDWriteHeaderCB *write_header_cb = gencb_alloc(sizeof(*write_header_cb),
+                                                    cb, opaque);
+
+    write_header_cb->s = s;
+    write_header_cb->nsectors = nsectors;
+    write_header_cb->buf = qemu_blockalign(s->bs, len);
+    write_header_cb->iov.iov_base = write_header_cb->buf;
+    write_header_cb->iov.iov_len = len;
+    qemu_iovec_init_external(&write_header_cb->qiov, &write_header_cb->iov, 1);
+
+    bdrv_aio_readv(s->bs->file, 0, &write_header_cb->qiov, nsectors,
+                   qed_write_header_read_cb, write_header_cb);
+}
+
+static uint64_t qed_max_image_size(uint32_t cluster_size, uint32_t table_size)
+{
+    uint64_t table_entries;
+    uint64_t l2_size;
+
+    table_entries = (table_size * cluster_size) / sizeof(uint64_t);
+    l2_size = table_entries * cluster_size;
+
+    return l2_size * table_entries;
+}
+
+static bool qed_is_cluster_size_valid(uint32_t cluster_size)
+{
+    if (cluster_size < QED_MIN_CLUSTER_SIZE ||
+        cluster_size > QED_MAX_CLUSTER_SIZE) {
+        return false;
+    }
+    if (cluster_size & (cluster_size - 1)) {
+        return false; /* not power of 2 */
+    }
+    return true;
+}
+
+static bool qed_is_table_size_valid(uint32_t table_size)
+{
+    if (table_size < QED_MIN_TABLE_SIZE ||
+        table_size > QED_MAX_TABLE_SIZE) {
+        return false;
+    }
+    if (table_size & (table_size - 1)) {
+        return false; /* not power of 2 */
+    }
+    return true;
+}
+
+static bool qed_is_image_size_valid(uint64_t image_size, uint32_t cluster_size,
+                                    uint32_t table_size)
+{
+    if (image_size % BDRV_SECTOR_SIZE != 0) {
+        return false; /* not multiple of sector size */
+    }
+    if (image_size > qed_max_image_size(cluster_size, table_size)) {
+        return false; /* image is too large */
+    }
+    return true;
+}
+
+/**
+ * Read a string of known length from the image file
+ *
+ * @file:       Image file
+ * @offset:     File offset to start of string, in bytes
+ * @n:          String length in bytes
+ * @buf:        Destination buffer
+ * @buflen:     Destination buffer length in bytes
+ * @ret:        0 on success, -errno on failure
+ *
+ * The string is NUL-terminated.
+ */
+static int qed_read_string(BlockDriverState *file, uint64_t offset, size_t n,
+                           char *buf, size_t buflen)
+{
+    int ret;
+    if (n >= buflen) {
+        return -EINVAL;
+    }
+    ret = bdrv_pread(file, offset, buf, n);
+    if (ret < 0) {
+        return ret;
+    }
+    buf[n] = '\0';
+    return 0;
+}
+
+/**
+ * Allocate new clusters
+ *
+ * @s:          QED state
+ * @n:          Number of contiguous clusters to allocate
+ * @ret:        Offset of first allocated cluster
+ *
+ * This function only produces the offset where the new clusters should be
+ * written.  It updates BDRVQEDState but does not make any changes to the image
+ * file.
+ */
+static uint64_t qed_alloc_clusters(BDRVQEDState *s, unsigned int n)
+{
+    uint64_t offset = s->file_size;
+    s->file_size += n * s->header.cluster_size;
+    return offset;
+}
+
+QEDTable *qed_alloc_table(BDRVQEDState *s)
+{
+    /* Honor O_DIRECT memory alignment requirements */
+    return qemu_blockalign(s->bs,
+                           s->header.cluster_size * s->header.table_size);
+}
+
+/**
+ * Allocate a new zeroed L2 table
+ */
+static CachedL2Table *qed_new_l2_table(BDRVQEDState *s)
+{
+    CachedL2Table *l2_table = qed_alloc_l2_cache_entry(&s->l2_cache);
+
+    l2_table->table = qed_alloc_table(s);
+    l2_table->offset = qed_alloc_clusters(s, s->header.table_size);
+
+    memset(l2_table->table->offsets, 0,
+           s->header.cluster_size * s->header.table_size);
+    return l2_table;
+}
+
+static void qed_aio_next_io(void *opaque, int ret);
+
+static void qed_plug_allocating_write_reqs(BDRVQEDState *s)
+{
+    assert(!s->allocating_write_reqs_plugged);
+
+    s->allocating_write_reqs_plugged = true;
+}
+
+static void qed_unplug_allocating_write_reqs(BDRVQEDState *s)
+{
+    QEDAIOCB *acb;
+
+    assert(s->allocating_write_reqs_plugged);
+
+    s->allocating_write_reqs_plugged = false;
+
+    acb = QSIMPLEQ_FIRST(&s->allocating_write_reqs);
+    if (acb) {
+        qed_aio_next_io(acb, 0);
+    }
+}
+
+static void qed_finish_clear_need_check(void *opaque, int ret)
+{
+    /* Do nothing */
+}
+
+static void qed_flush_after_clear_need_check(void *opaque, int ret)
+{
+    BDRVQEDState *s = opaque;
+
+    bdrv_aio_flush(s->bs, qed_finish_clear_need_check, s);
+
+    /* No need to wait until flush completes */
+    qed_unplug_allocating_write_reqs(s);
+}
+
+static void qed_clear_need_check(void *opaque, int ret)
+{
+    BDRVQEDState *s = opaque;
+
+    if (ret) {
+        qed_unplug_allocating_write_reqs(s);
+        return;
+    }
+
+    s->header.features &= ~QED_F_NEED_CHECK;
+    qed_write_header(s, qed_flush_after_clear_need_check, s);
+}
+
+static void qed_need_check_timer_cb(void *opaque)
+{
+    BDRVQEDState *s = opaque;
+
+    /* The timer should only fire when allocating writes have drained */
+    assert(!QSIMPLEQ_FIRST(&s->allocating_write_reqs));
+
+    trace_qed_need_check_timer_cb(s);
+
+    qed_plug_allocating_write_reqs(s);
+
+    /* Ensure writes are on disk before clearing flag */
+    bdrv_aio_flush(s->bs, qed_clear_need_check, s);
+}
+
+static void qed_start_need_check_timer(BDRVQEDState *s)
+{
+    trace_qed_start_need_check_timer(s);
+
+    /* Use vm_clock so we don't alter the image file while suspended for
+     * migration.
+     */
+    qemu_mod_timer(s->need_check_timer, qemu_get_clock_ns(vm_clock) +
+                   get_ticks_per_sec() * QED_NEED_CHECK_TIMEOUT);
+}
+
+/* It's okay to call this multiple times or when no timer is started */
+static void qed_cancel_need_check_timer(BDRVQEDState *s)
+{
+    trace_qed_cancel_need_check_timer(s);
+    qemu_del_timer(s->need_check_timer);
+}
+
+static void bdrv_qed_rebind(BlockDriverState *bs)
+{
+    BDRVQEDState *s = bs->opaque;
+    s->bs = bs;
+}
+
+static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags)
+{
+    BDRVQEDState *s = bs->opaque;
+    QEDHeader le_header;
+    int64_t file_size;
+    int ret;
+
+    s->bs = bs;
+    QSIMPLEQ_INIT(&s->allocating_write_reqs);
+
+    ret = bdrv_pread(bs->file, 0, &le_header, sizeof(le_header));
+    if (ret < 0) {
+        return ret;
+    }
+    qed_header_le_to_cpu(&le_header, &s->header);
+
+    if (s->header.magic != QED_MAGIC) {
+        return -EMEDIUMTYPE;
+    }
+    if (s->header.features & ~QED_FEATURE_MASK) {
+        /* image uses unsupported feature bits */
+        char buf[64];
+        snprintf(buf, sizeof(buf), "%" PRIx64,
+            s->header.features & ~QED_FEATURE_MASK);
+        qerror_report(QERR_UNKNOWN_BLOCK_FORMAT_FEATURE,
+            bs->device_name, "QED", buf);
+        return -ENOTSUP;
+    }
+    if (!qed_is_cluster_size_valid(s->header.cluster_size)) {
+        return -EINVAL;
+    }
+
+    /* Round down file size to the last cluster */
+    file_size = bdrv_getlength(bs->file);
+    if (file_size < 0) {
+        return file_size;
+    }
+    s->file_size = qed_start_of_cluster(s, file_size);
+
+    if (!qed_is_table_size_valid(s->header.table_size)) {
+        return -EINVAL;
+    }
+    if (!qed_is_image_size_valid(s->header.image_size,
+                                 s->header.cluster_size,
+                                 s->header.table_size)) {
+        return -EINVAL;
+    }
+    if (!qed_check_table_offset(s, s->header.l1_table_offset)) {
+        return -EINVAL;
+    }
+
+    s->table_nelems = (s->header.cluster_size * s->header.table_size) /
+                      sizeof(uint64_t);
+    s->l2_shift = ffs(s->header.cluster_size) - 1;
+    s->l2_mask = s->table_nelems - 1;
+    s->l1_shift = s->l2_shift + ffs(s->table_nelems) - 1;
+
+    if ((s->header.features & QED_F_BACKING_FILE)) {
+        if ((uint64_t)s->header.backing_filename_offset +
+            s->header.backing_filename_size >
+            s->header.cluster_size * s->header.header_size) {
+            return -EINVAL;
+        }
+
+        ret = qed_read_string(bs->file, s->header.backing_filename_offset,
+                              s->header.backing_filename_size, bs->backing_file,
+                              sizeof(bs->backing_file));
+        if (ret < 0) {
+            return ret;
+        }
+
+        if (s->header.features & QED_F_BACKING_FORMAT_NO_PROBE) {
+            pstrcpy(bs->backing_format, sizeof(bs->backing_format), "raw");
+        }
+    }
+
+    /* Reset unknown autoclear feature bits.  This is a backwards
+     * compatibility mechanism that allows images to be opened by older
+     * programs, which "knock out" unknown feature bits.  When an image is
+     * opened by a newer program again it can detect that the autoclear
+     * feature is no longer valid.
+     */
+    if ((s->header.autoclear_features & ~QED_AUTOCLEAR_FEATURE_MASK) != 0 &&
+        !bdrv_is_read_only(bs->file) && !(flags & BDRV_O_INCOMING)) {
+        s->header.autoclear_features &= QED_AUTOCLEAR_FEATURE_MASK;
+
+        ret = qed_write_header_sync(s);
+        if (ret) {
+            return ret;
+        }
+
+        /* From here on only known autoclear feature bits are valid */
+        bdrv_flush(bs->file);
+    }
+
+    s->l1_table = qed_alloc_table(s);
+    qed_init_l2_cache(&s->l2_cache);
+
+    ret = qed_read_l1_table_sync(s);
+    if (ret) {
+        goto out;
+    }
+
+    /* If image was not closed cleanly, check consistency */
+    if (!(flags & BDRV_O_CHECK) && (s->header.features & QED_F_NEED_CHECK)) {
+        /* Read-only images cannot be fixed.  There is no risk of corruption
+         * since write operations are not possible.  Therefore, allow
+         * potentially inconsistent images to be opened read-only.  This can
+         * aid data recovery from an otherwise inconsistent image.
+         */
+        if (!bdrv_is_read_only(bs->file) &&
+            !(flags & BDRV_O_INCOMING)) {
+            BdrvCheckResult result = {0};
+
+            ret = qed_check(s, &result, true);
+            if (ret) {
+                goto out;
+            }
+        }
+    }
+
+    s->need_check_timer = qemu_new_timer_ns(vm_clock,
+                                            qed_need_check_timer_cb, s);
+
+out:
+    if (ret) {
+        qed_free_l2_cache(&s->l2_cache);
+        qemu_vfree(s->l1_table);
+    }
+    return ret;
+}
+
+/* We have nothing to do for QED reopen, stubs just return
+ * success */
+static int bdrv_qed_reopen_prepare(BDRVReopenState *state,
+                                   BlockReopenQueue *queue, Error **errp)
+{
+    return 0;
+}
+
+static void bdrv_qed_close(BlockDriverState *bs)
+{
+    BDRVQEDState *s = bs->opaque;
+
+    qed_cancel_need_check_timer(s);
+    qemu_free_timer(s->need_check_timer);
+
+    /* Ensure writes reach stable storage */
+    bdrv_flush(bs->file);
+
+    /* Clean shutdown, no check required on next open */
+    if (s->header.features & QED_F_NEED_CHECK) {
+        s->header.features &= ~QED_F_NEED_CHECK;
+        qed_write_header_sync(s);
+    }
+
+    qed_free_l2_cache(&s->l2_cache);
+    qemu_vfree(s->l1_table);
+}
+
+static int qed_create(const char *filename, uint32_t cluster_size,
+                      uint64_t image_size, uint32_t table_size,
+                      const char *backing_file, const char *backing_fmt)
+{
+    QEDHeader header = {
+        .magic = QED_MAGIC,
+        .cluster_size = cluster_size,
+        .table_size = table_size,
+        .header_size = 1,
+        .features = 0,
+        .compat_features = 0,
+        .l1_table_offset = cluster_size,
+        .image_size = image_size,
+    };
+    QEDHeader le_header;
+    uint8_t *l1_table = NULL;
+    size_t l1_size = header.cluster_size * header.table_size;
+    int ret = 0;
+    BlockDriverState *bs = NULL;
+
+    ret = bdrv_create_file(filename, NULL);
+    if (ret < 0) {
+        return ret;
+    }
+
+    ret = bdrv_file_open(&bs, filename, NULL, BDRV_O_RDWR | BDRV_O_CACHE_WB);
+    if (ret < 0) {
+        return ret;
+    }
+
+    /* File must start empty and grow, check truncate is supported */
+    ret = bdrv_truncate(bs, 0);
+    if (ret < 0) {
+        goto out;
+    }
+
+    if (backing_file) {
+        header.features |= QED_F_BACKING_FILE;
+        header.backing_filename_offset = sizeof(le_header);
+        header.backing_filename_size = strlen(backing_file);
+
+        if (qed_fmt_is_raw(backing_fmt)) {
+            header.features |= QED_F_BACKING_FORMAT_NO_PROBE;
+        }
+    }
+
+    qed_header_cpu_to_le(&header, &le_header);
+    ret = bdrv_pwrite(bs, 0, &le_header, sizeof(le_header));
+    if (ret < 0) {
+        goto out;
+    }
+    ret = bdrv_pwrite(bs, sizeof(le_header), backing_file,
+                      header.backing_filename_size);
+    if (ret < 0) {
+        goto out;
+    }
+
+    l1_table = g_malloc0(l1_size);
+    ret = bdrv_pwrite(bs, header.l1_table_offset, l1_table, l1_size);
+    if (ret < 0) {
+        goto out;
+    }
+
+    ret = 0; /* success */
+out:
+    g_free(l1_table);
+    bdrv_delete(bs);
+    return ret;
+}
+
+static int bdrv_qed_create(const char *filename, QEMUOptionParameter *options)
+{
+    uint64_t image_size = 0;
+    uint32_t cluster_size = QED_DEFAULT_CLUSTER_SIZE;
+    uint32_t table_size = QED_DEFAULT_TABLE_SIZE;
+    const char *backing_file = NULL;
+    const char *backing_fmt = NULL;
+
+    while (options && options->name) {
+        if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
+            image_size = options->value.n;
+        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
+            backing_file = options->value.s;
+        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FMT)) {
+            backing_fmt = options->value.s;
+        } else if (!strcmp(options->name, BLOCK_OPT_CLUSTER_SIZE)) {
+            if (options->value.n) {
+                cluster_size = options->value.n;
+            }
+        } else if (!strcmp(options->name, BLOCK_OPT_TABLE_SIZE)) {
+            if (options->value.n) {
+                table_size = options->value.n;
+            }
+        }
+        options++;
+    }
+
+    if (!qed_is_cluster_size_valid(cluster_size)) {
+        fprintf(stderr, "QED cluster size must be within range [%u, %u] and power of 2\n",
+                QED_MIN_CLUSTER_SIZE, QED_MAX_CLUSTER_SIZE);
+        return -EINVAL;
+    }
+    if (!qed_is_table_size_valid(table_size)) {
+        fprintf(stderr, "QED table size must be within range [%u, %u] and power of 2\n",
+                QED_MIN_TABLE_SIZE, QED_MAX_TABLE_SIZE);
+        return -EINVAL;
+    }
+    if (!qed_is_image_size_valid(image_size, cluster_size, table_size)) {
+        fprintf(stderr, "QED image size must be a non-zero multiple of "
+                        "cluster size and less than %" PRIu64 " bytes\n",
+                qed_max_image_size(cluster_size, table_size));
+        return -EINVAL;
+    }
+
+    return qed_create(filename, cluster_size, image_size, table_size,
+                      backing_file, backing_fmt);
+}
+
+typedef struct {
+    Coroutine *co;
+    int is_allocated;
+    int *pnum;
+} QEDIsAllocatedCB;
+
+static void qed_is_allocated_cb(void *opaque, int ret, uint64_t offset, size_t len)
+{
+    QEDIsAllocatedCB *cb = opaque;
+    *cb->pnum = len / BDRV_SECTOR_SIZE;
+    cb->is_allocated = (ret == QED_CLUSTER_FOUND || ret == QED_CLUSTER_ZERO);
+    if (cb->co) {
+        qemu_coroutine_enter(cb->co, NULL);
+    }
+}
+
+static int coroutine_fn bdrv_qed_co_is_allocated(BlockDriverState *bs,
+                                                 int64_t sector_num,
+                                                 int nb_sectors, int *pnum)
+{
+    BDRVQEDState *s = bs->opaque;
+    uint64_t pos = (uint64_t)sector_num * BDRV_SECTOR_SIZE;
+    size_t len = (size_t)nb_sectors * BDRV_SECTOR_SIZE;
+    QEDIsAllocatedCB cb = {
+        .is_allocated = -1,
+        .pnum = pnum,
+    };
+    QEDRequest request = { .l2_table = NULL };
+
+    qed_find_cluster(s, &request, pos, len, qed_is_allocated_cb, &cb);
+
+    /* Now sleep if the callback wasn't invoked immediately */
+    while (cb.is_allocated == -1) {
+        cb.co = qemu_coroutine_self();
+        qemu_coroutine_yield();
+    }
+
+    qed_unref_l2_cache_entry(request.l2_table);
+
+    return cb.is_allocated;
+}
+
+static int bdrv_qed_make_empty(BlockDriverState *bs)
+{
+    return -ENOTSUP;
+}
+
+static BDRVQEDState *acb_to_s(QEDAIOCB *acb)
+{
+    return acb->common.bs->opaque;
+}
+
+/**
+ * Read from the backing file or zero-fill if no backing file
+ *
+ * @s:          QED state
+ * @pos:        Byte position in device
+ * @qiov:       Destination I/O vector
+ * @cb:         Completion function
+ * @opaque:     User data for completion function
+ *
+ * This function reads qiov->size bytes starting at pos from the backing file.
+ * If there is no backing file then zeroes are read.
+ */
+static void qed_read_backing_file(BDRVQEDState *s, uint64_t pos,
+                                  QEMUIOVector *qiov,
+                                  BlockDriverCompletionFunc *cb, void *opaque)
+{
+    uint64_t backing_length = 0;
+    size_t size;
+
+    /* If there is a backing file, get its length.  Treat the absence of a
+     * backing file like a zero length backing file.
+     */
+    if (s->bs->backing_hd) {
+        int64_t l = bdrv_getlength(s->bs->backing_hd);
+        if (l < 0) {
+            cb(opaque, l);
+            return;
+        }
+        backing_length = l;
+    }
+
+    /* Zero all sectors if reading beyond the end of the backing file */
+    if (pos >= backing_length ||
+        pos + qiov->size > backing_length) {
+        qemu_iovec_memset(qiov, 0, 0, qiov->size);
+    }
+
+    /* Complete now if there are no backing file sectors to read */
+    if (pos >= backing_length) {
+        cb(opaque, 0);
+        return;
+    }
+
+    /* If the read straddles the end of the backing file, shorten it */
+    size = MIN((uint64_t)backing_length - pos, qiov->size);
+
+    BLKDBG_EVENT(s->bs->file, BLKDBG_READ_BACKING_AIO);
+    bdrv_aio_readv(s->bs->backing_hd, pos / BDRV_SECTOR_SIZE,
+                   qiov, size / BDRV_SECTOR_SIZE, cb, opaque);
+}
+
+typedef struct {
+    GenericCB gencb;
+    BDRVQEDState *s;
+    QEMUIOVector qiov;
+    struct iovec iov;
+    uint64_t offset;
+} CopyFromBackingFileCB;
+
+static void qed_copy_from_backing_file_cb(void *opaque, int ret)
+{
+    CopyFromBackingFileCB *copy_cb = opaque;
+    qemu_vfree(copy_cb->iov.iov_base);
+    gencb_complete(&copy_cb->gencb, ret);
+}
+
+static void qed_copy_from_backing_file_write(void *opaque, int ret)
+{
+    CopyFromBackingFileCB *copy_cb = opaque;
+    BDRVQEDState *s = copy_cb->s;
+
+    if (ret) {
+        qed_copy_from_backing_file_cb(copy_cb, ret);
+        return;
+    }
+
+    BLKDBG_EVENT(s->bs->file, BLKDBG_COW_WRITE);
+    bdrv_aio_writev(s->bs->file, copy_cb->offset / BDRV_SECTOR_SIZE,
+                    &copy_cb->qiov, copy_cb->qiov.size / BDRV_SECTOR_SIZE,
+                    qed_copy_from_backing_file_cb, copy_cb);
+}
+
+/**
+ * Copy data from backing file into the image
+ *
+ * @s:          QED state
+ * @pos:        Byte position in device
+ * @len:        Number of bytes
+ * @offset:     Byte offset in image file
+ * @cb:         Completion function
+ * @opaque:     User data for completion function
+ */
+static void qed_copy_from_backing_file(BDRVQEDState *s, uint64_t pos,
+                                       uint64_t len, uint64_t offset,
+                                       BlockDriverCompletionFunc *cb,
+                                       void *opaque)
+{
+    CopyFromBackingFileCB *copy_cb;
+
+    /* Skip copy entirely if there is no work to do */
+    if (len == 0) {
+        cb(opaque, 0);
+        return;
+    }
+
+    copy_cb = gencb_alloc(sizeof(*copy_cb), cb, opaque);
+    copy_cb->s = s;
+    copy_cb->offset = offset;
+    copy_cb->iov.iov_base = qemu_blockalign(s->bs, len);
+    copy_cb->iov.iov_len = len;
+    qemu_iovec_init_external(&copy_cb->qiov, &copy_cb->iov, 1);
+
+    qed_read_backing_file(s, pos, &copy_cb->qiov,
+                          qed_copy_from_backing_file_write, copy_cb);
+}
+
+/**
+ * Link one or more contiguous clusters into a table
+ *
+ * @s:              QED state
+ * @table:          L2 table
+ * @index:          First cluster index
+ * @n:              Number of contiguous clusters
+ * @cluster:        First cluster offset
+ *
+ * The cluster offset may be an allocated byte offset in the image file, the
+ * zero cluster marker, or the unallocated cluster marker.
+ */
+static void qed_update_l2_table(BDRVQEDState *s, QEDTable *table, int index,
+                                unsigned int n, uint64_t cluster)
+{
+    int i;
+    for (i = index; i < index + n; i++) {
+        table->offsets[i] = cluster;
+        if (!qed_offset_is_unalloc_cluster(cluster) &&
+            !qed_offset_is_zero_cluster(cluster)) {
+            cluster += s->header.cluster_size;
+        }
+    }
+}
+
+static void qed_aio_complete_bh(void *opaque)
+{
+    QEDAIOCB *acb = opaque;
+    BlockDriverCompletionFunc *cb = acb->common.cb;
+    void *user_opaque = acb->common.opaque;
+    int ret = acb->bh_ret;
+    bool *finished = acb->finished;
+
+    qemu_bh_delete(acb->bh);
+    qemu_aio_release(acb);
+
+    /* Invoke callback */
+    cb(user_opaque, ret);
+
+    /* Signal cancel completion */
+    if (finished) {
+        *finished = true;
+    }
+}
+
+static void qed_aio_complete(QEDAIOCB *acb, int ret)
+{
+    BDRVQEDState *s = acb_to_s(acb);
+
+    trace_qed_aio_complete(s, acb, ret);
+
+    /* Free resources */
+    qemu_iovec_destroy(&acb->cur_qiov);
+    qed_unref_l2_cache_entry(acb->request.l2_table);
+
+    /* Free the buffer we may have allocated for zero writes */
+    if (acb->flags & QED_AIOCB_ZERO) {
+        qemu_vfree(acb->qiov->iov[0].iov_base);
+        acb->qiov->iov[0].iov_base = NULL;
+    }
+
+    /* Arrange for a bh to invoke the completion function */
+    acb->bh_ret = ret;
+    acb->bh = qemu_bh_new(qed_aio_complete_bh, acb);
+    qemu_bh_schedule(acb->bh);
+
+    /* Start next allocating write request waiting behind this one.  Note that
+     * requests enqueue themselves when they first hit an unallocated cluster
+     * but they wait until the entire request is finished before waking up the
+     * next request in the queue.  This ensures that we don't cycle through
+     * requests multiple times but rather finish one at a time completely.
+     */
+    if (acb == QSIMPLEQ_FIRST(&s->allocating_write_reqs)) {
+        QSIMPLEQ_REMOVE_HEAD(&s->allocating_write_reqs, next);
+        acb = QSIMPLEQ_FIRST(&s->allocating_write_reqs);
+        if (acb) {
+            qed_aio_next_io(acb, 0);
+        } else if (s->header.features & QED_F_NEED_CHECK) {
+            qed_start_need_check_timer(s);
+        }
+    }
+}
+
+/**
+ * Commit the current L2 table to the cache
+ */
+static void qed_commit_l2_update(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    CachedL2Table *l2_table = acb->request.l2_table;
+    uint64_t l2_offset = l2_table->offset;
+
+    qed_commit_l2_cache_entry(&s->l2_cache, l2_table);
+
+    /* This is guaranteed to succeed because we just committed the entry to the
+     * cache.
+     */
+    acb->request.l2_table = qed_find_l2_cache_entry(&s->l2_cache, l2_offset);
+    assert(acb->request.l2_table != NULL);
+
+    qed_aio_next_io(opaque, ret);
+}
+
+/**
+ * Update L1 table with new L2 table offset and write it out
+ */
+static void qed_aio_write_l1_update(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    int index;
+
+    if (ret) {
+        qed_aio_complete(acb, ret);
+        return;
+    }
+
+    index = qed_l1_index(s, acb->cur_pos);
+    s->l1_table->offsets[index] = acb->request.l2_table->offset;
+
+    qed_write_l1_table(s, index, 1, qed_commit_l2_update, acb);
+}
+
+/**
+ * Update L2 table with new cluster offsets and write them out
+ */
+static void qed_aio_write_l2_update(QEDAIOCB *acb, int ret, uint64_t offset)
+{
+    BDRVQEDState *s = acb_to_s(acb);
+    bool need_alloc = acb->find_cluster_ret == QED_CLUSTER_L1;
+    int index;
+
+    if (ret) {
+        goto err;
+    }
+
+    if (need_alloc) {
+        qed_unref_l2_cache_entry(acb->request.l2_table);
+        acb->request.l2_table = qed_new_l2_table(s);
+    }
+
+    index = qed_l2_index(s, acb->cur_pos);
+    qed_update_l2_table(s, acb->request.l2_table->table, index, acb->cur_nclusters,
+                         offset);
+
+    if (need_alloc) {
+        /* Write out the whole new L2 table */
+        qed_write_l2_table(s, &acb->request, 0, s->table_nelems, true,
+                            qed_aio_write_l1_update, acb);
+    } else {
+        /* Write out only the updated part of the L2 table */
+        qed_write_l2_table(s, &acb->request, index, acb->cur_nclusters, false,
+                            qed_aio_next_io, acb);
+    }
+    return;
+
+err:
+    qed_aio_complete(acb, ret);
+}
+
+static void qed_aio_write_l2_update_cb(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    qed_aio_write_l2_update(acb, ret, acb->cur_cluster);
+}
+
+/**
+ * Flush new data clusters before updating the L2 table
+ *
+ * This flush is necessary when a backing file is in use.  A crash during an
+ * allocating write could result in empty clusters in the image.  If the write
+ * only touched a subregion of the cluster, then backing image sectors have
+ * been lost in the untouched region.  The solution is to flush after writing a
+ * new data cluster and before updating the L2 table.
+ */
+static void qed_aio_write_flush_before_l2_update(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+
+    if (!bdrv_aio_flush(s->bs->file, qed_aio_write_l2_update_cb, opaque)) {
+        qed_aio_complete(acb, -EIO);
+    }
+}
+
+/**
+ * Write data to the image file
+ */
+static void qed_aio_write_main(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    uint64_t offset = acb->cur_cluster +
+                      qed_offset_into_cluster(s, acb->cur_pos);
+    BlockDriverCompletionFunc *next_fn;
+
+    trace_qed_aio_write_main(s, acb, ret, offset, acb->cur_qiov.size);
+
+    if (ret) {
+        qed_aio_complete(acb, ret);
+        return;
+    }
+
+    if (acb->find_cluster_ret == QED_CLUSTER_FOUND) {
+        next_fn = qed_aio_next_io;
+    } else {
+        if (s->bs->backing_hd) {
+            next_fn = qed_aio_write_flush_before_l2_update;
+        } else {
+            next_fn = qed_aio_write_l2_update_cb;
+        }
+    }
+
+    BLKDBG_EVENT(s->bs->file, BLKDBG_WRITE_AIO);
+    bdrv_aio_writev(s->bs->file, offset / BDRV_SECTOR_SIZE,
+                    &acb->cur_qiov, acb->cur_qiov.size / BDRV_SECTOR_SIZE,
+                    next_fn, acb);
+}
+
+/**
+ * Populate back untouched region of new data cluster
+ */
+static void qed_aio_write_postfill(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    uint64_t start = acb->cur_pos + acb->cur_qiov.size;
+    uint64_t len =
+        qed_start_of_cluster(s, start + s->header.cluster_size - 1) - start;
+    uint64_t offset = acb->cur_cluster +
+                      qed_offset_into_cluster(s, acb->cur_pos) +
+                      acb->cur_qiov.size;
+
+    if (ret) {
+        qed_aio_complete(acb, ret);
+        return;
+    }
+
+    trace_qed_aio_write_postfill(s, acb, start, len, offset);
+    qed_copy_from_backing_file(s, start, len, offset,
+                                qed_aio_write_main, acb);
+}
+
+/**
+ * Populate front untouched region of new data cluster
+ */
+static void qed_aio_write_prefill(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    uint64_t start = qed_start_of_cluster(s, acb->cur_pos);
+    uint64_t len = qed_offset_into_cluster(s, acb->cur_pos);
+
+    trace_qed_aio_write_prefill(s, acb, start, len, acb->cur_cluster);
+    qed_copy_from_backing_file(s, start, len, acb->cur_cluster,
+                                qed_aio_write_postfill, acb);
+}
+
+/**
+ * Check if the QED_F_NEED_CHECK bit should be set during allocating write
+ */
+static bool qed_should_set_need_check(BDRVQEDState *s)
+{
+    /* The flush before L2 update path ensures consistency */
+    if (s->bs->backing_hd) {
+        return false;
+    }
+
+    return !(s->header.features & QED_F_NEED_CHECK);
+}
+
+static void qed_aio_write_zero_cluster(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+
+    if (ret) {
+        qed_aio_complete(acb, ret);
+        return;
+    }
+
+    qed_aio_write_l2_update(acb, 0, 1);
+}
+
+/**
+ * Write new data cluster
+ *
+ * @acb:        Write request
+ * @len:        Length in bytes
+ *
+ * This path is taken when writing to previously unallocated clusters.
+ */
+static void qed_aio_write_alloc(QEDAIOCB *acb, size_t len)
+{
+    BDRVQEDState *s = acb_to_s(acb);
+    BlockDriverCompletionFunc *cb;
+
+    /* Cancel timer when the first allocating request comes in */
+    if (QSIMPLEQ_EMPTY(&s->allocating_write_reqs)) {
+        qed_cancel_need_check_timer(s);
+    }
+
+    /* Freeze this request if another allocating write is in progress */
+    if (acb != QSIMPLEQ_FIRST(&s->allocating_write_reqs)) {
+        QSIMPLEQ_INSERT_TAIL(&s->allocating_write_reqs, acb, next);
+    }
+    if (acb != QSIMPLEQ_FIRST(&s->allocating_write_reqs) ||
+        s->allocating_write_reqs_plugged) {
+        return; /* wait for existing request to finish */
+    }
+
+    acb->cur_nclusters = qed_bytes_to_clusters(s,
+            qed_offset_into_cluster(s, acb->cur_pos) + len);
+    qemu_iovec_concat(&acb->cur_qiov, acb->qiov, acb->qiov_offset, len);
+
+    if (acb->flags & QED_AIOCB_ZERO) {
+        /* Skip ahead if the clusters are already zero */
+        if (acb->find_cluster_ret == QED_CLUSTER_ZERO) {
+            qed_aio_next_io(acb, 0);
+            return;
+        }
+
+        cb = qed_aio_write_zero_cluster;
+    } else {
+        cb = qed_aio_write_prefill;
+        acb->cur_cluster = qed_alloc_clusters(s, acb->cur_nclusters);
+    }
+
+    if (qed_should_set_need_check(s)) {
+        s->header.features |= QED_F_NEED_CHECK;
+        qed_write_header(s, cb, acb);
+    } else {
+        cb(acb, 0);
+    }
+}
+
+/**
+ * Write data cluster in place
+ *
+ * @acb:        Write request
+ * @offset:     Cluster offset in bytes
+ * @len:        Length in bytes
+ *
+ * This path is taken when writing to already allocated clusters.
+ */
+static void qed_aio_write_inplace(QEDAIOCB *acb, uint64_t offset, size_t len)
+{
+    /* Allocate buffer for zero writes */
+    if (acb->flags & QED_AIOCB_ZERO) {
+        struct iovec *iov = acb->qiov->iov;
+
+        if (!iov->iov_base) {
+            iov->iov_base = qemu_blockalign(acb->common.bs, iov->iov_len);
+            memset(iov->iov_base, 0, iov->iov_len);
+        }
+    }
+
+    /* Calculate the I/O vector */
+    acb->cur_cluster = offset;
+    qemu_iovec_concat(&acb->cur_qiov, acb->qiov, acb->qiov_offset, len);
+
+    /* Do the actual write */
+    qed_aio_write_main(acb, 0);
+}
+
+/**
+ * Write data cluster
+ *
+ * @opaque:     Write request
+ * @ret:        QED_CLUSTER_FOUND, QED_CLUSTER_L2, QED_CLUSTER_L1,
+ *              or -errno
+ * @offset:     Cluster offset in bytes
+ * @len:        Length in bytes
+ *
+ * Callback from qed_find_cluster().
+ */
+static void qed_aio_write_data(void *opaque, int ret,
+                               uint64_t offset, size_t len)
+{
+    QEDAIOCB *acb = opaque;
+
+    trace_qed_aio_write_data(acb_to_s(acb), acb, ret, offset, len);
+
+    acb->find_cluster_ret = ret;
+
+    switch (ret) {
+    case QED_CLUSTER_FOUND:
+        qed_aio_write_inplace(acb, offset, len);
+        break;
+
+    case QED_CLUSTER_L2:
+    case QED_CLUSTER_L1:
+    case QED_CLUSTER_ZERO:
+        qed_aio_write_alloc(acb, len);
+        break;
+
+    default:
+        qed_aio_complete(acb, ret);
+        break;
+    }
+}
+
+/**
+ * Read data cluster
+ *
+ * @opaque:     Read request
+ * @ret:        QED_CLUSTER_FOUND, QED_CLUSTER_L2, QED_CLUSTER_L1,
+ *              or -errno
+ * @offset:     Cluster offset in bytes
+ * @len:        Length in bytes
+ *
+ * Callback from qed_find_cluster().
+ */
+static void qed_aio_read_data(void *opaque, int ret,
+                              uint64_t offset, size_t len)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    BlockDriverState *bs = acb->common.bs;
+
+    /* Adjust offset into cluster */
+    offset += qed_offset_into_cluster(s, acb->cur_pos);
+
+    trace_qed_aio_read_data(s, acb, ret, offset, len);
+
+    if (ret < 0) {
+        goto err;
+    }
+
+    qemu_iovec_concat(&acb->cur_qiov, acb->qiov, acb->qiov_offset, len);
+
+    /* Handle zero cluster and backing file reads */
+    if (ret == QED_CLUSTER_ZERO) {
+        qemu_iovec_memset(&acb->cur_qiov, 0, 0, acb->cur_qiov.size);
+        qed_aio_next_io(acb, 0);
+        return;
+    } else if (ret != QED_CLUSTER_FOUND) {
+        qed_read_backing_file(s, acb->cur_pos, &acb->cur_qiov,
+                              qed_aio_next_io, acb);
+        return;
+    }
+
+    BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
+    bdrv_aio_readv(bs->file, offset / BDRV_SECTOR_SIZE,
+                   &acb->cur_qiov, acb->cur_qiov.size / BDRV_SECTOR_SIZE,
+                   qed_aio_next_io, acb);
+    return;
+
+err:
+    qed_aio_complete(acb, ret);
+}
+
+/**
+ * Begin next I/O or complete the request
+ */
+static void qed_aio_next_io(void *opaque, int ret)
+{
+    QEDAIOCB *acb = opaque;
+    BDRVQEDState *s = acb_to_s(acb);
+    QEDFindClusterFunc *io_fn = (acb->flags & QED_AIOCB_WRITE) ?
+                                qed_aio_write_data : qed_aio_read_data;
+
+    trace_qed_aio_next_io(s, acb, ret, acb->cur_pos + acb->cur_qiov.size);
+
+    /* Handle I/O error */
+    if (ret) {
+        qed_aio_complete(acb, ret);
+        return;
+    }
+
+    acb->qiov_offset += acb->cur_qiov.size;
+    acb->cur_pos += acb->cur_qiov.size;
+    qemu_iovec_reset(&acb->cur_qiov);
+
+    /* Complete request */
+    if (acb->cur_pos >= acb->end_pos) {
+        qed_aio_complete(acb, 0);
+        return;
+    }
+
+    /* Find next cluster and start I/O */
+    qed_find_cluster(s, &acb->request,
+                      acb->cur_pos, acb->end_pos - acb->cur_pos,
+                      io_fn, acb);
+}
+
+static BlockDriverAIOCB *qed_aio_setup(BlockDriverState *bs,
+                                       int64_t sector_num,
+                                       QEMUIOVector *qiov, int nb_sectors,
+                                       BlockDriverCompletionFunc *cb,
+                                       void *opaque, int flags)
+{
+    QEDAIOCB *acb = qemu_aio_get(&qed_aiocb_info, bs, cb, opaque);
+
+    trace_qed_aio_setup(bs->opaque, acb, sector_num, nb_sectors,
+                        opaque, flags);
+
+    acb->flags = flags;
+    acb->finished = NULL;
+    acb->qiov = qiov;
+    acb->qiov_offset = 0;
+    acb->cur_pos = (uint64_t)sector_num * BDRV_SECTOR_SIZE;
+    acb->end_pos = acb->cur_pos + nb_sectors * BDRV_SECTOR_SIZE;
+    acb->request.l2_table = NULL;
+    qemu_iovec_init(&acb->cur_qiov, qiov->niov);
+
+    /* Start request */
+    qed_aio_next_io(acb, 0);
+    return &acb->common;
+}
+
+static BlockDriverAIOCB *bdrv_qed_aio_readv(BlockDriverState *bs,
+                                            int64_t sector_num,
+                                            QEMUIOVector *qiov, int nb_sectors,
+                                            BlockDriverCompletionFunc *cb,
+                                            void *opaque)
+{
+    return qed_aio_setup(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
+}
+
+static BlockDriverAIOCB *bdrv_qed_aio_writev(BlockDriverState *bs,
+                                             int64_t sector_num,
+                                             QEMUIOVector *qiov, int nb_sectors,
+                                             BlockDriverCompletionFunc *cb,
+                                             void *opaque)
+{
+    return qed_aio_setup(bs, sector_num, qiov, nb_sectors, cb,
+                         opaque, QED_AIOCB_WRITE);
+}
+
+typedef struct {
+    Coroutine *co;
+    int ret;
+    bool done;
+} QEDWriteZeroesCB;
+
+static void coroutine_fn qed_co_write_zeroes_cb(void *opaque, int ret)
+{
+    QEDWriteZeroesCB *cb = opaque;
+
+    cb->done = true;
+    cb->ret = ret;
+    if (cb->co) {
+        qemu_coroutine_enter(cb->co, NULL);
+    }
+}
+
+static int coroutine_fn bdrv_qed_co_write_zeroes(BlockDriverState *bs,
+                                                 int64_t sector_num,
+                                                 int nb_sectors)
+{
+    BlockDriverAIOCB *blockacb;
+    BDRVQEDState *s = bs->opaque;
+    QEDWriteZeroesCB cb = { .done = false };
+    QEMUIOVector qiov;
+    struct iovec iov;
+
+    /* Refuse if there are untouched backing file sectors */
+    if (bs->backing_hd) {
+        if (qed_offset_into_cluster(s, sector_num * BDRV_SECTOR_SIZE) != 0) {
+            return -ENOTSUP;
+        }
+        if (qed_offset_into_cluster(s, nb_sectors * BDRV_SECTOR_SIZE) != 0) {
+            return -ENOTSUP;
+        }
+    }
+
+    /* Zero writes start without an I/O buffer.  If a buffer becomes necessary
+     * then it will be allocated during request processing.
+     */
+    iov.iov_base = NULL,
+    iov.iov_len  = nb_sectors * BDRV_SECTOR_SIZE,
+
+    qemu_iovec_init_external(&qiov, &iov, 1);
+    blockacb = qed_aio_setup(bs, sector_num, &qiov, nb_sectors,
+                             qed_co_write_zeroes_cb, &cb,
+                             QED_AIOCB_WRITE | QED_AIOCB_ZERO);
+    if (!blockacb) {
+        return -EIO;
+    }
+    if (!cb.done) {
+        cb.co = qemu_coroutine_self();
+        qemu_coroutine_yield();
+    }
+    assert(cb.done);
+    return cb.ret;
+}
+
+static int bdrv_qed_truncate(BlockDriverState *bs, int64_t offset)
+{
+    BDRVQEDState *s = bs->opaque;
+    uint64_t old_image_size;
+    int ret;
+
+    if (!qed_is_image_size_valid(offset, s->header.cluster_size,
+                                 s->header.table_size)) {
+        return -EINVAL;
+    }
+
+    /* Shrinking is currently not supported */
+    if ((uint64_t)offset < s->header.image_size) {
+        return -ENOTSUP;
+    }
+
+    old_image_size = s->header.image_size;
+    s->header.image_size = offset;
+    ret = qed_write_header_sync(s);
+    if (ret < 0) {
+        s->header.image_size = old_image_size;
+    }
+    return ret;
+}
+
+static int64_t bdrv_qed_getlength(BlockDriverState *bs)
+{
+    BDRVQEDState *s = bs->opaque;
+    return s->header.image_size;
+}
+
+static int bdrv_qed_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
+{
+    BDRVQEDState *s = bs->opaque;
+
+    memset(bdi, 0, sizeof(*bdi));
+    bdi->cluster_size = s->header.cluster_size;
+    bdi->is_dirty = s->header.features & QED_F_NEED_CHECK;
+    return 0;
+}
+
+static int bdrv_qed_change_backing_file(BlockDriverState *bs,
+                                        const char *backing_file,
+                                        const char *backing_fmt)
+{
+    BDRVQEDState *s = bs->opaque;
+    QEDHeader new_header, le_header;
+    void *buffer;
+    size_t buffer_len, backing_file_len;
+    int ret;
+
+    /* Refuse to set backing filename if unknown compat feature bits are
+     * active.  If the image uses an unknown compat feature then we may not
+     * know the layout of data following the header structure and cannot safely
+     * add a new string.
+     */
+    if (backing_file && (s->header.compat_features &
+                         ~QED_COMPAT_FEATURE_MASK)) {
+        return -ENOTSUP;
+    }
+
+    memcpy(&new_header, &s->header, sizeof(new_header));
+
+    new_header.features &= ~(QED_F_BACKING_FILE |
+                             QED_F_BACKING_FORMAT_NO_PROBE);
+
+    /* Adjust feature flags */
+    if (backing_file) {
+        new_header.features |= QED_F_BACKING_FILE;
+
+        if (qed_fmt_is_raw(backing_fmt)) {
+            new_header.features |= QED_F_BACKING_FORMAT_NO_PROBE;
+        }
+    }
+
+    /* Calculate new header size */
+    backing_file_len = 0;
+
+    if (backing_file) {
+        backing_file_len = strlen(backing_file);
+    }
+
+    buffer_len = sizeof(new_header);
+    new_header.backing_filename_offset = buffer_len;
+    new_header.backing_filename_size = backing_file_len;
+    buffer_len += backing_file_len;
+
+    /* Make sure we can rewrite header without failing */
+    if (buffer_len > new_header.header_size * new_header.cluster_size) {
+        return -ENOSPC;
+    }
+
+    /* Prepare new header */
+    buffer = g_malloc(buffer_len);
+
+    qed_header_cpu_to_le(&new_header, &le_header);
+    memcpy(buffer, &le_header, sizeof(le_header));
+    buffer_len = sizeof(le_header);
+
+    if (backing_file) {
+        memcpy(buffer + buffer_len, backing_file, backing_file_len);
+        buffer_len += backing_file_len;
+    }
+
+    /* Write new header */
+    ret = bdrv_pwrite_sync(bs->file, 0, buffer, buffer_len);
+    g_free(buffer);
+    if (ret == 0) {
+        memcpy(&s->header, &new_header, sizeof(new_header));
+    }
+    return ret;
+}
+
+static void bdrv_qed_invalidate_cache(BlockDriverState *bs)
+{
+    BDRVQEDState *s = bs->opaque;
+
+    bdrv_qed_close(bs);
+    memset(s, 0, sizeof(BDRVQEDState));
+    bdrv_qed_open(bs, NULL, bs->open_flags);
+}
+
+static int bdrv_qed_check(BlockDriverState *bs, BdrvCheckResult *result,
+                          BdrvCheckMode fix)
+{
+    BDRVQEDState *s = bs->opaque;
+
+    return qed_check(s, result, !!fix);
+}
+
+static QEMUOptionParameter qed_create_options[] = {
+    {
+        .name = BLOCK_OPT_SIZE,
+        .type = OPT_SIZE,
+        .help = "Virtual disk size (in bytes)"
+    }, {
+        .name = BLOCK_OPT_BACKING_FILE,
+        .type = OPT_STRING,
+        .help = "File name of a base image"
+    }, {
+        .name = BLOCK_OPT_BACKING_FMT,
+        .type = OPT_STRING,
+        .help = "Image format of the base image"
+    }, {
+        .name = BLOCK_OPT_CLUSTER_SIZE,
+        .type = OPT_SIZE,
+        .help = "Cluster size (in bytes)",
+        .value = { .n = QED_DEFAULT_CLUSTER_SIZE },
+    }, {
+        .name = BLOCK_OPT_TABLE_SIZE,
+        .type = OPT_SIZE,
+        .help = "L1/L2 table size (in clusters)"
+    },
+    { /* end of list */ }
+};
+
+static BlockDriver bdrv_qed = {
+    .format_name              = "qed",
+    .instance_size            = sizeof(BDRVQEDState),
+    .create_options           = qed_create_options,
+
+    .bdrv_probe               = bdrv_qed_probe,
+    .bdrv_rebind              = bdrv_qed_rebind,
+    .bdrv_open                = bdrv_qed_open,
+    .bdrv_close               = bdrv_qed_close,
+    .bdrv_reopen_prepare      = bdrv_qed_reopen_prepare,
+    .bdrv_create              = bdrv_qed_create,
+    .bdrv_has_zero_init       = bdrv_has_zero_init_1,
+    .bdrv_co_is_allocated     = bdrv_qed_co_is_allocated,
+    .bdrv_make_empty          = bdrv_qed_make_empty,
+    .bdrv_aio_readv           = bdrv_qed_aio_readv,
+    .bdrv_aio_writev          = bdrv_qed_aio_writev,
+    .bdrv_co_write_zeroes     = bdrv_qed_co_write_zeroes,
+    .bdrv_truncate            = bdrv_qed_truncate,
+    .bdrv_getlength           = bdrv_qed_getlength,
+    .bdrv_get_info            = bdrv_qed_get_info,
+    .bdrv_change_backing_file = bdrv_qed_change_backing_file,
+    .bdrv_invalidate_cache    = bdrv_qed_invalidate_cache,
+    .bdrv_check               = bdrv_qed_check,
+};
+
+static void bdrv_qed_init(void)
+{
+    bdrv_register(&bdrv_qed);
+}
+
+block_init(bdrv_qed_init);
diff --git a/contrib/qemu/block/qed.h b/contrib/qemu/block/qed.h
new file mode 100644
index 000000000..2b4ddedf3
--- /dev/null
+++ b/contrib/qemu/block/qed.h
@@ -0,0 +1,344 @@
+/*
+ * QEMU Enhanced Disk Format
+ *
+ * Copyright IBM, Corp. 2010
+ *
+ * Authors:
+ *  Stefan Hajnoczi   <stefanha@linux.vnet.ibm.com>
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#ifndef BLOCK_QED_H
+#define BLOCK_QED_H
+
+#include "block/block_int.h"
+
+/* The layout of a QED file is as follows:
+ *
+ * +--------+----------+----------+----------+-----+
+ * | header | L1 table | cluster0 | cluster1 | ... |
+ * +--------+----------+----------+----------+-----+
+ *
+ * There is a 2-level pagetable for cluster allocation:
+ *
+ *                     +----------+
+ *                     | L1 table |
+ *                     +----------+
+ *                ,------'  |  '------.
+ *           +----------+   |    +----------+
+ *           | L2 table |  ...   | L2 table |
+ *           +----------+        +----------+
+ *       ,------'  |  '------.
+ *  +----------+   |    +----------+
+ *  |   Data   |  ...   |   Data   |
+ *  +----------+        +----------+
+ *
+ * The L1 table is fixed size and always present.  L2 tables are allocated on
+ * demand.  The L1 table size determines the maximum possible image size; it
+ * can be influenced using the cluster_size and table_size values.
+ *
+ * All fields are little-endian on disk.
+ */
+
+enum {
+    QED_MAGIC = 'Q' | 'E' << 8 | 'D' << 16 | '\0' << 24,
+
+    /* The image supports a backing file */
+    QED_F_BACKING_FILE = 0x01,
+
+    /* The image needs a consistency check before use */
+    QED_F_NEED_CHECK = 0x02,
+
+    /* The backing file format must not be probed, treat as raw image */
+    QED_F_BACKING_FORMAT_NO_PROBE = 0x04,
+
+    /* Feature bits must be used when the on-disk format changes */
+    QED_FEATURE_MASK = QED_F_BACKING_FILE | /* supported feature bits */
+                       QED_F_NEED_CHECK |
+                       QED_F_BACKING_FORMAT_NO_PROBE,
+    QED_COMPAT_FEATURE_MASK = 0,            /* supported compat feature bits */
+    QED_AUTOCLEAR_FEATURE_MASK = 0,         /* supported autoclear feature bits */
+
+    /* Data is stored in groups of sectors called clusters.  Cluster size must
+     * be large to avoid keeping too much metadata.  I/O requests that have
+     * sub-cluster size will require read-modify-write.
+     */
+    QED_MIN_CLUSTER_SIZE = 4 * 1024, /* in bytes */
+    QED_MAX_CLUSTER_SIZE = 64 * 1024 * 1024,
+    QED_DEFAULT_CLUSTER_SIZE = 64 * 1024,
+
+    /* Allocated clusters are tracked using a 2-level pagetable.  Table size is
+     * a multiple of clusters so large maximum image sizes can be supported
+     * without jacking up the cluster size too much.
+     */
+    QED_MIN_TABLE_SIZE = 1,        /* in clusters */
+    QED_MAX_TABLE_SIZE = 16,
+    QED_DEFAULT_TABLE_SIZE = 4,
+
+    /* Delay to flush and clean image after last allocating write completes */
+    QED_NEED_CHECK_TIMEOUT = 5,    /* in seconds */
+};
+
+typedef struct {
+    uint32_t magic;                 /* QED\0 */
+
+    uint32_t cluster_size;          /* in bytes */
+    uint32_t table_size;            /* for L1 and L2 tables, in clusters */
+    uint32_t header_size;           /* in clusters */
+
+    uint64_t features;              /* format feature bits */
+    uint64_t compat_features;       /* compatible feature bits */
+    uint64_t autoclear_features;    /* self-resetting feature bits */
+
+    uint64_t l1_table_offset;       /* in bytes */
+    uint64_t image_size;            /* total logical image size, in bytes */
+
+    /* if (features & QED_F_BACKING_FILE) */
+    uint32_t backing_filename_offset; /* in bytes from start of header */
+    uint32_t backing_filename_size;   /* in bytes */
+} QEDHeader;
+
+typedef struct {
+    uint64_t offsets[0];            /* in bytes */
+} QEDTable;
+
+/* The L2 cache is a simple write-through cache for L2 structures */
+typedef struct CachedL2Table {
+    QEDTable *table;
+    uint64_t offset;    /* offset=0 indicates an invalidate entry */
+    QTAILQ_ENTRY(CachedL2Table) node;
+    int ref;
+} CachedL2Table;
+
+typedef struct {
+    QTAILQ_HEAD(, CachedL2Table) entries;
+    unsigned int n_entries;
+} L2TableCache;
+
+typedef struct QEDRequest {
+    CachedL2Table *l2_table;
+} QEDRequest;
+
+enum {
+    QED_AIOCB_WRITE = 0x0001,       /* read or write? */
+    QED_AIOCB_ZERO  = 0x0002,       /* zero write, used with QED_AIOCB_WRITE */
+};
+
+typedef struct QEDAIOCB {
+    BlockDriverAIOCB common;
+    QEMUBH *bh;
+    int bh_ret;                     /* final return status for completion bh */
+    QSIMPLEQ_ENTRY(QEDAIOCB) next;  /* next request */
+    int flags;                      /* QED_AIOCB_* bits ORed together */
+    bool *finished;                 /* signal for cancel completion */
+    uint64_t end_pos;               /* request end on block device, in bytes */
+
+    /* User scatter-gather list */
+    QEMUIOVector *qiov;
+    size_t qiov_offset;             /* byte count already processed */
+
+    /* Current cluster scatter-gather list */
+    QEMUIOVector cur_qiov;
+    uint64_t cur_pos;               /* position on block device, in bytes */
+    uint64_t cur_cluster;           /* cluster offset in image file */
+    unsigned int cur_nclusters;     /* number of clusters being accessed */
+    int find_cluster_ret;           /* used for L1/L2 update */
+
+    QEDRequest request;
+} QEDAIOCB;
+
+typedef struct {
+    BlockDriverState *bs;           /* device */
+    uint64_t file_size;             /* length of image file, in bytes */
+
+    QEDHeader header;               /* always cpu-endian */
+    QEDTable *l1_table;
+    L2TableCache l2_cache;          /* l2 table cache */
+    uint32_t table_nelems;
+    uint32_t l1_shift;
+    uint32_t l2_shift;
+    uint32_t l2_mask;
+
+    /* Allocating write request queue */
+    QSIMPLEQ_HEAD(, QEDAIOCB) allocating_write_reqs;
+    bool allocating_write_reqs_plugged;
+
+    /* Periodic flush and clear need check flag */
+    QEMUTimer *need_check_timer;
+} BDRVQEDState;
+
+enum {
+    QED_CLUSTER_FOUND,         /* cluster found */
+    QED_CLUSTER_ZERO,          /* zero cluster found */
+    QED_CLUSTER_L2,            /* cluster missing in L2 */
+    QED_CLUSTER_L1,            /* cluster missing in L1 */
+};
+
+/**
+ * qed_find_cluster() completion callback
+ *
+ * @opaque:     User data for completion callback
+ * @ret:        QED_CLUSTER_FOUND   Success
+ *              QED_CLUSTER_L2      Data cluster unallocated in L2
+ *              QED_CLUSTER_L1      L2 unallocated in L1
+ *              -errno              POSIX error occurred
+ * @offset:     Data cluster offset
+ * @len:        Contiguous bytes starting from cluster offset
+ *
+ * This function is invoked when qed_find_cluster() completes.
+ *
+ * On success ret is QED_CLUSTER_FOUND and offset/len are a contiguous range
+ * in the image file.
+ *
+ * On failure ret is QED_CLUSTER_L2 or QED_CLUSTER_L1 for missing L2 or L1
+ * table offset, respectively.  len is number of contiguous unallocated bytes.
+ */
+typedef void QEDFindClusterFunc(void *opaque, int ret, uint64_t offset, size_t len);
+
+/**
+ * Generic callback for chaining async callbacks
+ */
+typedef struct {
+    BlockDriverCompletionFunc *cb;
+    void *opaque;
+} GenericCB;
+
+void *gencb_alloc(size_t len, BlockDriverCompletionFunc *cb, void *opaque);
+void gencb_complete(void *opaque, int ret);
+
+/**
+ * Header functions
+ */
+int qed_write_header_sync(BDRVQEDState *s);
+
+/**
+ * L2 cache functions
+ */
+void qed_init_l2_cache(L2TableCache *l2_cache);
+void qed_free_l2_cache(L2TableCache *l2_cache);
+CachedL2Table *qed_alloc_l2_cache_entry(L2TableCache *l2_cache);
+void qed_unref_l2_cache_entry(CachedL2Table *entry);
+CachedL2Table *qed_find_l2_cache_entry(L2TableCache *l2_cache, uint64_t offset);
+void qed_commit_l2_cache_entry(L2TableCache *l2_cache, CachedL2Table *l2_table);
+
+/**
+ * Table I/O functions
+ */
+int qed_read_l1_table_sync(BDRVQEDState *s);
+void qed_write_l1_table(BDRVQEDState *s, unsigned int index, unsigned int n,
+                        BlockDriverCompletionFunc *cb, void *opaque);
+int qed_write_l1_table_sync(BDRVQEDState *s, unsigned int index,
+                            unsigned int n);
+int qed_read_l2_table_sync(BDRVQEDState *s, QEDRequest *request,
+                           uint64_t offset);
+void qed_read_l2_table(BDRVQEDState *s, QEDRequest *request, uint64_t offset,
+                       BlockDriverCompletionFunc *cb, void *opaque);
+void qed_write_l2_table(BDRVQEDState *s, QEDRequest *request,
+                        unsigned int index, unsigned int n, bool flush,
+                        BlockDriverCompletionFunc *cb, void *opaque);
+int qed_write_l2_table_sync(BDRVQEDState *s, QEDRequest *request,
+                            unsigned int index, unsigned int n, bool flush);
+
+/**
+ * Cluster functions
+ */
+void qed_find_cluster(BDRVQEDState *s, QEDRequest *request, uint64_t pos,
+                      size_t len, QEDFindClusterFunc *cb, void *opaque);
+
+/**
+ * Consistency check
+ */
+int qed_check(BDRVQEDState *s, BdrvCheckResult *result, bool fix);
+
+QEDTable *qed_alloc_table(BDRVQEDState *s);
+
+/**
+ * Round down to the start of a cluster
+ */
+static inline uint64_t qed_start_of_cluster(BDRVQEDState *s, uint64_t offset)
+{
+    return offset & ~(uint64_t)(s->header.cluster_size - 1);
+}
+
+static inline uint64_t qed_offset_into_cluster(BDRVQEDState *s, uint64_t offset)
+{
+    return offset & (s->header.cluster_size - 1);
+}
+
+static inline uint64_t qed_bytes_to_clusters(BDRVQEDState *s, uint64_t bytes)
+{
+    return qed_start_of_cluster(s, bytes + (s->header.cluster_size - 1)) /
+           (s->header.cluster_size - 1);
+}
+
+static inline unsigned int qed_l1_index(BDRVQEDState *s, uint64_t pos)
+{
+    return pos >> s->l1_shift;
+}
+
+static inline unsigned int qed_l2_index(BDRVQEDState *s, uint64_t pos)
+{
+    return (pos >> s->l2_shift) & s->l2_mask;
+}
+
+/**
+ * Test if a cluster offset is valid
+ */
+static inline bool qed_check_cluster_offset(BDRVQEDState *s, uint64_t offset)
+{
+    uint64_t header_size = (uint64_t)s->header.header_size *
+                           s->header.cluster_size;
+
+    if (offset & (s->header.cluster_size - 1)) {
+        return false;
+    }
+    return offset >= header_size && offset < s->file_size;
+}
+
+/**
+ * Test if a table offset is valid
+ */
+static inline bool qed_check_table_offset(BDRVQEDState *s, uint64_t offset)
+{
+    uint64_t end_offset = offset + (s->header.table_size - 1) *
+                          s->header.cluster_size;
+
+    /* Overflow check */
+    if (end_offset <= offset) {
+        return false;
+    }
+
+    return qed_check_cluster_offset(s, offset) &&
+           qed_check_cluster_offset(s, end_offset);
+}
+
+static inline bool qed_offset_is_cluster_aligned(BDRVQEDState *s,
+                                                 uint64_t offset)
+{
+    if (qed_offset_into_cluster(s, offset)) {
+        return false;
+    }
+    return true;
+}
+
+static inline bool qed_offset_is_unalloc_cluster(uint64_t offset)
+{
+    if (offset == 0) {
+        return true;
+    }
+    return false;
+}
+
+static inline bool qed_offset_is_zero_cluster(uint64_t offset)
+{
+    if (offset == 1) {
+        return true;
+    }
+    return false;
+}
+
+#endif /* BLOCK_QED_H */
diff --git a/contrib/qemu/block/snapshot.c b/contrib/qemu/block/snapshot.c
new file mode 100644
index 000000000..6c6d9deea
--- /dev/null
+++ b/contrib/qemu/block/snapshot.c
@@ -0,0 +1,157 @@
+/*
+ * Block layer snapshot related functions
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "block/snapshot.h"
+#include "block/block_int.h"
+
+int bdrv_snapshot_find(BlockDriverState *bs, QEMUSnapshotInfo *sn_info,
+                       const char *name)
+{
+    QEMUSnapshotInfo *sn_tab, *sn;
+    int nb_sns, i, ret;
+
+    ret = -ENOENT;
+    nb_sns = bdrv_snapshot_list(bs, &sn_tab);
+    if (nb_sns < 0) {
+        return ret;
+    }
+    for (i = 0; i < nb_sns; i++) {
+        sn = &sn_tab[i];
+        if (!strcmp(sn->id_str, name) || !strcmp(sn->name, name)) {
+            *sn_info = *sn;
+            ret = 0;
+            break;
+        }
+    }
+    g_free(sn_tab);
+    return ret;
+}
+
+int bdrv_can_snapshot(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
+        return 0;
+    }
+
+    if (!drv->bdrv_snapshot_create) {
+        if (bs->file != NULL) {
+            return bdrv_can_snapshot(bs->file);
+        }
+        return 0;
+    }
+
+    return 1;
+}
+
+int bdrv_snapshot_create(BlockDriverState *bs,
+                         QEMUSnapshotInfo *sn_info)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (drv->bdrv_snapshot_create) {
+        return drv->bdrv_snapshot_create(bs, sn_info);
+    }
+    if (bs->file) {
+        return bdrv_snapshot_create(bs->file, sn_info);
+    }
+    return -ENOTSUP;
+}
+
+int bdrv_snapshot_goto(BlockDriverState *bs,
+                       const char *snapshot_id)
+{
+    BlockDriver *drv = bs->drv;
+    int ret, open_ret;
+
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (drv->bdrv_snapshot_goto) {
+        return drv->bdrv_snapshot_goto(bs, snapshot_id);
+    }
+
+    if (bs->file) {
+        drv->bdrv_close(bs);
+        ret = bdrv_snapshot_goto(bs->file, snapshot_id);
+        open_ret = drv->bdrv_open(bs, NULL, bs->open_flags);
+        if (open_ret < 0) {
+            bdrv_delete(bs->file);
+            bs->drv = NULL;
+            return open_ret;
+        }
+        return ret;
+    }
+
+    return -ENOTSUP;
+}
+
+int bdrv_snapshot_delete(BlockDriverState *bs, const char *snapshot_id)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (drv->bdrv_snapshot_delete) {
+        return drv->bdrv_snapshot_delete(bs, snapshot_id);
+    }
+    if (bs->file) {
+        return bdrv_snapshot_delete(bs->file, snapshot_id);
+    }
+    return -ENOTSUP;
+}
+
+int bdrv_snapshot_list(BlockDriverState *bs,
+                       QEMUSnapshotInfo **psn_info)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (drv->bdrv_snapshot_list) {
+        return drv->bdrv_snapshot_list(bs, psn_info);
+    }
+    if (bs->file) {
+        return bdrv_snapshot_list(bs->file, psn_info);
+    }
+    return -ENOTSUP;
+}
+
+int bdrv_snapshot_load_tmp(BlockDriverState *bs,
+        const char *snapshot_name)
+{
+    BlockDriver *drv = bs->drv;
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (!bs->read_only) {
+        return -EINVAL;
+    }
+    if (drv->bdrv_snapshot_load_tmp) {
+        return drv->bdrv_snapshot_load_tmp(bs, snapshot_name);
+    }
+    return -ENOTSUP;
+}
diff --git a/contrib/qemu/config-host.h b/contrib/qemu/config-host.h
new file mode 100644
index 000000000..874b04053
--- /dev/null
+++ b/contrib/qemu/config-host.h
@@ -0,0 +1,74 @@
+/* Automatically generated by create_config - do not modify */
+#define CONFIG_QEMU_CONFDIR "/usr/local/etc/qemu"
+#define CONFIG_QEMU_DATADIR "/usr/local/share/qemu"
+#define CONFIG_QEMU_DOCDIR "/usr/local/share/doc/qemu"
+#define CONFIG_QEMU_LOCALSTATEDIR "/usr/local/var"
+#define CONFIG_QEMU_HELPERDIR "/usr/local/libexec"
+#define CONFIG_QEMU_LOCALEDIR "/usr/local/share/locale"
+#define HOST_X86_64 1
+#define CONFIG_QEMU_LDST_OPTIMIZATION 1
+#define CONFIG_POSIX 1
+#define CONFIG_LINUX 1
+#define CONFIG_SLIRP 1
+#define CONFIG_SMBD_COMMAND "/usr/sbin/smbd"
+#define CONFIG_AUDIO_DRIVERS \
+    &oss_audio_driver,\
+
+#define CONFIG_OSS 1
+#define CONFIG_BDRV_RW_WHITELIST\
+    NULL
+#define CONFIG_BDRV_RO_WHITELIST\
+    NULL
+#define CONFIG_VNC 1
+#define CONFIG_VNC_TLS 1
+#define CONFIG_VNC_SASL 1
+#define CONFIG_VNC_WS 1
+#define CONFIG_FNMATCH 1
+#define CONFIG_UUID 1
+#define CONFIG_XFS 1
+#define QEMU_VERSION "1.5.50"
+#define QEMU_PKGVERSION ""
+#define CONFIG_CURSES 1
+#define CONFIG_UTIMENSAT 1
+#define CONFIG_PIPE2 1
+#define CONFIG_ACCEPT4 1
+#define CONFIG_SPLICE 1
+#define CONFIG_EVENTFD 1
+#define CONFIG_FALLOCATE 1
+#define CONFIG_FALLOCATE_PUNCH_HOLE 1
+#define CONFIG_SYNC_FILE_RANGE 1
+#define CONFIG_FIEMAP 1
+#define CONFIG_DUP3 1
+#define CONFIG_EPOLL 1
+#define CONFIG_EPOLL_CREATE1 1
+#define CONFIG_EPOLL_PWAIT 1
+#define CONFIG_SENDFILE 1
+#define CONFIG_INOTIFY 1
+#define CONFIG_INOTIFY1 1
+#define CONFIG_BYTESWAP_H 1
+#define CONFIG_CURL 1
+#define CONFIG_LINUX_AIO 1
+#define CONFIG_ATTR 1
+#define CONFIG_VHOST_SCSI 1
+#define CONFIG_IOVEC 1
+#define CONFIG_PREADV 1
+#define CONFIG_FDT 1
+#define CONFIG_SIGNALFD 1
+#define CONFIG_FDATASYNC 1
+#define CONFIG_MADVISE 1
+#define CONFIG_POSIX_MADVISE 1
+#define CONFIG_SIGEV_THREAD_ID 1
+#define CONFIG_UNAME_RELEASE ""
+#define CONFIG_QOM_CAST_DEBUG 1
+#define CONFIG_COROUTINE_BACKEND ucontext
+#define CONFIG_OPEN_BY_HANDLE 1
+#define CONFIG_LINUX_MAGIC_H 1
+#define CONFIG_PRAGMA_DIAGNOSTIC_AVAILABLE 1
+#define CONFIG_VALGRIND_H 1
+#define CONFIG_HAS_ENVIRON 1
+#define CONFIG_CPUID_H 1
+#define CONFIG_INT128 1
+#define CONFIG_VIRTIO_BLK_DATA_PLANE $(CONFIG_VIRTIO)
+#define CONFIG_TRACE_NOP 1
+#define CONFIG_TRACE_FILE trace
+#define CONFIG_TRACE_DEFAULT 1
diff --git a/contrib/qemu/include/block/aio.h b/contrib/qemu/include/block/aio.h
new file mode 100644
index 000000000..183679374
--- /dev/null
+++ b/contrib/qemu/include/block/aio.h
@@ -0,0 +1,247 @@
+/*
+ * QEMU aio implementation
+ *
+ * Copyright IBM, Corp. 2008
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef QEMU_AIO_H
+#define QEMU_AIO_H
+
+#include "qemu-common.h"
+#include "qemu/queue.h"
+#include "qemu/event_notifier.h"
+
+typedef struct BlockDriverAIOCB BlockDriverAIOCB;
+typedef void BlockDriverCompletionFunc(void *opaque, int ret);
+
+typedef struct AIOCBInfo {
+    void (*cancel)(BlockDriverAIOCB *acb);
+    size_t aiocb_size;
+} AIOCBInfo;
+
+struct BlockDriverAIOCB {
+    const AIOCBInfo *aiocb_info;
+    BlockDriverState *bs;
+    BlockDriverCompletionFunc *cb;
+    void *opaque;
+};
+
+void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
+                   BlockDriverCompletionFunc *cb, void *opaque);
+void qemu_aio_release(void *p);
+
+typedef struct AioHandler AioHandler;
+typedef void QEMUBHFunc(void *opaque);
+typedef void IOHandler(void *opaque);
+
+typedef struct AioContext {
+    GSource source;
+
+    /* The list of registered AIO handlers */
+    QLIST_HEAD(, AioHandler) aio_handlers;
+
+    /* This is a simple lock used to protect the aio_handlers list.
+     * Specifically, it's used to ensure that no callbacks are removed while
+     * we're walking and dispatching callbacks.
+     */
+    int walking_handlers;
+
+    /* Anchor of the list of Bottom Halves belonging to the context */
+    struct QEMUBH *first_bh;
+
+    /* A simple lock used to protect the first_bh list, and ensure that
+     * no callbacks are removed while we're walking and dispatching callbacks.
+     */
+    int walking_bh;
+
+    /* Used for aio_notify.  */
+    EventNotifier notifier;
+
+    /* GPollFDs for aio_poll() */
+    GArray *pollfds;
+
+    /* Thread pool for performing work and receiving completion callbacks */
+    struct ThreadPool *thread_pool;
+} AioContext;
+
+/* Returns 1 if there are still outstanding AIO requests; 0 otherwise */
+typedef int (AioFlushEventNotifierHandler)(EventNotifier *e);
+
+/**
+ * aio_context_new: Allocate a new AioContext.
+ *
+ * AioContext provide a mini event-loop that can be waited on synchronously.
+ * They also provide bottom halves, a service to execute a piece of code
+ * as soon as possible.
+ */
+AioContext *aio_context_new(void);
+
+/**
+ * aio_context_ref:
+ * @ctx: The AioContext to operate on.
+ *
+ * Add a reference to an AioContext.
+ */
+void aio_context_ref(AioContext *ctx);
+
+/**
+ * aio_context_unref:
+ * @ctx: The AioContext to operate on.
+ *
+ * Drop a reference to an AioContext.
+ */
+void aio_context_unref(AioContext *ctx);
+
+/**
+ * aio_bh_new: Allocate a new bottom half structure.
+ *
+ * Bottom halves are lightweight callbacks whose invocation is guaranteed
+ * to be wait-free, thread-safe and signal-safe.  The #QEMUBH structure
+ * is opaque and must be allocated prior to its use.
+ */
+QEMUBH *aio_bh_new(AioContext *ctx, QEMUBHFunc *cb, void *opaque);
+
+/**
+ * aio_notify: Force processing of pending events.
+ *
+ * Similar to signaling a condition variable, aio_notify forces
+ * aio_wait to exit, so that the next call will re-examine pending events.
+ * The caller of aio_notify will usually call aio_wait again very soon,
+ * or go through another iteration of the GLib main loop.  Hence, aio_notify
+ * also has the side effect of recalculating the sets of file descriptors
+ * that the main loop waits for.
+ *
+ * Calling aio_notify is rarely necessary, because for example scheduling
+ * a bottom half calls it already.
+ */
+void aio_notify(AioContext *ctx);
+
+/**
+ * aio_bh_poll: Poll bottom halves for an AioContext.
+ *
+ * These are internal functions used by the QEMU main loop.
+ */
+int aio_bh_poll(AioContext *ctx);
+
+/**
+ * qemu_bh_schedule: Schedule a bottom half.
+ *
+ * Scheduling a bottom half interrupts the main loop and causes the
+ * execution of the callback that was passed to qemu_bh_new.
+ *
+ * Bottom halves that are scheduled from a bottom half handler are instantly
+ * invoked.  This can create an infinite loop if a bottom half handler
+ * schedules itself.
+ *
+ * @bh: The bottom half to be scheduled.
+ */
+void qemu_bh_schedule(QEMUBH *bh);
+
+/**
+ * qemu_bh_cancel: Cancel execution of a bottom half.
+ *
+ * Canceling execution of a bottom half undoes the effect of calls to
+ * qemu_bh_schedule without freeing its resources yet.  While cancellation
+ * itself is also wait-free and thread-safe, it can of course race with the
+ * loop that executes bottom halves unless you are holding the iothread
+ * mutex.  This makes it mostly useless if you are not holding the mutex.
+ *
+ * @bh: The bottom half to be canceled.
+ */
+void qemu_bh_cancel(QEMUBH *bh);
+
+/**
+ *qemu_bh_delete: Cancel execution of a bottom half and free its resources.
+ *
+ * Deleting a bottom half frees the memory that was allocated for it by
+ * qemu_bh_new.  It also implies canceling the bottom half if it was
+ * scheduled.
+ *
+ * @bh: The bottom half to be deleted.
+ */
+void qemu_bh_delete(QEMUBH *bh);
+
+/* Return whether there are any pending callbacks from the GSource
+ * attached to the AioContext.
+ *
+ * This is used internally in the implementation of the GSource.
+ */
+bool aio_pending(AioContext *ctx);
+
+/* Progress in completing AIO work to occur.  This can issue new pending
+ * aio as a result of executing I/O completion or bh callbacks.
+ *
+ * If there is no pending AIO operation or completion (bottom half),
+ * return false.  If there are pending AIO operations of bottom halves,
+ * return true.
+ *
+ * If there are no pending bottom halves, but there are pending AIO
+ * operations, it may not be possible to make any progress without
+ * blocking.  If @blocking is true, this function will wait until one
+ * or more AIO events have completed, to ensure something has moved
+ * before returning.
+ */
+bool aio_poll(AioContext *ctx, bool blocking);
+
+#ifdef CONFIG_POSIX
+/* Returns 1 if there are still outstanding AIO requests; 0 otherwise */
+typedef int (AioFlushHandler)(void *opaque);
+
+/* Register a file descriptor and associated callbacks.  Behaves very similarly
+ * to qemu_set_fd_handler2.  Unlike qemu_set_fd_handler2, these callbacks will
+ * be invoked when using qemu_aio_wait().
+ *
+ * Code that invokes AIO completion functions should rely on this function
+ * instead of qemu_set_fd_handler[2].
+ */
+void aio_set_fd_handler(AioContext *ctx,
+                        int fd,
+                        IOHandler *io_read,
+                        IOHandler *io_write,
+                        AioFlushHandler *io_flush,
+                        void *opaque);
+#endif
+
+/* Register an event notifier and associated callbacks.  Behaves very similarly
+ * to event_notifier_set_handler.  Unlike event_notifier_set_handler, these callbacks
+ * will be invoked when using qemu_aio_wait().
+ *
+ * Code that invokes AIO completion functions should rely on this function
+ * instead of event_notifier_set_handler.
+ */
+void aio_set_event_notifier(AioContext *ctx,
+                            EventNotifier *notifier,
+                            EventNotifierHandler *io_read,
+                            AioFlushEventNotifierHandler *io_flush);
+
+/* Return a GSource that lets the main loop poll the file descriptors attached
+ * to this AioContext.
+ */
+GSource *aio_get_g_source(AioContext *ctx);
+
+/* Return the ThreadPool bound to this AioContext */
+struct ThreadPool *aio_get_thread_pool(AioContext *ctx);
+
+/* Functions to operate on the main QEMU AioContext.  */
+
+bool qemu_aio_wait(void);
+void qemu_aio_set_event_notifier(EventNotifier *notifier,
+                                 EventNotifierHandler *io_read,
+                                 AioFlushEventNotifierHandler *io_flush);
+
+#ifdef CONFIG_POSIX
+void qemu_aio_set_fd_handler(int fd,
+                             IOHandler *io_read,
+                             IOHandler *io_write,
+                             AioFlushHandler *io_flush,
+                             void *opaque);
+#endif
+
+#endif
diff --git a/contrib/qemu/include/block/block.h b/contrib/qemu/include/block/block.h
new file mode 100644
index 000000000..b6b9014a9
--- /dev/null
+++ b/contrib/qemu/include/block/block.h
@@ -0,0 +1,443 @@
+#ifndef BLOCK_H
+#define BLOCK_H
+
+#include "block/aio.h"
+#include "qemu-common.h"
+#include "qemu/option.h"
+#include "block/coroutine.h"
+#include "qapi/qmp/qobject.h"
+#include "qapi-types.h"
+
+/* block.c */
+typedef struct BlockDriver BlockDriver;
+typedef struct BlockJob BlockJob;
+
+typedef struct BlockDriverInfo {
+    /* in bytes, 0 if irrelevant */
+    int cluster_size;
+    /* offset at which the VM state can be saved (0 if not possible) */
+    int64_t vm_state_offset;
+    bool is_dirty;
+} BlockDriverInfo;
+
+typedef struct BlockFragInfo {
+    uint64_t allocated_clusters;
+    uint64_t total_clusters;
+    uint64_t fragmented_clusters;
+    uint64_t compressed_clusters;
+} BlockFragInfo;
+
+/* Callbacks for block device models */
+typedef struct BlockDevOps {
+    /*
+     * Runs when virtual media changed (monitor commands eject, change)
+     * Argument load is true on load and false on eject.
+     * Beware: doesn't run when a host device's physical media
+     * changes.  Sure would be useful if it did.
+     * Device models with removable media must implement this callback.
+     */
+    void (*change_media_cb)(void *opaque, bool load);
+    /*
+     * Runs when an eject request is issued from the monitor, the tray
+     * is closed, and the medium is locked.
+     * Device models that do not implement is_medium_locked will not need
+     * this callback.  Device models that can lock the medium or tray might
+     * want to implement the callback and unlock the tray when "force" is
+     * true, even if they do not support eject requests.
+     */
+    void (*eject_request_cb)(void *opaque, bool force);
+    /*
+     * Is the virtual tray open?
+     * Device models implement this only when the device has a tray.
+     */
+    bool (*is_tray_open)(void *opaque);
+    /*
+     * Is the virtual medium locked into the device?
+     * Device models implement this only when device has such a lock.
+     */
+    bool (*is_medium_locked)(void *opaque);
+    /*
+     * Runs when the size changed (e.g. monitor command block_resize)
+     */
+    void (*resize_cb)(void *opaque);
+} BlockDevOps;
+
+#define BDRV_O_RDWR        0x0002
+#define BDRV_O_SNAPSHOT    0x0008 /* open the file read only and save writes in a snapshot */
+#define BDRV_O_NOCACHE     0x0020 /* do not use the host page cache */
+#define BDRV_O_CACHE_WB    0x0040 /* use write-back caching */
+#define BDRV_O_NATIVE_AIO  0x0080 /* use native AIO instead of the thread pool */
+#define BDRV_O_NO_BACKING  0x0100 /* don't open the backing file */
+#define BDRV_O_NO_FLUSH    0x0200 /* disable flushing on this disk */
+#define BDRV_O_COPY_ON_READ 0x0400 /* copy read backing sectors into image */
+#define BDRV_O_INCOMING    0x0800  /* consistency hint for incoming migration */
+#define BDRV_O_CHECK       0x1000  /* open solely for consistency check */
+#define BDRV_O_ALLOW_RDWR  0x2000  /* allow reopen to change from r/o to r/w */
+#define BDRV_O_UNMAP       0x4000  /* execute guest UNMAP/TRIM operations */
+
+#define BDRV_O_CACHE_MASK  (BDRV_O_NOCACHE | BDRV_O_CACHE_WB | BDRV_O_NO_FLUSH)
+
+#define BDRV_SECTOR_BITS   9
+#define BDRV_SECTOR_SIZE   (1ULL << BDRV_SECTOR_BITS)
+#define BDRV_SECTOR_MASK   ~(BDRV_SECTOR_SIZE - 1)
+
+typedef enum {
+    BDRV_ACTION_REPORT, BDRV_ACTION_IGNORE, BDRV_ACTION_STOP
+} BlockErrorAction;
+
+typedef QSIMPLEQ_HEAD(BlockReopenQueue, BlockReopenQueueEntry) BlockReopenQueue;
+
+typedef struct BDRVReopenState {
+    BlockDriverState *bs;
+    int flags;
+    void *opaque;
+} BDRVReopenState;
+
+
+void bdrv_iostatus_enable(BlockDriverState *bs);
+void bdrv_iostatus_reset(BlockDriverState *bs);
+void bdrv_iostatus_disable(BlockDriverState *bs);
+bool bdrv_iostatus_is_enabled(const BlockDriverState *bs);
+void bdrv_iostatus_set_err(BlockDriverState *bs, int error);
+void bdrv_info_print(Monitor *mon, const QObject *data);
+void bdrv_info(Monitor *mon, QObject **ret_data);
+void bdrv_stats_print(Monitor *mon, const QObject *data);
+void bdrv_info_stats(Monitor *mon, QObject **ret_data);
+
+/* disk I/O throttling */
+void bdrv_io_limits_enable(BlockDriverState *bs);
+void bdrv_io_limits_disable(BlockDriverState *bs);
+bool bdrv_io_limits_enabled(BlockDriverState *bs);
+
+void bdrv_init(void);
+void bdrv_init_with_whitelist(void);
+BlockDriver *bdrv_find_protocol(const char *filename,
+                                bool allow_protocol_prefix);
+BlockDriver *bdrv_find_format(const char *format_name);
+BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
+                                          bool readonly);
+int bdrv_create(BlockDriver *drv, const char* filename,
+    QEMUOptionParameter *options);
+int bdrv_create_file(const char* filename, QEMUOptionParameter *options);
+BlockDriverState *bdrv_new(const char *device_name);
+void bdrv_make_anon(BlockDriverState *bs);
+void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old);
+void bdrv_append(BlockDriverState *bs_new, BlockDriverState *bs_top);
+void bdrv_delete(BlockDriverState *bs);
+int bdrv_parse_cache_flags(const char *mode, int *flags);
+int bdrv_parse_discard_flags(const char *mode, int *flags);
+int bdrv_file_open(BlockDriverState **pbs, const char *filename,
+                   QDict *options, int flags);
+int bdrv_open_backing_file(BlockDriverState *bs, QDict *options);
+int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
+              int flags, BlockDriver *drv);
+BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue,
+                                    BlockDriverState *bs, int flags);
+int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp);
+int bdrv_reopen(BlockDriverState *bs, int bdrv_flags, Error **errp);
+int bdrv_reopen_prepare(BDRVReopenState *reopen_state,
+                        BlockReopenQueue *queue, Error **errp);
+void bdrv_reopen_commit(BDRVReopenState *reopen_state);
+void bdrv_reopen_abort(BDRVReopenState *reopen_state);
+void bdrv_close(BlockDriverState *bs);
+void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify);
+int bdrv_attach_dev(BlockDriverState *bs, void *dev);
+void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev);
+void bdrv_detach_dev(BlockDriverState *bs, void *dev);
+void *bdrv_get_attached_dev(BlockDriverState *bs);
+void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
+                      void *opaque);
+void bdrv_dev_eject_request(BlockDriverState *bs, bool force);
+bool bdrv_dev_has_removable_media(BlockDriverState *bs);
+bool bdrv_dev_is_tray_open(BlockDriverState *bs);
+bool bdrv_dev_is_medium_locked(BlockDriverState *bs);
+int bdrv_read(BlockDriverState *bs, int64_t sector_num,
+              uint8_t *buf, int nb_sectors);
+int bdrv_read_unthrottled(BlockDriverState *bs, int64_t sector_num,
+                          uint8_t *buf, int nb_sectors);
+int bdrv_write(BlockDriverState *bs, int64_t sector_num,
+               const uint8_t *buf, int nb_sectors);
+int bdrv_writev(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov);
+int bdrv_pread(BlockDriverState *bs, int64_t offset,
+               void *buf, int count);
+int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
+                const void *buf, int count);
+int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov);
+int bdrv_pwrite_sync(BlockDriverState *bs, int64_t offset,
+    const void *buf, int count);
+int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
+    int nb_sectors, QEMUIOVector *qiov);
+int coroutine_fn bdrv_co_copy_on_readv(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov);
+int coroutine_fn bdrv_co_writev(BlockDriverState *bs, int64_t sector_num,
+    int nb_sectors, QEMUIOVector *qiov);
+/*
+ * Efficiently zero a region of the disk image.  Note that this is a regular
+ * I/O request like read or write and should have a reasonable size.  This
+ * function is not suitable for zeroing the entire image in a single request
+ * because it may allocate memory for the entire region.
+ */
+int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs, int64_t sector_num,
+    int nb_sectors);
+int coroutine_fn bdrv_co_is_allocated(BlockDriverState *bs, int64_t sector_num,
+    int nb_sectors, int *pnum);
+int coroutine_fn bdrv_co_is_allocated_above(BlockDriverState *top,
+                                            BlockDriverState *base,
+                                            int64_t sector_num,
+                                            int nb_sectors, int *pnum);
+BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
+    const char *backing_file);
+int bdrv_get_backing_file_depth(BlockDriverState *bs);
+int bdrv_truncate(BlockDriverState *bs, int64_t offset);
+int64_t bdrv_getlength(BlockDriverState *bs);
+int64_t bdrv_get_allocated_file_size(BlockDriverState *bs);
+void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr);
+int bdrv_commit(BlockDriverState *bs);
+int bdrv_commit_all(void);
+int bdrv_change_backing_file(BlockDriverState *bs,
+    const char *backing_file, const char *backing_fmt);
+void bdrv_register(BlockDriver *bdrv);
+int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
+                           BlockDriverState *base);
+BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
+                                    BlockDriverState *bs);
+BlockDriverState *bdrv_find_base(BlockDriverState *bs);
+
+
+typedef struct BdrvCheckResult {
+    int corruptions;
+    int leaks;
+    int check_errors;
+    int corruptions_fixed;
+    int leaks_fixed;
+    int64_t image_end_offset;
+    BlockFragInfo bfi;
+} BdrvCheckResult;
+
+typedef enum {
+    BDRV_FIX_LEAKS    = 1,
+    BDRV_FIX_ERRORS   = 2,
+} BdrvCheckMode;
+
+int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix);
+
+/* async block I/O */
+typedef void BlockDriverDirtyHandler(BlockDriverState *bs, int64_t sector,
+                                     int sector_num);
+BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
+                                 QEMUIOVector *iov, int nb_sectors,
+                                 BlockDriverCompletionFunc *cb, void *opaque);
+BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
+                                  QEMUIOVector *iov, int nb_sectors,
+                                  BlockDriverCompletionFunc *cb, void *opaque);
+BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
+                                 BlockDriverCompletionFunc *cb, void *opaque);
+BlockDriverAIOCB *bdrv_aio_discard(BlockDriverState *bs,
+                                   int64_t sector_num, int nb_sectors,
+                                   BlockDriverCompletionFunc *cb, void *opaque);
+void bdrv_aio_cancel(BlockDriverAIOCB *acb);
+
+typedef struct BlockRequest {
+    /* Fields to be filled by multiwrite caller */
+    int64_t sector;
+    int nb_sectors;
+    QEMUIOVector *qiov;
+    BlockDriverCompletionFunc *cb;
+    void *opaque;
+
+    /* Filled by multiwrite implementation */
+    int error;
+} BlockRequest;
+
+int bdrv_aio_multiwrite(BlockDriverState *bs, BlockRequest *reqs,
+    int num_reqs);
+
+/* sg packet commands */
+int bdrv_ioctl(BlockDriverState *bs, unsigned long int req, void *buf);
+BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
+        unsigned long int req, void *buf,
+        BlockDriverCompletionFunc *cb, void *opaque);
+
+/* Invalidate any cached metadata used by image formats */
+void bdrv_invalidate_cache(BlockDriverState *bs);
+void bdrv_invalidate_cache_all(void);
+
+void bdrv_clear_incoming_migration_all(void);
+
+/* Ensure contents are flushed to disk.  */
+int bdrv_flush(BlockDriverState *bs);
+int coroutine_fn bdrv_co_flush(BlockDriverState *bs);
+int bdrv_flush_all(void);
+void bdrv_close_all(void);
+void bdrv_drain_all(void);
+
+int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors);
+int bdrv_co_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors);
+int bdrv_has_zero_init_1(BlockDriverState *bs);
+int bdrv_has_zero_init(BlockDriverState *bs);
+int bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num, int nb_sectors,
+                      int *pnum);
+int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base,
+                            int64_t sector_num, int nb_sectors, int *pnum);
+
+void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
+                       BlockdevOnError on_write_error);
+BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read);
+BlockErrorAction bdrv_get_error_action(BlockDriverState *bs, bool is_read, int error);
+void bdrv_error_action(BlockDriverState *bs, BlockErrorAction action,
+                       bool is_read, int error);
+int bdrv_is_read_only(BlockDriverState *bs);
+int bdrv_is_sg(BlockDriverState *bs);
+int bdrv_enable_write_cache(BlockDriverState *bs);
+void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce);
+int bdrv_is_inserted(BlockDriverState *bs);
+int bdrv_media_changed(BlockDriverState *bs);
+void bdrv_lock_medium(BlockDriverState *bs, bool locked);
+void bdrv_eject(BlockDriverState *bs, bool eject_flag);
+const char *bdrv_get_format_name(BlockDriverState *bs);
+BlockDriverState *bdrv_find(const char *name);
+BlockDriverState *bdrv_next(BlockDriverState *bs);
+void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs),
+                  void *opaque);
+int bdrv_is_encrypted(BlockDriverState *bs);
+int bdrv_key_required(BlockDriverState *bs);
+int bdrv_set_key(BlockDriverState *bs, const char *key);
+int bdrv_query_missing_keys(void);
+void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
+                         void *opaque);
+const char *bdrv_get_device_name(BlockDriverState *bs);
+int bdrv_get_flags(BlockDriverState *bs);
+int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
+                          const uint8_t *buf, int nb_sectors);
+int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi);
+void bdrv_round_to_clusters(BlockDriverState *bs,
+                            int64_t sector_num, int nb_sectors,
+                            int64_t *cluster_sector_num,
+                            int *cluster_nb_sectors);
+
+const char *bdrv_get_encrypted_filename(BlockDriverState *bs);
+void bdrv_get_backing_filename(BlockDriverState *bs,
+                               char *filename, int filename_size);
+void bdrv_get_full_backing_filename(BlockDriverState *bs,
+                                    char *dest, size_t sz);
+int bdrv_is_snapshot(BlockDriverState *bs);
+
+int path_is_absolute(const char *path);
+void path_combine(char *dest, int dest_size,
+                  const char *base_path,
+                  const char *filename);
+
+int bdrv_writev_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos);
+int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
+                      int64_t pos, int size);
+
+int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
+                      int64_t pos, int size);
+
+void bdrv_img_create(const char *filename, const char *fmt,
+                     const char *base_filename, const char *base_fmt,
+                     char *options, uint64_t img_size, int flags,
+                     Error **errp, bool quiet);
+
+void bdrv_set_buffer_alignment(BlockDriverState *bs, int align);
+void *qemu_blockalign(BlockDriverState *bs, size_t size);
+bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov);
+
+struct HBitmapIter;
+void bdrv_set_dirty_tracking(BlockDriverState *bs, int granularity);
+int bdrv_get_dirty(BlockDriverState *bs, int64_t sector);
+void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector, int nr_sectors);
+void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector, int nr_sectors);
+void bdrv_dirty_iter_init(BlockDriverState *bs, struct HBitmapIter *hbi);
+int64_t bdrv_get_dirty_count(BlockDriverState *bs);
+
+void bdrv_enable_copy_on_read(BlockDriverState *bs);
+void bdrv_disable_copy_on_read(BlockDriverState *bs);
+
+void bdrv_set_in_use(BlockDriverState *bs, int in_use);
+int bdrv_in_use(BlockDriverState *bs);
+
+#ifdef CONFIG_LINUX_AIO
+int raw_get_aio_fd(BlockDriverState *bs);
+#else
+static inline int raw_get_aio_fd(BlockDriverState *bs)
+{
+    return -ENOTSUP;
+}
+#endif
+
+enum BlockAcctType {
+    BDRV_ACCT_READ,
+    BDRV_ACCT_WRITE,
+    BDRV_ACCT_FLUSH,
+    BDRV_MAX_IOTYPE,
+};
+
+typedef struct BlockAcctCookie {
+    int64_t bytes;
+    int64_t start_time_ns;
+    enum BlockAcctType type;
+} BlockAcctCookie;
+
+void bdrv_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie,
+        int64_t bytes, enum BlockAcctType type);
+void bdrv_acct_done(BlockDriverState *bs, BlockAcctCookie *cookie);
+
+typedef enum {
+    BLKDBG_L1_UPDATE,
+
+    BLKDBG_L1_GROW_ALLOC_TABLE,
+    BLKDBG_L1_GROW_WRITE_TABLE,
+    BLKDBG_L1_GROW_ACTIVATE_TABLE,
+
+    BLKDBG_L2_LOAD,
+    BLKDBG_L2_UPDATE,
+    BLKDBG_L2_UPDATE_COMPRESSED,
+    BLKDBG_L2_ALLOC_COW_READ,
+    BLKDBG_L2_ALLOC_WRITE,
+
+    BLKDBG_READ_AIO,
+    BLKDBG_READ_BACKING_AIO,
+    BLKDBG_READ_COMPRESSED,
+
+    BLKDBG_WRITE_AIO,
+    BLKDBG_WRITE_COMPRESSED,
+
+    BLKDBG_VMSTATE_LOAD,
+    BLKDBG_VMSTATE_SAVE,
+
+    BLKDBG_COW_READ,
+    BLKDBG_COW_WRITE,
+
+    BLKDBG_REFTABLE_LOAD,
+    BLKDBG_REFTABLE_GROW,
+
+    BLKDBG_REFBLOCK_LOAD,
+    BLKDBG_REFBLOCK_UPDATE,
+    BLKDBG_REFBLOCK_UPDATE_PART,
+    BLKDBG_REFBLOCK_ALLOC,
+    BLKDBG_REFBLOCK_ALLOC_HOOKUP,
+    BLKDBG_REFBLOCK_ALLOC_WRITE,
+    BLKDBG_REFBLOCK_ALLOC_WRITE_BLOCKS,
+    BLKDBG_REFBLOCK_ALLOC_WRITE_TABLE,
+    BLKDBG_REFBLOCK_ALLOC_SWITCH_TABLE,
+
+    BLKDBG_CLUSTER_ALLOC,
+    BLKDBG_CLUSTER_ALLOC_BYTES,
+    BLKDBG_CLUSTER_FREE,
+
+    BLKDBG_FLUSH_TO_OS,
+    BLKDBG_FLUSH_TO_DISK,
+
+    BLKDBG_EVENT_MAX,
+} BlkDebugEvent;
+
+#define BLKDBG_EVENT(bs, evt) bdrv_debug_event(bs, evt)
+void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event);
+
+int bdrv_debug_breakpoint(BlockDriverState *bs, const char *event,
+                           const char *tag);
+int bdrv_debug_resume(BlockDriverState *bs, const char *tag);
+bool bdrv_debug_is_suspended(BlockDriverState *bs, const char *tag);
+
+#endif
diff --git a/contrib/qemu/include/block/block_int.h b/contrib/qemu/include/block/block_int.h
new file mode 100644
index 000000000..c6ac871e2
--- /dev/null
+++ b/contrib/qemu/include/block/block_int.h
@@ -0,0 +1,421 @@
+/*
+ * QEMU System Emulator block driver
+ *
+ * Copyright (c) 2003 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#ifndef BLOCK_INT_H
+#define BLOCK_INT_H
+
+#include "block/block.h"
+#include "qemu/option.h"
+#include "qemu/queue.h"
+#include "block/coroutine.h"
+#include "qemu/timer.h"
+#include "qapi-types.h"
+#include "qapi/qmp/qerror.h"
+#include "monitor/monitor.h"
+#include "qemu/hbitmap.h"
+#include "block/snapshot.h"
+
+#define BLOCK_FLAG_ENCRYPT          1
+#define BLOCK_FLAG_COMPAT6          4
+#define BLOCK_FLAG_LAZY_REFCOUNTS   8
+
+#define BLOCK_IO_LIMIT_READ     0
+#define BLOCK_IO_LIMIT_WRITE    1
+#define BLOCK_IO_LIMIT_TOTAL    2
+
+#define BLOCK_IO_SLICE_TIME     100000000
+#define NANOSECONDS_PER_SECOND  1000000000.0
+
+#define BLOCK_OPT_SIZE              "size"
+#define BLOCK_OPT_ENCRYPT           "encryption"
+#define BLOCK_OPT_COMPAT6           "compat6"
+#define BLOCK_OPT_BACKING_FILE      "backing_file"
+#define BLOCK_OPT_BACKING_FMT       "backing_fmt"
+#define BLOCK_OPT_CLUSTER_SIZE      "cluster_size"
+#define BLOCK_OPT_TABLE_SIZE        "table_size"
+#define BLOCK_OPT_PREALLOC          "preallocation"
+#define BLOCK_OPT_SUBFMT            "subformat"
+#define BLOCK_OPT_COMPAT_LEVEL      "compat"
+#define BLOCK_OPT_LAZY_REFCOUNTS    "lazy_refcounts"
+#define BLOCK_OPT_ADAPTER_TYPE      "adapter_type"
+
+typedef struct BdrvTrackedRequest {
+    BlockDriverState *bs;
+    int64_t sector_num;
+    int nb_sectors;
+    bool is_write;
+    QLIST_ENTRY(BdrvTrackedRequest) list;
+    Coroutine *co; /* owner, used for deadlock detection */
+    CoQueue wait_queue; /* coroutines blocked on this request */
+} BdrvTrackedRequest;
+
+
+typedef struct BlockIOLimit {
+    int64_t bps[3];
+    int64_t iops[3];
+} BlockIOLimit;
+
+typedef struct BlockIOBaseValue {
+    uint64_t bytes[2];
+    uint64_t ios[2];
+} BlockIOBaseValue;
+
+struct BlockDriver {
+    const char *format_name;
+    int instance_size;
+    int (*bdrv_probe)(const uint8_t *buf, int buf_size, const char *filename);
+    int (*bdrv_probe_device)(const char *filename);
+
+    /* Any driver implementing this callback is expected to be able to handle
+     * NULL file names in its .bdrv_open() implementation */
+    void (*bdrv_parse_filename)(const char *filename, QDict *options, Error **errp);
+
+    /* For handling image reopen for split or non-split files */
+    int (*bdrv_reopen_prepare)(BDRVReopenState *reopen_state,
+                               BlockReopenQueue *queue, Error **errp);
+    void (*bdrv_reopen_commit)(BDRVReopenState *reopen_state);
+    void (*bdrv_reopen_abort)(BDRVReopenState *reopen_state);
+
+    int (*bdrv_open)(BlockDriverState *bs, QDict *options, int flags);
+    int (*bdrv_file_open)(BlockDriverState *bs, QDict *options, int flags);
+    int (*bdrv_read)(BlockDriverState *bs, int64_t sector_num,
+                     uint8_t *buf, int nb_sectors);
+    int (*bdrv_write)(BlockDriverState *bs, int64_t sector_num,
+                      const uint8_t *buf, int nb_sectors);
+    void (*bdrv_close)(BlockDriverState *bs);
+    void (*bdrv_rebind)(BlockDriverState *bs);
+    int (*bdrv_create)(const char *filename, QEMUOptionParameter *options);
+    int (*bdrv_set_key)(BlockDriverState *bs, const char *key);
+    int (*bdrv_make_empty)(BlockDriverState *bs);
+    /* aio */
+    BlockDriverAIOCB *(*bdrv_aio_readv)(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque);
+    BlockDriverAIOCB *(*bdrv_aio_writev)(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque);
+    BlockDriverAIOCB *(*bdrv_aio_flush)(BlockDriverState *bs,
+        BlockDriverCompletionFunc *cb, void *opaque);
+    BlockDriverAIOCB *(*bdrv_aio_discard)(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque);
+
+    int coroutine_fn (*bdrv_co_readv)(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors, QEMUIOVector *qiov);
+    int coroutine_fn (*bdrv_co_writev)(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors, QEMUIOVector *qiov);
+    /*
+     * Efficiently zero a region of the disk image.  Typically an image format
+     * would use a compact metadata representation to implement this.  This
+     * function pointer may be NULL and .bdrv_co_writev() will be called
+     * instead.
+     */
+    int coroutine_fn (*bdrv_co_write_zeroes)(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors);
+    int coroutine_fn (*bdrv_co_discard)(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors);
+    int coroutine_fn (*bdrv_co_is_allocated)(BlockDriverState *bs,
+        int64_t sector_num, int nb_sectors, int *pnum);
+
+    /*
+     * Invalidate any cached meta-data.
+     */
+    void (*bdrv_invalidate_cache)(BlockDriverState *bs);
+
+    /*
+     * Flushes all data that was already written to the OS all the way down to
+     * the disk (for example raw-posix calls fsync()).
+     */
+    int coroutine_fn (*bdrv_co_flush_to_disk)(BlockDriverState *bs);
+
+    /*
+     * Flushes all internal caches to the OS. The data may still sit in a
+     * writeback cache of the host OS, but it will survive a crash of the qemu
+     * process.
+     */
+    int coroutine_fn (*bdrv_co_flush_to_os)(BlockDriverState *bs);
+
+    const char *protocol_name;
+    int (*bdrv_truncate)(BlockDriverState *bs, int64_t offset);
+    int64_t (*bdrv_getlength)(BlockDriverState *bs);
+    int64_t (*bdrv_get_allocated_file_size)(BlockDriverState *bs);
+    int (*bdrv_write_compressed)(BlockDriverState *bs, int64_t sector_num,
+                                 const uint8_t *buf, int nb_sectors);
+
+    int (*bdrv_snapshot_create)(BlockDriverState *bs,
+                                QEMUSnapshotInfo *sn_info);
+    int (*bdrv_snapshot_goto)(BlockDriverState *bs,
+                              const char *snapshot_id);
+    int (*bdrv_snapshot_delete)(BlockDriverState *bs, const char *snapshot_id);
+    int (*bdrv_snapshot_list)(BlockDriverState *bs,
+                              QEMUSnapshotInfo **psn_info);
+    int (*bdrv_snapshot_load_tmp)(BlockDriverState *bs,
+                                  const char *snapshot_name);
+    int (*bdrv_get_info)(BlockDriverState *bs, BlockDriverInfo *bdi);
+
+    int (*bdrv_save_vmstate)(BlockDriverState *bs, QEMUIOVector *qiov,
+                             int64_t pos);
+    int (*bdrv_load_vmstate)(BlockDriverState *bs, uint8_t *buf,
+                             int64_t pos, int size);
+
+    int (*bdrv_change_backing_file)(BlockDriverState *bs,
+        const char *backing_file, const char *backing_fmt);
+
+    /* removable device specific */
+    int (*bdrv_is_inserted)(BlockDriverState *bs);
+    int (*bdrv_media_changed)(BlockDriverState *bs);
+    void (*bdrv_eject)(BlockDriverState *bs, bool eject_flag);
+    void (*bdrv_lock_medium)(BlockDriverState *bs, bool locked);
+
+    /* to control generic scsi devices */
+    int (*bdrv_ioctl)(BlockDriverState *bs, unsigned long int req, void *buf);
+    BlockDriverAIOCB *(*bdrv_aio_ioctl)(BlockDriverState *bs,
+        unsigned long int req, void *buf,
+        BlockDriverCompletionFunc *cb, void *opaque);
+
+    /* List of options for creating images, terminated by name == NULL */
+    QEMUOptionParameter *create_options;
+
+
+    /*
+     * Returns 0 for completed check, -errno for internal errors.
+     * The check results are stored in result.
+     */
+    int (*bdrv_check)(BlockDriverState* bs, BdrvCheckResult *result,
+        BdrvCheckMode fix);
+
+    void (*bdrv_debug_event)(BlockDriverState *bs, BlkDebugEvent event);
+
+    /* TODO Better pass a option string/QDict/QemuOpts to add any rule? */
+    int (*bdrv_debug_breakpoint)(BlockDriverState *bs, const char *event,
+        const char *tag);
+    int (*bdrv_debug_resume)(BlockDriverState *bs, const char *tag);
+    bool (*bdrv_debug_is_suspended)(BlockDriverState *bs, const char *tag);
+
+    /*
+     * Returns 1 if newly created images are guaranteed to contain only
+     * zeros, 0 otherwise.
+     */
+    int (*bdrv_has_zero_init)(BlockDriverState *bs);
+
+    QLIST_ENTRY(BlockDriver) list;
+};
+
+/*
+ * Note: the function bdrv_append() copies and swaps contents of
+ * BlockDriverStates, so if you add new fields to this struct, please
+ * inspect bdrv_append() to determine if the new fields need to be
+ * copied as well.
+ */
+struct BlockDriverState {
+    int64_t total_sectors; /* if we are reading a disk image, give its
+                              size in sectors */
+    int read_only; /* if true, the media is read only */
+    int open_flags; /* flags used to open the file, re-used for re-open */
+    int encrypted; /* if true, the media is encrypted */
+    int valid_key; /* if true, a valid encryption key has been set */
+    int sg;        /* if true, the device is a /dev/sg* */
+    int copy_on_read; /* if true, copy read backing sectors into image
+                         note this is a reference count */
+
+    BlockDriver *drv; /* NULL means no media */
+    void *opaque;
+
+    void *dev;                  /* attached device model, if any */
+    /* TODO change to DeviceState when all users are qdevified */
+    const BlockDevOps *dev_ops;
+    void *dev_opaque;
+
+    char filename[1024];
+    char backing_file[1024]; /* if non zero, the image is a diff of
+                                this file image */
+    char backing_format[16]; /* if non-zero and backing_file exists */
+    int is_temporary;
+
+    BlockDriverState *backing_hd;
+    BlockDriverState *file;
+
+    NotifierList close_notifiers;
+
+    /* Callback before write request is processed */
+    NotifierWithReturnList before_write_notifiers;
+
+    /* number of in-flight copy-on-read requests */
+    unsigned int copy_on_read_in_flight;
+
+    /* the time for latest disk I/O */
+    int64_t slice_start;
+    int64_t slice_end;
+    BlockIOLimit io_limits;
+    BlockIOBaseValue slice_submitted;
+    CoQueue      throttled_reqs;
+    QEMUTimer    *block_timer;
+    bool         io_limits_enabled;
+
+    /* I/O stats (display with "info blockstats"). */
+    uint64_t nr_bytes[BDRV_MAX_IOTYPE];
+    uint64_t nr_ops[BDRV_MAX_IOTYPE];
+    uint64_t total_time_ns[BDRV_MAX_IOTYPE];
+    uint64_t wr_highest_sector;
+
+    /* Whether the disk can expand beyond total_sectors */
+    int growable;
+
+    /* the memory alignment required for the buffers handled by this driver */
+    int buffer_alignment;
+
+    /* do we need to tell the quest if we have a volatile write cache? */
+    int enable_write_cache;
+
+    /* NOTE: the following infos are only hints for real hardware
+       drivers. They are not used by the block driver */
+    BlockdevOnError on_read_error, on_write_error;
+    bool iostatus_enabled;
+    BlockDeviceIoStatus iostatus;
+    char device_name[32];
+    HBitmap *dirty_bitmap;
+    int in_use; /* users other than guest access, eg. block migration */
+    QTAILQ_ENTRY(BlockDriverState) list;
+
+    QLIST_HEAD(, BdrvTrackedRequest) tracked_requests;
+
+    /* long-running background operation */
+    BlockJob *job;
+
+    QDict *options;
+};
+
+int get_tmp_filename(char *filename, int size);
+
+void bdrv_set_io_limits(BlockDriverState *bs,
+                        BlockIOLimit *io_limits);
+
+/**
+ * bdrv_add_before_write_notifier:
+ *
+ * Register a callback that is invoked before write requests are processed but
+ * after any throttling or waiting for overlapping requests.
+ */
+void bdrv_add_before_write_notifier(BlockDriverState *bs,
+                                    NotifierWithReturn *notifier);
+
+/**
+ * bdrv_get_aio_context:
+ *
+ * Returns: the currently bound #AioContext
+ */
+AioContext *bdrv_get_aio_context(BlockDriverState *bs);
+
+#ifdef _WIN32
+int is_windows_drive(const char *filename);
+#endif
+void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
+                               enum MonitorEvent ev,
+                               BlockErrorAction action, bool is_read);
+
+/**
+ * stream_start:
+ * @bs: Block device to operate on.
+ * @base: Block device that will become the new base, or %NULL to
+ * flatten the whole backing file chain onto @bs.
+ * @base_id: The file name that will be written to @bs as the new
+ * backing file if the job completes.  Ignored if @base is %NULL.
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @on_error: The action to take upon error.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ * @errp: Error object.
+ *
+ * Start a streaming operation on @bs.  Clusters that are unallocated
+ * in @bs, but allocated in any image between @base and @bs (both
+ * exclusive) will be written to @bs.  At the end of a successful
+ * streaming job, the backing file of @bs will be changed to
+ * @base_id in the written image and to @base in the live BlockDriverState.
+ */
+void stream_start(BlockDriverState *bs, BlockDriverState *base,
+                  const char *base_id, int64_t speed, BlockdevOnError on_error,
+                  BlockDriverCompletionFunc *cb,
+                  void *opaque, Error **errp);
+
+/**
+ * commit_start:
+ * @bs: Top Block device
+ * @base: Block device that will be written into, and become the new top
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @on_error: The action to take upon error.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ * @errp: Error object.
+ *
+ */
+void commit_start(BlockDriverState *bs, BlockDriverState *base,
+                 BlockDriverState *top, int64_t speed,
+                 BlockdevOnError on_error, BlockDriverCompletionFunc *cb,
+                 void *opaque, Error **errp);
+
+/*
+ * mirror_start:
+ * @bs: Block device to operate on.
+ * @target: Block device to write to.
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @granularity: The chosen granularity for the dirty bitmap.
+ * @buf_size: The amount of data that can be in flight at one time.
+ * @mode: Whether to collapse all images in the chain to the target.
+ * @on_source_error: The action to take upon error reading from the source.
+ * @on_target_error: The action to take upon error writing to the target.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ * @errp: Error object.
+ *
+ * Start a mirroring operation on @bs.  Clusters that are allocated
+ * in @bs will be written to @bs until the job is cancelled or
+ * manually completed.  At the end of a successful mirroring job,
+ * @bs will be switched to read from @target.
+ */
+void mirror_start(BlockDriverState *bs, BlockDriverState *target,
+                  int64_t speed, int64_t granularity, int64_t buf_size,
+                  MirrorSyncMode mode, BlockdevOnError on_source_error,
+                  BlockdevOnError on_target_error,
+                  BlockDriverCompletionFunc *cb,
+                  void *opaque, Error **errp);
+
+/*
+ * backup_start:
+ * @bs: Block device to operate on.
+ * @target: Block device to write to.
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @on_source_error: The action to take upon error reading from the source.
+ * @on_target_error: The action to take upon error writing to the target.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ *
+ * Start a backup operation on @bs.  Clusters in @bs are written to @target
+ * until the job is cancelled or manually completed.
+ */
+void backup_start(BlockDriverState *bs, BlockDriverState *target,
+                  int64_t speed, BlockdevOnError on_source_error,
+                  BlockdevOnError on_target_error,
+                  BlockDriverCompletionFunc *cb, void *opaque,
+                  Error **errp);
+
+#endif /* BLOCK_INT_H */
diff --git a/contrib/qemu/include/block/blockjob.h b/contrib/qemu/include/block/blockjob.h
new file mode 100644
index 000000000..c290d07bb
--- /dev/null
+++ b/contrib/qemu/include/block/blockjob.h
@@ -0,0 +1,278 @@
+/*
+ * Declarations for long-running block device operations
+ *
+ * Copyright (c) 2011 IBM Corp.
+ * Copyright (c) 2012 Red Hat, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#ifndef BLOCKJOB_H
+#define BLOCKJOB_H 1
+
+#include "block/block.h"
+
+/**
+ * BlockJobType:
+ *
+ * A class type for block job objects.
+ */
+typedef struct BlockJobType {
+    /** Derived BlockJob struct size */
+    size_t instance_size;
+
+    /** String describing the operation, part of query-block-jobs QMP API */
+    const char *job_type;
+
+    /** Optional callback for job types that support setting a speed limit */
+    void (*set_speed)(BlockJob *job, int64_t speed, Error **errp);
+
+    /** Optional callback for job types that need to forward I/O status reset */
+    void (*iostatus_reset)(BlockJob *job);
+
+    /**
+     * Optional callback for job types whose completion must be triggered
+     * manually.
+     */
+    void (*complete)(BlockJob *job, Error **errp);
+} BlockJobType;
+
+/**
+ * BlockJob:
+ *
+ * Long-running operation on a BlockDriverState.
+ */
+struct BlockJob {
+    /** The job type, including the job vtable.  */
+    const BlockJobType *job_type;
+
+    /** The block device on which the job is operating.  */
+    BlockDriverState *bs;
+
+    /**
+     * The coroutine that executes the job.  If not NULL, it is
+     * reentered when busy is false and the job is cancelled.
+     */
+    Coroutine *co;
+
+    /**
+     * Set to true if the job should cancel itself.  The flag must
+     * always be tested just before toggling the busy flag from false
+     * to true.  After a job has been cancelled, it should only yield
+     * if #qemu_aio_wait will ("sooner or later") reenter the coroutine.
+     */
+    bool cancelled;
+
+    /**
+     * Set to true if the job is either paused, or will pause itself
+     * as soon as possible (if busy == true).
+     */
+    bool paused;
+
+    /**
+     * Set to false by the job while it is in a quiescent state, where
+     * no I/O is pending and the job has yielded on any condition
+     * that is not detected by #qemu_aio_wait, such as a timer.
+     */
+    bool busy;
+
+    /** Status that is published by the query-block-jobs QMP API */
+    BlockDeviceIoStatus iostatus;
+
+    /** Offset that is published by the query-block-jobs QMP API */
+    int64_t offset;
+
+    /** Length that is published by the query-block-jobs QMP API */
+    int64_t len;
+
+    /** Speed that was set with @block_job_set_speed.  */
+    int64_t speed;
+
+    /** The completion function that will be called when the job completes.  */
+    BlockDriverCompletionFunc *cb;
+
+    /** The opaque value that is passed to the completion function.  */
+    void *opaque;
+};
+
+/**
+ * block_job_create:
+ * @job_type: The class object for the newly-created job.
+ * @bs: The block
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ * @errp: Error object.
+ *
+ * Create a new long-running block device job and return it.  The job
+ * will call @cb asynchronously when the job completes.  Note that
+ * @bs may have been closed at the time the @cb it is called.  If
+ * this is the case, the job may be reported as either cancelled or
+ * completed.
+ *
+ * This function is not part of the public job interface; it should be
+ * called from a wrapper that is specific to the job type.
+ */
+void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
+                       int64_t speed, BlockDriverCompletionFunc *cb,
+                       void *opaque, Error **errp);
+
+/**
+ * block_job_sleep_ns:
+ * @job: The job that calls the function.
+ * @clock: The clock to sleep on.
+ * @ns: How many nanoseconds to stop for.
+ *
+ * Put the job to sleep (assuming that it wasn't canceled) for @ns
+ * nanoseconds.  Canceling the job will interrupt the wait immediately.
+ */
+void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns);
+
+/**
+ * block_job_completed:
+ * @job: The job being completed.
+ * @ret: The status code.
+ *
+ * Call the completion function that was registered at creation time, and
+ * free @job.
+ */
+void block_job_completed(BlockJob *job, int ret);
+
+/**
+ * block_job_set_speed:
+ * @job: The job to set the speed for.
+ * @speed: The new value
+ * @errp: Error object.
+ *
+ * Set a rate-limiting parameter for the job; the actual meaning may
+ * vary depending on the job type.
+ */
+void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp);
+
+/**
+ * block_job_cancel:
+ * @job: The job to be canceled.
+ *
+ * Asynchronously cancel the specified job.
+ */
+void block_job_cancel(BlockJob *job);
+
+/**
+ * block_job_complete:
+ * @job: The job to be completed.
+ * @errp: Error object.
+ *
+ * Asynchronously complete the specified job.
+ */
+void block_job_complete(BlockJob *job, Error **errp);
+
+/**
+ * block_job_is_cancelled:
+ * @job: The job being queried.
+ *
+ * Returns whether the job is scheduled for cancellation.
+ */
+bool block_job_is_cancelled(BlockJob *job);
+
+/**
+ * block_job_query:
+ * @job: The job to get information about.
+ *
+ * Return information about a job.
+ */
+BlockJobInfo *block_job_query(BlockJob *job);
+
+/**
+ * block_job_pause:
+ * @job: The job to be paused.
+ *
+ * Asynchronously pause the specified job.
+ */
+void block_job_pause(BlockJob *job);
+
+/**
+ * block_job_resume:
+ * @job: The job to be resumed.
+ *
+ * Resume the specified job.
+ */
+void block_job_resume(BlockJob *job);
+
+/**
+ * qobject_from_block_job:
+ * @job: The job whose information is requested.
+ *
+ * Return a QDict corresponding to @job's query-block-jobs entry.
+ */
+QObject *qobject_from_block_job(BlockJob *job);
+
+/**
+ * block_job_ready:
+ * @job: The job which is now ready to complete.
+ *
+ * Send a BLOCK_JOB_READY event for the specified job.
+ */
+void block_job_ready(BlockJob *job);
+
+/**
+ * block_job_is_paused:
+ * @job: The job being queried.
+ *
+ * Returns whether the job is currently paused, or will pause
+ * as soon as it reaches a sleeping point.
+ */
+bool block_job_is_paused(BlockJob *job);
+
+/**
+ * block_job_cancel_sync:
+ * @job: The job to be canceled.
+ *
+ * Synchronously cancel the job.  The completion callback is called
+ * before the function returns.  The job may actually complete
+ * instead of canceling itself; the circumstances under which this
+ * happens depend on the kind of job that is active.
+ *
+ * Returns the return value from the job if the job actually completed
+ * during the call, or -ECANCELED if it was canceled.
+ */
+int block_job_cancel_sync(BlockJob *job);
+
+/**
+ * block_job_iostatus_reset:
+ * @job: The job whose I/O status should be reset.
+ *
+ * Reset I/O status on @job and on BlockDriverState objects it uses,
+ * other than job->bs.
+ */
+void block_job_iostatus_reset(BlockJob *job);
+
+/**
+ * block_job_error_action:
+ * @job: The job to signal an error for.
+ * @bs: The block device on which to set an I/O error.
+ * @on_err: The error action setting.
+ * @is_read: Whether the operation was a read.
+ * @error: The error that was reported.
+ *
+ * Report an I/O error for a block job and possibly stop the VM.  Return the
+ * action that was selected based on @on_err and @error.
+ */
+BlockErrorAction block_job_error_action(BlockJob *job, BlockDriverState *bs,
+                                        BlockdevOnError on_err,
+                                        int is_read, int error);
+#endif
diff --git a/contrib/qemu/include/block/coroutine.h b/contrib/qemu/include/block/coroutine.h
new file mode 100644
index 000000000..377805a3b
--- /dev/null
+++ b/contrib/qemu/include/block/coroutine.h
@@ -0,0 +1,218 @@
+/*
+ * QEMU coroutine implementation
+ *
+ * Copyright IBM, Corp. 2011
+ *
+ * Authors:
+ *  Stefan Hajnoczi    <stefanha@linux.vnet.ibm.com>
+ *  Kevin Wolf         <kwolf@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#ifndef QEMU_COROUTINE_H
+#define QEMU_COROUTINE_H
+
+#include <stdbool.h>
+#include "qemu/queue.h"
+#include "qemu/timer.h"
+
+/**
+ * Coroutines are a mechanism for stack switching and can be used for
+ * cooperative userspace threading.  These functions provide a simple but
+ * useful flavor of coroutines that is suitable for writing sequential code,
+ * rather than callbacks, for operations that need to give up control while
+ * waiting for events to complete.
+ *
+ * These functions are re-entrant and may be used outside the global mutex.
+ */
+
+/**
+ * Mark a function that executes in coroutine context
+ *
+ * Functions that execute in coroutine context cannot be called directly from
+ * normal functions.  In the future it would be nice to enable compiler or
+ * static checker support for catching such errors.  This annotation might make
+ * it possible and in the meantime it serves as documentation.
+ *
+ * For example:
+ *
+ *   static void coroutine_fn foo(void) {
+ *       ....
+ *   }
+ */
+#define coroutine_fn
+
+typedef struct Coroutine Coroutine;
+
+/**
+ * Coroutine entry point
+ *
+ * When the coroutine is entered for the first time, opaque is passed in as an
+ * argument.
+ *
+ * When this function returns, the coroutine is destroyed automatically and
+ * execution continues in the caller who last entered the coroutine.
+ */
+typedef void coroutine_fn CoroutineEntry(void *opaque);
+
+/**
+ * Create a new coroutine
+ *
+ * Use qemu_coroutine_enter() to actually transfer control to the coroutine.
+ */
+Coroutine *qemu_coroutine_create(CoroutineEntry *entry);
+
+/**
+ * Transfer control to a coroutine
+ *
+ * The opaque argument is passed as the argument to the entry point when
+ * entering the coroutine for the first time.  It is subsequently ignored.
+ */
+void qemu_coroutine_enter(Coroutine *coroutine, void *opaque);
+
+/**
+ * Transfer control back to a coroutine's caller
+ *
+ * This function does not return until the coroutine is re-entered using
+ * qemu_coroutine_enter().
+ */
+void coroutine_fn qemu_coroutine_yield(void);
+
+/**
+ * Get the currently executing coroutine
+ */
+Coroutine *coroutine_fn qemu_coroutine_self(void);
+
+/**
+ * Return whether or not currently inside a coroutine
+ *
+ * This can be used to write functions that work both when in coroutine context
+ * and when not in coroutine context.  Note that such functions cannot use the
+ * coroutine_fn annotation since they work outside coroutine context.
+ */
+bool qemu_in_coroutine(void);
+
+
+
+/**
+ * CoQueues are a mechanism to queue coroutines in order to continue executing
+ * them later. They provide the fundamental primitives on which coroutine locks
+ * are built.
+ */
+typedef struct CoQueue {
+    QTAILQ_HEAD(, Coroutine) entries;
+    AioContext *ctx;
+} CoQueue;
+
+/**
+ * Initialise a CoQueue. This must be called before any other operation is used
+ * on the CoQueue.
+ */
+void qemu_co_queue_init(CoQueue *queue);
+
+/**
+ * Adds the current coroutine to the CoQueue and transfers control to the
+ * caller of the coroutine.
+ */
+void coroutine_fn qemu_co_queue_wait(CoQueue *queue);
+
+/**
+ * Adds the current coroutine to the head of the CoQueue and transfers control to the
+ * caller of the coroutine.
+ */
+void coroutine_fn qemu_co_queue_wait_insert_head(CoQueue *queue);
+
+/**
+ * Restarts the next coroutine in the CoQueue and removes it from the queue.
+ *
+ * Returns true if a coroutine was restarted, false if the queue is empty.
+ */
+bool qemu_co_queue_next(CoQueue *queue);
+
+/**
+ * Restarts all coroutines in the CoQueue and leaves the queue empty.
+ */
+void qemu_co_queue_restart_all(CoQueue *queue);
+
+/**
+ * Checks if the CoQueue is empty.
+ */
+bool qemu_co_queue_empty(CoQueue *queue);
+
+
+/**
+ * Provides a mutex that can be used to synchronise coroutines
+ */
+typedef struct CoMutex {
+    bool locked;
+    CoQueue queue;
+} CoMutex;
+
+/**
+ * Initialises a CoMutex. This must be called before any other operation is used
+ * on the CoMutex.
+ */
+void qemu_co_mutex_init(CoMutex *mutex);
+
+/**
+ * Locks the mutex. If the lock cannot be taken immediately, control is
+ * transferred to the caller of the current coroutine.
+ */
+void coroutine_fn qemu_co_mutex_lock(CoMutex *mutex);
+
+/**
+ * Unlocks the mutex and schedules the next coroutine that was waiting for this
+ * lock to be run.
+ */
+void coroutine_fn qemu_co_mutex_unlock(CoMutex *mutex);
+
+typedef struct CoRwlock {
+    bool writer;
+    int reader;
+    CoQueue queue;
+} CoRwlock;
+
+/**
+ * Initialises a CoRwlock. This must be called before any other operation
+ * is used on the CoRwlock
+ */
+void qemu_co_rwlock_init(CoRwlock *lock);
+
+/**
+ * Read locks the CoRwlock. If the lock cannot be taken immediately because
+ * of a parallel writer, control is transferred to the caller of the current
+ * coroutine.
+ */
+void qemu_co_rwlock_rdlock(CoRwlock *lock);
+
+/**
+ * Write Locks the mutex. If the lock cannot be taken immediately because
+ * of a parallel reader, control is transferred to the caller of the current
+ * coroutine.
+ */
+void qemu_co_rwlock_wrlock(CoRwlock *lock);
+
+/**
+ * Unlocks the read/write lock and schedules the next coroutine that was
+ * waiting for this lock to be run.
+ */
+void qemu_co_rwlock_unlock(CoRwlock *lock);
+
+/**
+ * Yield the coroutine for a given duration
+ *
+ * Note this function uses timers and hence only works when a main loop is in
+ * use.  See main-loop.h and do not use from qemu-tool programs.
+ */
+void coroutine_fn co_sleep_ns(QEMUClock *clock, int64_t ns);
+
+/**
+ * Yield until a file descriptor becomes readable
+ *
+ * Note that this function clobbers the handlers for the file descriptor.
+ */
+void coroutine_fn yield_until_fd_readable(int fd);
+#endif /* QEMU_COROUTINE_H */
diff --git a/contrib/qemu/include/block/coroutine_int.h b/contrib/qemu/include/block/coroutine_int.h
new file mode 100644
index 000000000..f133d65af
--- /dev/null
+++ b/contrib/qemu/include/block/coroutine_int.h
@@ -0,0 +1,53 @@
+/*
+ * Coroutine internals
+ *
+ * Copyright (c) 2011 Kevin Wolf <kwolf@redhat.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef QEMU_COROUTINE_INT_H
+#define QEMU_COROUTINE_INT_H
+
+#include "qemu/queue.h"
+#include "block/coroutine.h"
+
+typedef enum {
+    COROUTINE_YIELD = 1,
+    COROUTINE_TERMINATE = 2,
+} CoroutineAction;
+
+struct Coroutine {
+    CoroutineEntry *entry;
+    void *entry_arg;
+    Coroutine *caller;
+    QSLIST_ENTRY(Coroutine) pool_next;
+
+    /* Coroutines that should be woken up when we yield or terminate */
+    QTAILQ_HEAD(, Coroutine) co_queue_wakeup;
+    QTAILQ_ENTRY(Coroutine) co_queue_next;
+};
+
+Coroutine *qemu_coroutine_new(void);
+void qemu_coroutine_delete(Coroutine *co);
+CoroutineAction qemu_coroutine_switch(Coroutine *from, Coroutine *to,
+                                      CoroutineAction action);
+void coroutine_fn qemu_co_queue_run_restart(Coroutine *co);
+
+#endif
diff --git a/contrib/qemu/include/block/snapshot.h b/contrib/qemu/include/block/snapshot.h
new file mode 100644
index 000000000..eaf61f032
--- /dev/null
+++ b/contrib/qemu/include/block/snapshot.h
@@ -0,0 +1,53 @@
+/*
+ * Block layer snapshot related functions
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#ifndef SNAPSHOT_H
+#define SNAPSHOT_H
+
+#include "qemu-common.h"
+
+typedef struct QEMUSnapshotInfo {
+    char id_str[128]; /* unique snapshot id */
+    /* the following fields are informative. They are not needed for
+       the consistency of the snapshot */
+    char name[256]; /* user chosen name */
+    uint64_t vm_state_size; /* VM state info size */
+    uint32_t date_sec; /* UTC date of the snapshot */
+    uint32_t date_nsec;
+    uint64_t vm_clock_nsec; /* VM clock relative to boot */
+} QEMUSnapshotInfo;
+
+int bdrv_snapshot_find(BlockDriverState *bs, QEMUSnapshotInfo *sn_info,
+                       const char *name);
+int bdrv_can_snapshot(BlockDriverState *bs);
+int bdrv_snapshot_create(BlockDriverState *bs,
+                         QEMUSnapshotInfo *sn_info);
+int bdrv_snapshot_goto(BlockDriverState *bs,
+                       const char *snapshot_id);
+int bdrv_snapshot_delete(BlockDriverState *bs, const char *snapshot_id);
+int bdrv_snapshot_list(BlockDriverState *bs,
+                       QEMUSnapshotInfo **psn_info);
+int bdrv_snapshot_load_tmp(BlockDriverState *bs,
+                           const char *snapshot_name);
+#endif
diff --git a/contrib/qemu/include/config.h b/contrib/qemu/include/config.h
new file mode 100644
index 000000000..e20f78696
--- /dev/null
+++ b/contrib/qemu/include/config.h
@@ -0,0 +1,2 @@
+#include "config-host.h"
+#include "config-target.h"
diff --git a/contrib/qemu/include/exec/cpu-common.h b/contrib/qemu/include/exec/cpu-common.h
new file mode 100644
index 000000000..e4996e19c
--- /dev/null
+++ b/contrib/qemu/include/exec/cpu-common.h
@@ -0,0 +1,124 @@
+#ifndef CPU_COMMON_H
+#define CPU_COMMON_H 1
+
+/* CPU interfaces that are target independent.  */
+
+#ifndef CONFIG_USER_ONLY
+#include "exec/hwaddr.h"
+#endif
+
+#ifndef NEED_CPU_H
+#include "exec/poison.h"
+#endif
+
+#include "qemu/bswap.h"
+#include "qemu/queue.h"
+
+/**
+ * CPUListState:
+ * @cpu_fprintf: Print function.
+ * @file: File to print to using @cpu_fprint.
+ *
+ * State commonly used for iterating over CPU models.
+ */
+typedef struct CPUListState {
+    fprintf_function cpu_fprintf;
+    FILE *file;
+} CPUListState;
+
+#if !defined(CONFIG_USER_ONLY)
+
+enum device_endian {
+    DEVICE_NATIVE_ENDIAN,
+    DEVICE_BIG_ENDIAN,
+    DEVICE_LITTLE_ENDIAN,
+};
+
+/* address in the RAM (different from a physical address) */
+#if defined(CONFIG_XEN_BACKEND)
+typedef uint64_t ram_addr_t;
+#  define RAM_ADDR_MAX UINT64_MAX
+#  define RAM_ADDR_FMT "%" PRIx64
+#else
+typedef uintptr_t ram_addr_t;
+#  define RAM_ADDR_MAX UINTPTR_MAX
+#  define RAM_ADDR_FMT "%" PRIxPTR
+#endif
+
+/* memory API */
+
+typedef void CPUWriteMemoryFunc(void *opaque, hwaddr addr, uint32_t value);
+typedef uint32_t CPUReadMemoryFunc(void *opaque, hwaddr addr);
+
+void qemu_ram_remap(ram_addr_t addr, ram_addr_t length);
+/* This should not be used by devices.  */
+MemoryRegion *qemu_ram_addr_from_host(void *ptr, ram_addr_t *ram_addr);
+void qemu_ram_set_idstr(ram_addr_t addr, const char *name, DeviceState *dev);
+
+void cpu_physical_memory_rw(hwaddr addr, uint8_t *buf,
+                            int len, int is_write);
+static inline void cpu_physical_memory_read(hwaddr addr,
+                                            void *buf, int len)
+{
+    cpu_physical_memory_rw(addr, buf, len, 0);
+}
+static inline void cpu_physical_memory_write(hwaddr addr,
+                                             const void *buf, int len)
+{
+    cpu_physical_memory_rw(addr, (void *)buf, len, 1);
+}
+void *cpu_physical_memory_map(hwaddr addr,
+                              hwaddr *plen,
+                              int is_write);
+void cpu_physical_memory_unmap(void *buffer, hwaddr len,
+                               int is_write, hwaddr access_len);
+void *cpu_register_map_client(void *opaque, void (*callback)(void *opaque));
+
+bool cpu_physical_memory_is_io(hwaddr phys_addr);
+
+/* Coalesced MMIO regions are areas where write operations can be reordered.
+ * This usually implies that write operations are side-effect free.  This allows
+ * batching which can make a major impact on performance when using
+ * virtualization.
+ */
+void qemu_flush_coalesced_mmio_buffer(void);
+
+uint32_t ldub_phys(hwaddr addr);
+uint32_t lduw_le_phys(hwaddr addr);
+uint32_t lduw_be_phys(hwaddr addr);
+uint32_t ldl_le_phys(hwaddr addr);
+uint32_t ldl_be_phys(hwaddr addr);
+uint64_t ldq_le_phys(hwaddr addr);
+uint64_t ldq_be_phys(hwaddr addr);
+void stb_phys(hwaddr addr, uint32_t val);
+void stw_le_phys(hwaddr addr, uint32_t val);
+void stw_be_phys(hwaddr addr, uint32_t val);
+void stl_le_phys(hwaddr addr, uint32_t val);
+void stl_be_phys(hwaddr addr, uint32_t val);
+void stq_le_phys(hwaddr addr, uint64_t val);
+void stq_be_phys(hwaddr addr, uint64_t val);
+
+#ifdef NEED_CPU_H
+uint32_t lduw_phys(hwaddr addr);
+uint32_t ldl_phys(hwaddr addr);
+uint64_t ldq_phys(hwaddr addr);
+void stl_phys_notdirty(hwaddr addr, uint32_t val);
+void stw_phys(hwaddr addr, uint32_t val);
+void stl_phys(hwaddr addr, uint32_t val);
+void stq_phys(hwaddr addr, uint64_t val);
+#endif
+
+void cpu_physical_memory_write_rom(hwaddr addr,
+                                   const uint8_t *buf, int len);
+
+extern struct MemoryRegion io_mem_rom;
+extern struct MemoryRegion io_mem_notdirty;
+
+typedef void (RAMBlockIterFunc)(void *host_addr,
+    ram_addr_t offset, ram_addr_t length, void *opaque);
+
+void qemu_ram_foreach_block(RAMBlockIterFunc func, void *opaque);
+
+#endif
+
+#endif /* !CPU_COMMON_H */
diff --git a/contrib/qemu/include/exec/hwaddr.h b/contrib/qemu/include/exec/hwaddr.h
new file mode 100644
index 000000000..c9eb78fba
--- /dev/null
+++ b/contrib/qemu/include/exec/hwaddr.h
@@ -0,0 +1,20 @@
+/* Define hwaddr if it exists.  */
+
+#ifndef HWADDR_H
+#define HWADDR_H
+
+#define HWADDR_BITS 64
+/* hwaddr is the type of a physical address (its size can
+   be different from 'target_ulong').  */
+
+typedef uint64_t hwaddr;
+#define HWADDR_MAX UINT64_MAX
+#define TARGET_FMT_plx "%016" PRIx64
+#define HWADDR_PRId PRId64
+#define HWADDR_PRIi PRIi64
+#define HWADDR_PRIo PRIo64
+#define HWADDR_PRIu PRIu64
+#define HWADDR_PRIx PRIx64
+#define HWADDR_PRIX PRIX64
+
+#endif
diff --git a/contrib/qemu/include/exec/poison.h b/contrib/qemu/include/exec/poison.h
new file mode 100644
index 000000000..2341a7504
--- /dev/null
+++ b/contrib/qemu/include/exec/poison.h
@@ -0,0 +1,63 @@
+/* Poison identifiers that should not be used when building
+   target independent device code.  */
+
+#ifndef HW_POISON_H
+#define HW_POISON_H
+#ifdef __GNUC__
+
+#pragma GCC poison TARGET_I386
+#pragma GCC poison TARGET_X86_64
+#pragma GCC poison TARGET_ALPHA
+#pragma GCC poison TARGET_ARM
+#pragma GCC poison TARGET_CRIS
+#pragma GCC poison TARGET_LM32
+#pragma GCC poison TARGET_M68K
+#pragma GCC poison TARGET_MIPS
+#pragma GCC poison TARGET_MIPS64
+#pragma GCC poison TARGET_OPENRISC
+#pragma GCC poison TARGET_PPC
+#pragma GCC poison TARGET_PPCEMB
+#pragma GCC poison TARGET_PPC64
+#pragma GCC poison TARGET_ABI32
+#pragma GCC poison TARGET_SH4
+#pragma GCC poison TARGET_SPARC
+#pragma GCC poison TARGET_SPARC64
+
+#pragma GCC poison TARGET_WORDS_BIGENDIAN
+#pragma GCC poison BSWAP_NEEDED
+
+#pragma GCC poison TARGET_LONG_BITS
+#pragma GCC poison TARGET_FMT_lx
+#pragma GCC poison TARGET_FMT_ld
+
+#pragma GCC poison TARGET_PAGE_SIZE
+#pragma GCC poison TARGET_PAGE_MASK
+#pragma GCC poison TARGET_PAGE_BITS
+#pragma GCC poison TARGET_PAGE_ALIGN
+
+#pragma GCC poison CPUArchState
+#pragma GCC poison env
+
+#pragma GCC poison lduw_phys
+#pragma GCC poison ldl_phys
+#pragma GCC poison ldq_phys
+#pragma GCC poison stl_phys_notdirty
+#pragma GCC poison stw_phys
+#pragma GCC poison stl_phys
+#pragma GCC poison stq_phys
+
+#pragma GCC poison CPU_INTERRUPT_HARD
+#pragma GCC poison CPU_INTERRUPT_EXITTB
+#pragma GCC poison CPU_INTERRUPT_HALT
+#pragma GCC poison CPU_INTERRUPT_DEBUG
+#pragma GCC poison CPU_INTERRUPT_TGT_EXT_0
+#pragma GCC poison CPU_INTERRUPT_TGT_EXT_1
+#pragma GCC poison CPU_INTERRUPT_TGT_EXT_2
+#pragma GCC poison CPU_INTERRUPT_TGT_EXT_3
+#pragma GCC poison CPU_INTERRUPT_TGT_EXT_4
+#pragma GCC poison CPU_INTERRUPT_TGT_INT_0
+#pragma GCC poison CPU_INTERRUPT_TGT_INT_1
+#pragma GCC poison CPU_INTERRUPT_TGT_INT_2
+
+#endif
+#endif
diff --git a/contrib/qemu/include/fpu/softfloat.h b/contrib/qemu/include/fpu/softfloat.h
new file mode 100644
index 000000000..f3927e241
--- /dev/null
+++ b/contrib/qemu/include/fpu/softfloat.h
@@ -0,0 +1,641 @@
+/*
+ * QEMU float support
+ *
+ * Derived from SoftFloat.
+ */
+
+/*============================================================================
+
+This C header file is part of the SoftFloat IEC/IEEE Floating-point Arithmetic
+Package, Release 2b.
+
+Written by John R. Hauser.  This work was made possible in part by the
+International Computer Science Institute, located at Suite 600, 1947 Center
+Street, Berkeley, California 94704.  Funding was partially provided by the
+National Science Foundation under grant MIP-9311980.  The original version
+of this code was written as part of a project to build a fixed-point vector
+processor in collaboration with the University of California at Berkeley,
+overseen by Profs. Nelson Morgan and John Wawrzynek.  More information
+is available through the Web page `http://www.cs.berkeley.edu/~jhauser/
+arithmetic/SoftFloat.html'.
+
+THIS SOFTWARE IS DISTRIBUTED AS IS, FOR FREE.  Although reasonable effort has
+been made to avoid it, THIS SOFTWARE MAY CONTAIN FAULTS THAT WILL AT TIMES
+RESULT IN INCORRECT BEHAVIOR.  USE OF THIS SOFTWARE IS RESTRICTED TO PERSONS
+AND ORGANIZATIONS WHO CAN AND WILL TAKE FULL RESPONSIBILITY FOR ALL LOSSES,
+COSTS, OR OTHER PROBLEMS THEY INCUR DUE TO THE SOFTWARE, AND WHO FURTHERMORE
+EFFECTIVELY INDEMNIFY JOHN HAUSER AND THE INTERNATIONAL COMPUTER SCIENCE
+INSTITUTE (possibly via similar legal warning) AGAINST ALL LOSSES, COSTS, OR
+OTHER PROBLEMS INCURRED BY THEIR CUSTOMERS AND CLIENTS DUE TO THE SOFTWARE.
+
+Derivative works are acceptable, even for commercial purposes, so long as
+(1) the source code for the derivative work includes prominent notice that
+the work is derivative, and (2) the source code includes prominent notice with
+these four paragraphs for those parts of this code that are retained.
+
+=============================================================================*/
+
+#ifndef SOFTFLOAT_H
+#define SOFTFLOAT_H
+
+#if defined(CONFIG_SOLARIS) && defined(CONFIG_NEEDS_LIBSUNMATH)
+#include <sunmath.h>
+#endif
+
+#include <inttypes.h>
+#include "config-host.h"
+#include "qemu/osdep.h"
+
+/*----------------------------------------------------------------------------
+| Each of the following `typedef's defines the most convenient type that holds
+| integers of at least as many bits as specified.  For example, `uint8' should
+| be the most convenient type that can hold unsigned integers of as many as
+| 8 bits.  The `flag' type must be able to hold either a 0 or 1.  For most
+| implementations of C, `flag', `uint8', and `int8' should all be `typedef'ed
+| to the same as `int'.
+*----------------------------------------------------------------------------*/
+typedef uint8_t flag;
+typedef uint8_t uint8;
+typedef int8_t int8;
+typedef unsigned int uint32;
+typedef signed int int32;
+typedef uint64_t uint64;
+typedef int64_t int64;
+
+#define LIT64( a ) a##LL
+#define INLINE static inline
+
+#define STATUS_PARAM , float_status *status
+#define STATUS(field) status->field
+#define STATUS_VAR , status
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE floating-point ordering relations
+*----------------------------------------------------------------------------*/
+enum {
+    float_relation_less      = -1,
+    float_relation_equal     =  0,
+    float_relation_greater   =  1,
+    float_relation_unordered =  2
+};
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE floating-point types.
+*----------------------------------------------------------------------------*/
+/* Use structures for soft-float types.  This prevents accidentally mixing
+   them with native int/float types.  A sufficiently clever compiler and
+   sane ABI should be able to see though these structs.  However
+   x86/gcc 3.x seems to struggle a bit, so leave them disabled by default.  */
+//#define USE_SOFTFLOAT_STRUCT_TYPES
+#ifdef USE_SOFTFLOAT_STRUCT_TYPES
+typedef struct {
+    uint16_t v;
+} float16;
+#define float16_val(x) (((float16)(x)).v)
+#define make_float16(x) __extension__ ({ float16 f16_val = {x}; f16_val; })
+#define const_float16(x) { x }
+typedef struct {
+    uint32_t v;
+} float32;
+/* The cast ensures an error if the wrong type is passed.  */
+#define float32_val(x) (((float32)(x)).v)
+#define make_float32(x) __extension__ ({ float32 f32_val = {x}; f32_val; })
+#define const_float32(x) { x }
+typedef struct {
+    uint64_t v;
+} float64;
+#define float64_val(x) (((float64)(x)).v)
+#define make_float64(x) __extension__ ({ float64 f64_val = {x}; f64_val; })
+#define const_float64(x) { x }
+#else
+typedef uint16_t float16;
+typedef uint32_t float32;
+typedef uint64_t float64;
+#define float16_val(x) (x)
+#define float32_val(x) (x)
+#define float64_val(x) (x)
+#define make_float16(x) (x)
+#define make_float32(x) (x)
+#define make_float64(x) (x)
+#define const_float16(x) (x)
+#define const_float32(x) (x)
+#define const_float64(x) (x)
+#endif
+typedef struct {
+    uint64_t low;
+    uint16_t high;
+} floatx80;
+#define make_floatx80(exp, mant) ((floatx80) { mant, exp })
+#define make_floatx80_init(exp, mant) { .low = mant, .high = exp }
+typedef struct {
+#ifdef HOST_WORDS_BIGENDIAN
+    uint64_t high, low;
+#else
+    uint64_t low, high;
+#endif
+} float128;
+#define make_float128(high_, low_) ((float128) { .high = high_, .low = low_ })
+#define make_float128_init(high_, low_) { .high = high_, .low = low_ }
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE floating-point underflow tininess-detection mode.
+*----------------------------------------------------------------------------*/
+enum {
+    float_tininess_after_rounding  = 0,
+    float_tininess_before_rounding = 1
+};
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE floating-point rounding mode.
+*----------------------------------------------------------------------------*/
+enum {
+    float_round_nearest_even = 0,
+    float_round_down         = 1,
+    float_round_up           = 2,
+    float_round_to_zero      = 3
+};
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE floating-point exception flags.
+*----------------------------------------------------------------------------*/
+enum {
+    float_flag_invalid   =  1,
+    float_flag_divbyzero =  4,
+    float_flag_overflow  =  8,
+    float_flag_underflow = 16,
+    float_flag_inexact   = 32,
+    float_flag_input_denormal = 64,
+    float_flag_output_denormal = 128
+};
+
+typedef struct float_status {
+    signed char float_detect_tininess;
+    signed char float_rounding_mode;
+    signed char float_exception_flags;
+    signed char floatx80_rounding_precision;
+    /* should denormalised results go to zero and set the inexact flag? */
+    flag flush_to_zero;
+    /* should denormalised inputs go to zero and set the input_denormal flag? */
+    flag flush_inputs_to_zero;
+    flag default_nan_mode;
+} float_status;
+
+void set_float_rounding_mode(int val STATUS_PARAM);
+void set_float_exception_flags(int val STATUS_PARAM);
+INLINE void set_float_detect_tininess(int val STATUS_PARAM)
+{
+    STATUS(float_detect_tininess) = val;
+}
+INLINE void set_flush_to_zero(flag val STATUS_PARAM)
+{
+    STATUS(flush_to_zero) = val;
+}
+INLINE void set_flush_inputs_to_zero(flag val STATUS_PARAM)
+{
+    STATUS(flush_inputs_to_zero) = val;
+}
+INLINE void set_default_nan_mode(flag val STATUS_PARAM)
+{
+    STATUS(default_nan_mode) = val;
+}
+INLINE int get_float_exception_flags(float_status *status)
+{
+    return STATUS(float_exception_flags);
+}
+void set_floatx80_rounding_precision(int val STATUS_PARAM);
+
+/*----------------------------------------------------------------------------
+| Routine to raise any or all of the software IEC/IEEE floating-point
+| exception flags.
+*----------------------------------------------------------------------------*/
+void float_raise( int8 flags STATUS_PARAM);
+
+/*----------------------------------------------------------------------------
+| Options to indicate which negations to perform in float*_muladd()
+| Using these differs from negating an input or output before calling
+| the muladd function in that this means that a NaN doesn't have its
+| sign bit inverted before it is propagated.
+*----------------------------------------------------------------------------*/
+enum {
+    float_muladd_negate_c = 1,
+    float_muladd_negate_product = 2,
+    float_muladd_negate_result = 4,
+};
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE integer-to-floating-point conversion routines.
+*----------------------------------------------------------------------------*/
+float32 int32_to_float32( int32 STATUS_PARAM );
+float64 int32_to_float64( int32 STATUS_PARAM );
+float32 uint32_to_float32( uint32 STATUS_PARAM );
+float64 uint32_to_float64( uint32 STATUS_PARAM );
+floatx80 int32_to_floatx80( int32 STATUS_PARAM );
+float128 int32_to_float128( int32 STATUS_PARAM );
+float32 int64_to_float32( int64 STATUS_PARAM );
+float32 uint64_to_float32( uint64 STATUS_PARAM );
+float64 int64_to_float64( int64 STATUS_PARAM );
+float64 uint64_to_float64( uint64 STATUS_PARAM );
+floatx80 int64_to_floatx80( int64 STATUS_PARAM );
+float128 int64_to_float128( int64 STATUS_PARAM );
+float128 uint64_to_float128( uint64 STATUS_PARAM );
+
+/*----------------------------------------------------------------------------
+| Software half-precision conversion routines.
+*----------------------------------------------------------------------------*/
+float16 float32_to_float16( float32, flag STATUS_PARAM );
+float32 float16_to_float32( float16, flag STATUS_PARAM );
+
+/*----------------------------------------------------------------------------
+| Software half-precision operations.
+*----------------------------------------------------------------------------*/
+int float16_is_quiet_nan( float16 );
+int float16_is_signaling_nan( float16 );
+float16 float16_maybe_silence_nan( float16 );
+
+INLINE int float16_is_any_nan(float16 a)
+{
+    return ((float16_val(a) & ~0x8000) > 0x7c00);
+}
+
+/*----------------------------------------------------------------------------
+| The pattern for a default generated half-precision NaN.
+*----------------------------------------------------------------------------*/
+extern const float16 float16_default_nan;
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE single-precision conversion routines.
+*----------------------------------------------------------------------------*/
+int_fast16_t float32_to_int16_round_to_zero(float32 STATUS_PARAM);
+uint_fast16_t float32_to_uint16_round_to_zero(float32 STATUS_PARAM);
+int32 float32_to_int32( float32 STATUS_PARAM );
+int32 float32_to_int32_round_to_zero( float32 STATUS_PARAM );
+uint32 float32_to_uint32( float32 STATUS_PARAM );
+uint32 float32_to_uint32_round_to_zero( float32 STATUS_PARAM );
+int64 float32_to_int64( float32 STATUS_PARAM );
+int64 float32_to_int64_round_to_zero( float32 STATUS_PARAM );
+float64 float32_to_float64( float32 STATUS_PARAM );
+floatx80 float32_to_floatx80( float32 STATUS_PARAM );
+float128 float32_to_float128( float32 STATUS_PARAM );
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE single-precision operations.
+*----------------------------------------------------------------------------*/
+float32 float32_round_to_int( float32 STATUS_PARAM );
+float32 float32_add( float32, float32 STATUS_PARAM );
+float32 float32_sub( float32, float32 STATUS_PARAM );
+float32 float32_mul( float32, float32 STATUS_PARAM );
+float32 float32_div( float32, float32 STATUS_PARAM );
+float32 float32_rem( float32, float32 STATUS_PARAM );
+float32 float32_muladd(float32, float32, float32, int STATUS_PARAM);
+float32 float32_sqrt( float32 STATUS_PARAM );
+float32 float32_exp2( float32 STATUS_PARAM );
+float32 float32_log2( float32 STATUS_PARAM );
+int float32_eq( float32, float32 STATUS_PARAM );
+int float32_le( float32, float32 STATUS_PARAM );
+int float32_lt( float32, float32 STATUS_PARAM );
+int float32_unordered( float32, float32 STATUS_PARAM );
+int float32_eq_quiet( float32, float32 STATUS_PARAM );
+int float32_le_quiet( float32, float32 STATUS_PARAM );
+int float32_lt_quiet( float32, float32 STATUS_PARAM );
+int float32_unordered_quiet( float32, float32 STATUS_PARAM );
+int float32_compare( float32, float32 STATUS_PARAM );
+int float32_compare_quiet( float32, float32 STATUS_PARAM );
+float32 float32_min(float32, float32 STATUS_PARAM);
+float32 float32_max(float32, float32 STATUS_PARAM);
+int float32_is_quiet_nan( float32 );
+int float32_is_signaling_nan( float32 );
+float32 float32_maybe_silence_nan( float32 );
+float32 float32_scalbn( float32, int STATUS_PARAM );
+
+INLINE float32 float32_abs(float32 a)
+{
+    /* Note that abs does *not* handle NaN specially, nor does
+     * it flush denormal inputs to zero.
+     */
+    return make_float32(float32_val(a) & 0x7fffffff);
+}
+
+INLINE float32 float32_chs(float32 a)
+{
+    /* Note that chs does *not* handle NaN specially, nor does
+     * it flush denormal inputs to zero.
+     */
+    return make_float32(float32_val(a) ^ 0x80000000);
+}
+
+INLINE int float32_is_infinity(float32 a)
+{
+    return (float32_val(a) & 0x7fffffff) == 0x7f800000;
+}
+
+INLINE int float32_is_neg(float32 a)
+{
+    return float32_val(a) >> 31;
+}
+
+INLINE int float32_is_zero(float32 a)
+{
+    return (float32_val(a) & 0x7fffffff) == 0;
+}
+
+INLINE int float32_is_any_nan(float32 a)
+{
+    return ((float32_val(a) & ~(1 << 31)) > 0x7f800000UL);
+}
+
+INLINE int float32_is_zero_or_denormal(float32 a)
+{
+    return (float32_val(a) & 0x7f800000) == 0;
+}
+
+INLINE float32 float32_set_sign(float32 a, int sign)
+{
+    return make_float32((float32_val(a) & 0x7fffffff) | (sign << 31));
+}
+
+#define float32_zero make_float32(0)
+#define float32_one make_float32(0x3f800000)
+#define float32_ln2 make_float32(0x3f317218)
+#define float32_pi make_float32(0x40490fdb)
+#define float32_half make_float32(0x3f000000)
+#define float32_infinity make_float32(0x7f800000)
+
+
+/*----------------------------------------------------------------------------
+| The pattern for a default generated single-precision NaN.
+*----------------------------------------------------------------------------*/
+extern const float32 float32_default_nan;
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE double-precision conversion routines.
+*----------------------------------------------------------------------------*/
+int_fast16_t float64_to_int16_round_to_zero(float64 STATUS_PARAM);
+uint_fast16_t float64_to_uint16_round_to_zero(float64 STATUS_PARAM);
+int32 float64_to_int32( float64 STATUS_PARAM );
+int32 float64_to_int32_round_to_zero( float64 STATUS_PARAM );
+uint32 float64_to_uint32( float64 STATUS_PARAM );
+uint32 float64_to_uint32_round_to_zero( float64 STATUS_PARAM );
+int64 float64_to_int64( float64 STATUS_PARAM );
+int64 float64_to_int64_round_to_zero( float64 STATUS_PARAM );
+uint64 float64_to_uint64 (float64 a STATUS_PARAM);
+uint64 float64_to_uint64_round_to_zero (float64 a STATUS_PARAM);
+float32 float64_to_float32( float64 STATUS_PARAM );
+floatx80 float64_to_floatx80( float64 STATUS_PARAM );
+float128 float64_to_float128( float64 STATUS_PARAM );
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE double-precision operations.
+*----------------------------------------------------------------------------*/
+float64 float64_round_to_int( float64 STATUS_PARAM );
+float64 float64_trunc_to_int( float64 STATUS_PARAM );
+float64 float64_add( float64, float64 STATUS_PARAM );
+float64 float64_sub( float64, float64 STATUS_PARAM );
+float64 float64_mul( float64, float64 STATUS_PARAM );
+float64 float64_div( float64, float64 STATUS_PARAM );
+float64 float64_rem( float64, float64 STATUS_PARAM );
+float64 float64_muladd(float64, float64, float64, int STATUS_PARAM);
+float64 float64_sqrt( float64 STATUS_PARAM );
+float64 float64_log2( float64 STATUS_PARAM );
+int float64_eq( float64, float64 STATUS_PARAM );
+int float64_le( float64, float64 STATUS_PARAM );
+int float64_lt( float64, float64 STATUS_PARAM );
+int float64_unordered( float64, float64 STATUS_PARAM );
+int float64_eq_quiet( float64, float64 STATUS_PARAM );
+int float64_le_quiet( float64, float64 STATUS_PARAM );
+int float64_lt_quiet( float64, float64 STATUS_PARAM );
+int float64_unordered_quiet( float64, float64 STATUS_PARAM );
+int float64_compare( float64, float64 STATUS_PARAM );
+int float64_compare_quiet( float64, float64 STATUS_PARAM );
+float64 float64_min(float64, float64 STATUS_PARAM);
+float64 float64_max(float64, float64 STATUS_PARAM);
+int float64_is_quiet_nan( float64 a );
+int float64_is_signaling_nan( float64 );
+float64 float64_maybe_silence_nan( float64 );
+float64 float64_scalbn( float64, int STATUS_PARAM );
+
+INLINE float64 float64_abs(float64 a)
+{
+    /* Note that abs does *not* handle NaN specially, nor does
+     * it flush denormal inputs to zero.
+     */
+    return make_float64(float64_val(a) & 0x7fffffffffffffffLL);
+}
+
+INLINE float64 float64_chs(float64 a)
+{
+    /* Note that chs does *not* handle NaN specially, nor does
+     * it flush denormal inputs to zero.
+     */
+    return make_float64(float64_val(a) ^ 0x8000000000000000LL);
+}
+
+INLINE int float64_is_infinity(float64 a)
+{
+    return (float64_val(a) & 0x7fffffffffffffffLL ) == 0x7ff0000000000000LL;
+}
+
+INLINE int float64_is_neg(float64 a)
+{
+    return float64_val(a) >> 63;
+}
+
+INLINE int float64_is_zero(float64 a)
+{
+    return (float64_val(a) & 0x7fffffffffffffffLL) == 0;
+}
+
+INLINE int float64_is_any_nan(float64 a)
+{
+    return ((float64_val(a) & ~(1ULL << 63)) > 0x7ff0000000000000ULL);
+}
+
+INLINE int float64_is_zero_or_denormal(float64 a)
+{
+    return (float64_val(a) & 0x7ff0000000000000LL) == 0;
+}
+
+INLINE float64 float64_set_sign(float64 a, int sign)
+{
+    return make_float64((float64_val(a) & 0x7fffffffffffffffULL)
+                        | ((int64_t)sign << 63));
+}
+
+#define float64_zero make_float64(0)
+#define float64_one make_float64(0x3ff0000000000000LL)
+#define float64_ln2 make_float64(0x3fe62e42fefa39efLL)
+#define float64_pi make_float64(0x400921fb54442d18LL)
+#define float64_half make_float64(0x3fe0000000000000LL)
+#define float64_infinity make_float64(0x7ff0000000000000LL)
+
+/*----------------------------------------------------------------------------
+| The pattern for a default generated double-precision NaN.
+*----------------------------------------------------------------------------*/
+extern const float64 float64_default_nan;
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE extended double-precision conversion routines.
+*----------------------------------------------------------------------------*/
+int32 floatx80_to_int32( floatx80 STATUS_PARAM );
+int32 floatx80_to_int32_round_to_zero( floatx80 STATUS_PARAM );
+int64 floatx80_to_int64( floatx80 STATUS_PARAM );
+int64 floatx80_to_int64_round_to_zero( floatx80 STATUS_PARAM );
+float32 floatx80_to_float32( floatx80 STATUS_PARAM );
+float64 floatx80_to_float64( floatx80 STATUS_PARAM );
+float128 floatx80_to_float128( floatx80 STATUS_PARAM );
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE extended double-precision operations.
+*----------------------------------------------------------------------------*/
+floatx80 floatx80_round_to_int( floatx80 STATUS_PARAM );
+floatx80 floatx80_add( floatx80, floatx80 STATUS_PARAM );
+floatx80 floatx80_sub( floatx80, floatx80 STATUS_PARAM );
+floatx80 floatx80_mul( floatx80, floatx80 STATUS_PARAM );
+floatx80 floatx80_div( floatx80, floatx80 STATUS_PARAM );
+floatx80 floatx80_rem( floatx80, floatx80 STATUS_PARAM );
+floatx80 floatx80_sqrt( floatx80 STATUS_PARAM );
+int floatx80_eq( floatx80, floatx80 STATUS_PARAM );
+int floatx80_le( floatx80, floatx80 STATUS_PARAM );
+int floatx80_lt( floatx80, floatx80 STATUS_PARAM );
+int floatx80_unordered( floatx80, floatx80 STATUS_PARAM );
+int floatx80_eq_quiet( floatx80, floatx80 STATUS_PARAM );
+int floatx80_le_quiet( floatx80, floatx80 STATUS_PARAM );
+int floatx80_lt_quiet( floatx80, floatx80 STATUS_PARAM );
+int floatx80_unordered_quiet( floatx80, floatx80 STATUS_PARAM );
+int floatx80_compare( floatx80, floatx80 STATUS_PARAM );
+int floatx80_compare_quiet( floatx80, floatx80 STATUS_PARAM );
+int floatx80_is_quiet_nan( floatx80 );
+int floatx80_is_signaling_nan( floatx80 );
+floatx80 floatx80_maybe_silence_nan( floatx80 );
+floatx80 floatx80_scalbn( floatx80, int STATUS_PARAM );
+
+INLINE floatx80 floatx80_abs(floatx80 a)
+{
+    a.high &= 0x7fff;
+    return a;
+}
+
+INLINE floatx80 floatx80_chs(floatx80 a)
+{
+    a.high ^= 0x8000;
+    return a;
+}
+
+INLINE int floatx80_is_infinity(floatx80 a)
+{
+    return (a.high & 0x7fff) == 0x7fff && a.low == 0x8000000000000000LL;
+}
+
+INLINE int floatx80_is_neg(floatx80 a)
+{
+    return a.high >> 15;
+}
+
+INLINE int floatx80_is_zero(floatx80 a)
+{
+    return (a.high & 0x7fff) == 0 && a.low == 0;
+}
+
+INLINE int floatx80_is_zero_or_denormal(floatx80 a)
+{
+    return (a.high & 0x7fff) == 0;
+}
+
+INLINE int floatx80_is_any_nan(floatx80 a)
+{
+    return ((a.high & 0x7fff) == 0x7fff) && (a.low<<1);
+}
+
+#define floatx80_zero make_floatx80(0x0000, 0x0000000000000000LL)
+#define floatx80_one make_floatx80(0x3fff, 0x8000000000000000LL)
+#define floatx80_ln2 make_floatx80(0x3ffe, 0xb17217f7d1cf79acLL)
+#define floatx80_pi make_floatx80(0x4000, 0xc90fdaa22168c235LL)
+#define floatx80_half make_floatx80(0x3ffe, 0x8000000000000000LL)
+#define floatx80_infinity make_floatx80(0x7fff, 0x8000000000000000LL)
+
+/*----------------------------------------------------------------------------
+| The pattern for a default generated extended double-precision NaN.
+*----------------------------------------------------------------------------*/
+extern const floatx80 floatx80_default_nan;
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE quadruple-precision conversion routines.
+*----------------------------------------------------------------------------*/
+int32 float128_to_int32( float128 STATUS_PARAM );
+int32 float128_to_int32_round_to_zero( float128 STATUS_PARAM );
+int64 float128_to_int64( float128 STATUS_PARAM );
+int64 float128_to_int64_round_to_zero( float128 STATUS_PARAM );
+float32 float128_to_float32( float128 STATUS_PARAM );
+float64 float128_to_float64( float128 STATUS_PARAM );
+floatx80 float128_to_floatx80( float128 STATUS_PARAM );
+
+/*----------------------------------------------------------------------------
+| Software IEC/IEEE quadruple-precision operations.
+*----------------------------------------------------------------------------*/
+float128 float128_round_to_int( float128 STATUS_PARAM );
+float128 float128_add( float128, float128 STATUS_PARAM );
+float128 float128_sub( float128, float128 STATUS_PARAM );
+float128 float128_mul( float128, float128 STATUS_PARAM );
+float128 float128_div( float128, float128 STATUS_PARAM );
+float128 float128_rem( float128, float128 STATUS_PARAM );
+float128 float128_sqrt( float128 STATUS_PARAM );
+int float128_eq( float128, float128 STATUS_PARAM );
+int float128_le( float128, float128 STATUS_PARAM );
+int float128_lt( float128, float128 STATUS_PARAM );
+int float128_unordered( float128, float128 STATUS_PARAM );
+int float128_eq_quiet( float128, float128 STATUS_PARAM );
+int float128_le_quiet( float128, float128 STATUS_PARAM );
+int float128_lt_quiet( float128, float128 STATUS_PARAM );
+int float128_unordered_quiet( float128, float128 STATUS_PARAM );
+int float128_compare( float128, float128 STATUS_PARAM );
+int float128_compare_quiet( float128, float128 STATUS_PARAM );
+int float128_is_quiet_nan( float128 );
+int float128_is_signaling_nan( float128 );
+float128 float128_maybe_silence_nan( float128 );
+float128 float128_scalbn( float128, int STATUS_PARAM );
+
+INLINE float128 float128_abs(float128 a)
+{
+    a.high &= 0x7fffffffffffffffLL;
+    return a;
+}
+
+INLINE float128 float128_chs(float128 a)
+{
+    a.high ^= 0x8000000000000000LL;
+    return a;
+}
+
+INLINE int float128_is_infinity(float128 a)
+{
+    return (a.high & 0x7fffffffffffffffLL) == 0x7fff000000000000LL && a.low == 0;
+}
+
+INLINE int float128_is_neg(float128 a)
+{
+    return a.high >> 63;
+}
+
+INLINE int float128_is_zero(float128 a)
+{
+    return (a.high & 0x7fffffffffffffffLL) == 0 && a.low == 0;
+}
+
+INLINE int float128_is_zero_or_denormal(float128 a)
+{
+    return (a.high & 0x7fff000000000000LL) == 0;
+}
+
+INLINE int float128_is_any_nan(float128 a)
+{
+    return ((a.high >> 48) & 0x7fff) == 0x7fff &&
+        ((a.low != 0) || ((a.high & 0xffffffffffffLL) != 0));
+}
+
+#define float128_zero make_float128(0, 0)
+
+/*----------------------------------------------------------------------------
+| The pattern for a default generated quadruple-precision NaN.
+*----------------------------------------------------------------------------*/
+extern const float128 float128_default_nan;
+
+#endif /* !SOFTFLOAT_H */
diff --git a/contrib/qemu/include/glib-compat.h b/contrib/qemu/include/glib-compat.h
new file mode 100644
index 000000000..8aa77afd6
--- /dev/null
+++ b/contrib/qemu/include/glib-compat.h
@@ -0,0 +1,27 @@
+/*
+ * GLIB Compatibility Functions
+ *
+ * Copyright IBM, Corp. 2013
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.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.
+ *
+ */
+
+#ifndef QEMU_GLIB_COMPAT_H
+#define QEMU_GLIB_COMPAT_H
+
+#include <glib.h>
+
+#if !GLIB_CHECK_VERSION(2, 14, 0)
+static inline guint g_timeout_add_seconds(guint interval, GSourceFunc function,
+                                          gpointer data)
+{
+    return g_timeout_add(interval * 1000, function, data);
+}
+#endif
+
+#endif
diff --git a/contrib/qemu/include/migration/migration.h b/contrib/qemu/include/migration/migration.h
new file mode 100644
index 000000000..bc9fde0b2
--- /dev/null
+++ b/contrib/qemu/include/migration/migration.h
@@ -0,0 +1,157 @@
+/*
+ * QEMU live migration
+ *
+ * Copyright IBM, Corp. 2008
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef QEMU_MIGRATION_H
+#define QEMU_MIGRATION_H
+
+#include "qapi/qmp/qdict.h"
+#include "qemu-common.h"
+#include "qemu/thread.h"
+#include "qemu/notify.h"
+#include "qapi/error.h"
+#include "migration/vmstate.h"
+#include "qapi-types.h"
+#include "exec/cpu-common.h"
+
+struct MigrationParams {
+    bool blk;
+    bool shared;
+};
+
+typedef struct MigrationState MigrationState;
+
+struct MigrationState
+{
+    int64_t bandwidth_limit;
+    size_t bytes_xfer;
+    size_t xfer_limit;
+    QemuThread thread;
+    QEMUBH *cleanup_bh;
+    QEMUFile *file;
+
+    int state;
+    MigrationParams params;
+    double mbps;
+    int64_t total_time;
+    int64_t downtime;
+    int64_t expected_downtime;
+    int64_t dirty_pages_rate;
+    int64_t dirty_bytes_rate;
+    bool enabled_capabilities[MIGRATION_CAPABILITY_MAX];
+    int64_t xbzrle_cache_size;
+};
+
+void process_incoming_migration(QEMUFile *f);
+
+void qemu_start_incoming_migration(const char *uri, Error **errp);
+
+uint64_t migrate_max_downtime(void);
+
+void do_info_migrate_print(Monitor *mon, const QObject *data);
+
+void do_info_migrate(Monitor *mon, QObject **ret_data);
+
+void exec_start_incoming_migration(const char *host_port, Error **errp);
+
+void exec_start_outgoing_migration(MigrationState *s, const char *host_port, Error **errp);
+
+void tcp_start_incoming_migration(const char *host_port, Error **errp);
+
+void tcp_start_outgoing_migration(MigrationState *s, const char *host_port, Error **errp);
+
+void unix_start_incoming_migration(const char *path, Error **errp);
+
+void unix_start_outgoing_migration(MigrationState *s, const char *path, Error **errp);
+
+void fd_start_incoming_migration(const char *path, Error **errp);
+
+void fd_start_outgoing_migration(MigrationState *s, const char *fdname, Error **errp);
+
+void migrate_fd_error(MigrationState *s);
+
+void migrate_fd_connect(MigrationState *s);
+
+int migrate_fd_close(MigrationState *s);
+
+void add_migration_state_change_notifier(Notifier *notify);
+void remove_migration_state_change_notifier(Notifier *notify);
+bool migration_is_active(MigrationState *);
+bool migration_has_finished(MigrationState *);
+bool migration_has_failed(MigrationState *);
+MigrationState *migrate_get_current(void);
+
+uint64_t ram_bytes_remaining(void);
+uint64_t ram_bytes_transferred(void);
+uint64_t ram_bytes_total(void);
+
+void acct_update_position(QEMUFile *f, size_t size, bool zero);
+
+extern SaveVMHandlers savevm_ram_handlers;
+
+uint64_t dup_mig_bytes_transferred(void);
+uint64_t dup_mig_pages_transferred(void);
+uint64_t skipped_mig_bytes_transferred(void);
+uint64_t skipped_mig_pages_transferred(void);
+uint64_t norm_mig_bytes_transferred(void);
+uint64_t norm_mig_pages_transferred(void);
+uint64_t xbzrle_mig_bytes_transferred(void);
+uint64_t xbzrle_mig_pages_transferred(void);
+uint64_t xbzrle_mig_pages_overflow(void);
+uint64_t xbzrle_mig_pages_cache_miss(void);
+
+/**
+ * @migrate_add_blocker - prevent migration from proceeding
+ *
+ * @reason - an error to be returned whenever migration is attempted
+ */
+void migrate_add_blocker(Error *reason);
+
+/**
+ * @migrate_del_blocker - remove a blocking error from migration
+ *
+ * @reason - the error blocking migration
+ */
+void migrate_del_blocker(Error *reason);
+
+bool migrate_rdma_pin_all(void);
+
+bool migrate_auto_converge(void);
+
+int xbzrle_encode_buffer(uint8_t *old_buf, uint8_t *new_buf, int slen,
+                         uint8_t *dst, int dlen);
+int xbzrle_decode_buffer(uint8_t *src, int slen, uint8_t *dst, int dlen);
+
+int migrate_use_xbzrle(void);
+int64_t migrate_xbzrle_cache_size(void);
+
+int64_t xbzrle_cache_resize(int64_t new_size);
+
+void ram_control_before_iterate(QEMUFile *f, uint64_t flags);
+void ram_control_after_iterate(QEMUFile *f, uint64_t flags);
+void ram_control_load_hook(QEMUFile *f, uint64_t flags);
+
+/* Whenever this is found in the data stream, the flags
+ * will be passed to ram_control_load_hook in the incoming-migration
+ * side. This lets before_ram_iterate/after_ram_iterate add
+ * transport-specific sections to the RAM migration data.
+ */
+#define RAM_SAVE_FLAG_HOOK     0x80
+
+#define RAM_SAVE_CONTROL_NOT_SUPP -1000
+#define RAM_SAVE_CONTROL_DELAYED  -2000
+
+size_t ram_control_save_page(QEMUFile *f, ram_addr_t block_offset,
+                             ram_addr_t offset, size_t size,
+                             int *bytes_sent);
+
+#endif
diff --git a/contrib/qemu/include/migration/qemu-file.h b/contrib/qemu/include/migration/qemu-file.h
new file mode 100644
index 000000000..0f757fbeb
--- /dev/null
+++ b/contrib/qemu/include/migration/qemu-file.h
@@ -0,0 +1,266 @@
+/*
+ * QEMU System Emulator
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#ifndef QEMU_FILE_H
+#define QEMU_FILE_H 1
+#include "exec/cpu-common.h"
+
+/* This function writes a chunk of data to a file at the given position.
+ * The pos argument can be ignored if the file is only being used for
+ * streaming.  The handler should try to write all of the data it can.
+ */
+typedef int (QEMUFilePutBufferFunc)(void *opaque, const uint8_t *buf,
+                                    int64_t pos, int size);
+
+/* Read a chunk of data from a file at the given position.  The pos argument
+ * can be ignored if the file is only be used for streaming.  The number of
+ * bytes actually read should be returned.
+ */
+typedef int (QEMUFileGetBufferFunc)(void *opaque, uint8_t *buf,
+                                    int64_t pos, int size);
+
+/* Close a file
+ *
+ * Return negative error number on error, 0 or positive value on success.
+ *
+ * The meaning of return value on success depends on the specific back-end being
+ * used.
+ */
+typedef int (QEMUFileCloseFunc)(void *opaque);
+
+/* Called to return the OS file descriptor associated to the QEMUFile.
+ */
+typedef int (QEMUFileGetFD)(void *opaque);
+
+/*
+ * This function writes an iovec to file.
+ */
+typedef ssize_t (QEMUFileWritevBufferFunc)(void *opaque, struct iovec *iov,
+                                           int iovcnt, int64_t pos);
+
+/*
+ * This function provides hooks around different
+ * stages of RAM migration.
+ */
+typedef int (QEMURamHookFunc)(QEMUFile *f, void *opaque, uint64_t flags);
+
+/*
+ * Constants used by ram_control_* hooks
+ */
+#define RAM_CONTROL_SETUP    0
+#define RAM_CONTROL_ROUND    1
+#define RAM_CONTROL_HOOK     2
+#define RAM_CONTROL_FINISH   3
+
+/*
+ * This function allows override of where the RAM page
+ * is saved (such as RDMA, for example.)
+ */
+typedef size_t (QEMURamSaveFunc)(QEMUFile *f, void *opaque,
+                               ram_addr_t block_offset,
+                               ram_addr_t offset,
+                               size_t size,
+                               int *bytes_sent);
+
+typedef struct QEMUFileOps {
+    QEMUFilePutBufferFunc *put_buffer;
+    QEMUFileGetBufferFunc *get_buffer;
+    QEMUFileCloseFunc *close;
+    QEMUFileGetFD *get_fd;
+    QEMUFileWritevBufferFunc *writev_buffer;
+    QEMURamHookFunc *before_ram_iterate;
+    QEMURamHookFunc *after_ram_iterate;
+    QEMURamHookFunc *hook_ram_load;
+    QEMURamSaveFunc *save_page;
+} QEMUFileOps;
+
+QEMUFile *qemu_fopen_ops(void *opaque, const QEMUFileOps *ops);
+QEMUFile *qemu_fopen(const char *filename, const char *mode);
+QEMUFile *qemu_fdopen(int fd, const char *mode);
+QEMUFile *qemu_fopen_socket(int fd, const char *mode);
+QEMUFile *qemu_popen_cmd(const char *command, const char *mode);
+int qemu_get_fd(QEMUFile *f);
+int qemu_fclose(QEMUFile *f);
+int64_t qemu_ftell(QEMUFile *f);
+void qemu_put_buffer(QEMUFile *f, const uint8_t *buf, int size);
+void qemu_put_byte(QEMUFile *f, int v);
+/*
+ * put_buffer without copying the buffer.
+ * The buffer should be available till it is sent asynchronously.
+ */
+void qemu_put_buffer_async(QEMUFile *f, const uint8_t *buf, int size);
+bool qemu_file_mode_is_not_valid(const char *mode);
+
+static inline void qemu_put_ubyte(QEMUFile *f, unsigned int v)
+{
+    qemu_put_byte(f, (int)v);
+}
+
+#define qemu_put_sbyte qemu_put_byte
+
+void qemu_put_be16(QEMUFile *f, unsigned int v);
+void qemu_put_be32(QEMUFile *f, unsigned int v);
+void qemu_put_be64(QEMUFile *f, uint64_t v);
+int qemu_get_buffer(QEMUFile *f, uint8_t *buf, int size);
+int qemu_get_byte(QEMUFile *f);
+void qemu_update_position(QEMUFile *f, size_t size);
+
+static inline unsigned int qemu_get_ubyte(QEMUFile *f)
+{
+    return (unsigned int)qemu_get_byte(f);
+}
+
+#define qemu_get_sbyte qemu_get_byte
+
+unsigned int qemu_get_be16(QEMUFile *f);
+unsigned int qemu_get_be32(QEMUFile *f);
+uint64_t qemu_get_be64(QEMUFile *f);
+
+int qemu_file_rate_limit(QEMUFile *f);
+void qemu_file_reset_rate_limit(QEMUFile *f);
+void qemu_file_set_rate_limit(QEMUFile *f, int64_t new_rate);
+int64_t qemu_file_get_rate_limit(QEMUFile *f);
+int qemu_file_get_error(QEMUFile *f);
+void qemu_fflush(QEMUFile *f);
+
+static inline void qemu_put_be64s(QEMUFile *f, const uint64_t *pv)
+{
+    qemu_put_be64(f, *pv);
+}
+
+static inline void qemu_put_be32s(QEMUFile *f, const uint32_t *pv)
+{
+    qemu_put_be32(f, *pv);
+}
+
+static inline void qemu_put_be16s(QEMUFile *f, const uint16_t *pv)
+{
+    qemu_put_be16(f, *pv);
+}
+
+static inline void qemu_put_8s(QEMUFile *f, const uint8_t *pv)
+{
+    qemu_put_byte(f, *pv);
+}
+
+static inline void qemu_get_be64s(QEMUFile *f, uint64_t *pv)
+{
+    *pv = qemu_get_be64(f);
+}
+
+static inline void qemu_get_be32s(QEMUFile *f, uint32_t *pv)
+{
+    *pv = qemu_get_be32(f);
+}
+
+static inline void qemu_get_be16s(QEMUFile *f, uint16_t *pv)
+{
+    *pv = qemu_get_be16(f);
+}
+
+static inline void qemu_get_8s(QEMUFile *f, uint8_t *pv)
+{
+    *pv = qemu_get_byte(f);
+}
+
+// Signed versions for type safety
+static inline void qemu_put_sbuffer(QEMUFile *f, const int8_t *buf, int size)
+{
+    qemu_put_buffer(f, (const uint8_t *)buf, size);
+}
+
+static inline void qemu_put_sbe16(QEMUFile *f, int v)
+{
+    qemu_put_be16(f, (unsigned int)v);
+}
+
+static inline void qemu_put_sbe32(QEMUFile *f, int v)
+{
+    qemu_put_be32(f, (unsigned int)v);
+}
+
+static inline void qemu_put_sbe64(QEMUFile *f, int64_t v)
+{
+    qemu_put_be64(f, (uint64_t)v);
+}
+
+static inline size_t qemu_get_sbuffer(QEMUFile *f, int8_t *buf, int size)
+{
+    return qemu_get_buffer(f, (uint8_t *)buf, size);
+}
+
+static inline int qemu_get_sbe16(QEMUFile *f)
+{
+    return (int)qemu_get_be16(f);
+}
+
+static inline int qemu_get_sbe32(QEMUFile *f)
+{
+    return (int)qemu_get_be32(f);
+}
+
+static inline int64_t qemu_get_sbe64(QEMUFile *f)
+{
+    return (int64_t)qemu_get_be64(f);
+}
+
+static inline void qemu_put_s8s(QEMUFile *f, const int8_t *pv)
+{
+    qemu_put_8s(f, (const uint8_t *)pv);
+}
+
+static inline void qemu_put_sbe16s(QEMUFile *f, const int16_t *pv)
+{
+    qemu_put_be16s(f, (const uint16_t *)pv);
+}
+
+static inline void qemu_put_sbe32s(QEMUFile *f, const int32_t *pv)
+{
+    qemu_put_be32s(f, (const uint32_t *)pv);
+}
+
+static inline void qemu_put_sbe64s(QEMUFile *f, const int64_t *pv)
+{
+    qemu_put_be64s(f, (const uint64_t *)pv);
+}
+
+static inline void qemu_get_s8s(QEMUFile *f, int8_t *pv)
+{
+    qemu_get_8s(f, (uint8_t *)pv);
+}
+
+static inline void qemu_get_sbe16s(QEMUFile *f, int16_t *pv)
+{
+    qemu_get_be16s(f, (uint16_t *)pv);
+}
+
+static inline void qemu_get_sbe32s(QEMUFile *f, int32_t *pv)
+{
+    qemu_get_be32s(f, (uint32_t *)pv);
+}
+
+static inline void qemu_get_sbe64s(QEMUFile *f, int64_t *pv)
+{
+    qemu_get_be64s(f, (uint64_t *)pv);
+}
+#endif
diff --git a/contrib/qemu/include/migration/vmstate.h b/contrib/qemu/include/migration/vmstate.h
new file mode 100644
index 000000000..1c31b5d6f
--- /dev/null
+++ b/contrib/qemu/include/migration/vmstate.h
@@ -0,0 +1,740 @@
+/*
+ * QEMU migration/snapshot declarations
+ *
+ * Copyright (c) 2009-2011 Red Hat, Inc.
+ *
+ * Original author: Juan Quintela <quintela@redhat.com>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#ifndef QEMU_VMSTATE_H
+#define QEMU_VMSTATE_H 1
+
+#ifndef CONFIG_USER_ONLY
+#include <migration/qemu-file.h>
+#endif
+
+typedef void SaveStateHandler(QEMUFile *f, void *opaque);
+typedef int LoadStateHandler(QEMUFile *f, void *opaque, int version_id);
+
+typedef struct SaveVMHandlers {
+    /* This runs inside the iothread lock.  */
+    void (*set_params)(const MigrationParams *params, void * opaque);
+    SaveStateHandler *save_state;
+
+    void (*cancel)(void *opaque);
+    int (*save_live_complete)(QEMUFile *f, void *opaque);
+
+    /* This runs both outside and inside the iothread lock.  */
+    bool (*is_active)(void *opaque);
+
+    /* This runs outside the iothread lock in the migration case, and
+     * within the lock in the savevm case.  The callback had better only
+     * use data that is local to the migration thread or protected
+     * by other locks.
+     */
+    int (*save_live_iterate)(QEMUFile *f, void *opaque);
+
+    /* This runs outside the iothread lock!  */
+    int (*save_live_setup)(QEMUFile *f, void *opaque);
+    uint64_t (*save_live_pending)(QEMUFile *f, void *opaque, uint64_t max_size);
+
+    LoadStateHandler *load_state;
+} SaveVMHandlers;
+
+int register_savevm(DeviceState *dev,
+                    const char *idstr,
+                    int instance_id,
+                    int version_id,
+                    SaveStateHandler *save_state,
+                    LoadStateHandler *load_state,
+                    void *opaque);
+
+int register_savevm_live(DeviceState *dev,
+                         const char *idstr,
+                         int instance_id,
+                         int version_id,
+                         SaveVMHandlers *ops,
+                         void *opaque);
+
+void unregister_savevm(DeviceState *dev, const char *idstr, void *opaque);
+void register_device_unmigratable(DeviceState *dev, const char *idstr,
+                                                                void *opaque);
+
+
+typedef struct VMStateInfo VMStateInfo;
+typedef struct VMStateDescription VMStateDescription;
+
+struct VMStateInfo {
+    const char *name;
+    int (*get)(QEMUFile *f, void *pv, size_t size);
+    void (*put)(QEMUFile *f, void *pv, size_t size);
+};
+
+enum VMStateFlags {
+    VMS_SINGLE           = 0x001,
+    VMS_POINTER          = 0x002,
+    VMS_ARRAY            = 0x004,
+    VMS_STRUCT           = 0x008,
+    VMS_VARRAY_INT32     = 0x010,  /* Array with size in int32_t field*/
+    VMS_BUFFER           = 0x020,  /* static sized buffer */
+    VMS_ARRAY_OF_POINTER = 0x040,
+    VMS_VARRAY_UINT16    = 0x080,  /* Array with size in uint16_t field */
+    VMS_VBUFFER          = 0x100,  /* Buffer with size in int32_t field */
+    VMS_MULTIPLY         = 0x200,  /* multiply "size" field by field_size */
+    VMS_VARRAY_UINT8     = 0x400,  /* Array with size in uint8_t field*/
+    VMS_VARRAY_UINT32    = 0x800,  /* Array with size in uint32_t field*/
+};
+
+typedef struct {
+    const char *name;
+    size_t offset;
+    size_t size;
+    size_t start;
+    int num;
+    size_t num_offset;
+    size_t size_offset;
+    const VMStateInfo *info;
+    enum VMStateFlags flags;
+    const VMStateDescription *vmsd;
+    int version_id;
+    bool (*field_exists)(void *opaque, int version_id);
+} VMStateField;
+
+typedef struct VMStateSubsection {
+    const VMStateDescription *vmsd;
+    bool (*needed)(void *opaque);
+} VMStateSubsection;
+
+struct VMStateDescription {
+    const char *name;
+    int unmigratable;
+    int version_id;
+    int minimum_version_id;
+    int minimum_version_id_old;
+    LoadStateHandler *load_state_old;
+    int (*pre_load)(void *opaque);
+    int (*post_load)(void *opaque, int version_id);
+    void (*pre_save)(void *opaque);
+    VMStateField *fields;
+    const VMStateSubsection *subsections;
+};
+
+#ifdef CONFIG_USER_ONLY
+extern const VMStateDescription vmstate_dummy;
+#endif
+
+extern const VMStateInfo vmstate_info_bool;
+
+extern const VMStateInfo vmstate_info_int8;
+extern const VMStateInfo vmstate_info_int16;
+extern const VMStateInfo vmstate_info_int32;
+extern const VMStateInfo vmstate_info_int64;
+
+extern const VMStateInfo vmstate_info_uint8_equal;
+extern const VMStateInfo vmstate_info_uint16_equal;
+extern const VMStateInfo vmstate_info_int32_equal;
+extern const VMStateInfo vmstate_info_uint32_equal;
+extern const VMStateInfo vmstate_info_uint64_equal;
+extern const VMStateInfo vmstate_info_int32_le;
+
+extern const VMStateInfo vmstate_info_uint8;
+extern const VMStateInfo vmstate_info_uint16;
+extern const VMStateInfo vmstate_info_uint32;
+extern const VMStateInfo vmstate_info_uint64;
+
+extern const VMStateInfo vmstate_info_float64;
+
+extern const VMStateInfo vmstate_info_timer;
+extern const VMStateInfo vmstate_info_buffer;
+extern const VMStateInfo vmstate_info_unused_buffer;
+extern const VMStateInfo vmstate_info_bitmap;
+
+#define type_check_2darray(t1,t2,n,m) ((t1(*)[n][m])0 - (t2*)0)
+#define type_check_array(t1,t2,n) ((t1(*)[n])0 - (t2*)0)
+#define type_check_pointer(t1,t2) ((t1**)0 - (t2*)0)
+
+#define vmstate_offset_value(_state, _field, _type)                  \
+    (offsetof(_state, _field) +                                      \
+     type_check(_type, typeof_field(_state, _field)))
+
+#define vmstate_offset_pointer(_state, _field, _type)                \
+    (offsetof(_state, _field) +                                      \
+     type_check_pointer(_type, typeof_field(_state, _field)))
+
+#define vmstate_offset_array(_state, _field, _type, _num)            \
+    (offsetof(_state, _field) +                                      \
+     type_check_array(_type, typeof_field(_state, _field), _num))
+
+#define vmstate_offset_2darray(_state, _field, _type, _n1, _n2)      \
+    (offsetof(_state, _field) +                                      \
+     type_check_2darray(_type, typeof_field(_state, _field), _n1, _n2))
+
+#define vmstate_offset_sub_array(_state, _field, _type, _start)      \
+    (offsetof(_state, _field[_start]))
+
+#define vmstate_offset_buffer(_state, _field)                        \
+    vmstate_offset_array(_state, _field, uint8_t,                    \
+                         sizeof(typeof_field(_state, _field)))
+
+#define VMSTATE_SINGLE_TEST(_field, _state, _test, _version, _info, _type) { \
+    .name         = (stringify(_field)),                             \
+    .version_id   = (_version),                                      \
+    .field_exists = (_test),                                         \
+    .size         = sizeof(_type),                                   \
+    .info         = &(_info),                                        \
+    .flags        = VMS_SINGLE,                                      \
+    .offset       = vmstate_offset_value(_state, _field, _type),     \
+}
+
+#define VMSTATE_POINTER(_field, _state, _version, _info, _type) {    \
+    .name       = (stringify(_field)),                               \
+    .version_id = (_version),                                        \
+    .info       = &(_info),                                          \
+    .size       = sizeof(_type),                                     \
+    .flags      = VMS_SINGLE|VMS_POINTER,                            \
+    .offset     = vmstate_offset_value(_state, _field, _type),       \
+}
+
+#define VMSTATE_POINTER_TEST(_field, _state, _test, _info, _type) {  \
+    .name       = (stringify(_field)),                               \
+    .info       = &(_info),                                          \
+    .field_exists = (_test),                                         \
+    .size       = sizeof(_type),                                     \
+    .flags      = VMS_SINGLE|VMS_POINTER,                            \
+    .offset     = vmstate_offset_value(_state, _field, _type),       \
+}
+
+#define VMSTATE_ARRAY(_field, _state, _num, _version, _info, _type) {\
+    .name       = (stringify(_field)),                               \
+    .version_id = (_version),                                        \
+    .num        = (_num),                                            \
+    .info       = &(_info),                                          \
+    .size       = sizeof(_type),                                     \
+    .flags      = VMS_ARRAY,                                         \
+    .offset     = vmstate_offset_array(_state, _field, _type, _num), \
+}
+
+#define VMSTATE_2DARRAY(_field, _state, _n1, _n2, _version, _info, _type) { \
+    .name       = (stringify(_field)),                                      \
+    .version_id = (_version),                                               \
+    .num        = (_n1) * (_n2),                                            \
+    .info       = &(_info),                                                 \
+    .size       = sizeof(_type),                                            \
+    .flags      = VMS_ARRAY,                                                \
+    .offset     = vmstate_offset_2darray(_state, _field, _type, _n1, _n2),  \
+}
+
+#define VMSTATE_ARRAY_TEST(_field, _state, _num, _test, _info, _type) {\
+    .name         = (stringify(_field)),                              \
+    .field_exists = (_test),                                          \
+    .num          = (_num),                                           \
+    .info         = &(_info),                                         \
+    .size         = sizeof(_type),                                    \
+    .flags        = VMS_ARRAY,                                        \
+    .offset       = vmstate_offset_array(_state, _field, _type, _num),\
+}
+
+#define VMSTATE_SUB_ARRAY(_field, _state, _start, _num, _version, _info, _type) { \
+    .name       = (stringify(_field)),                               \
+    .version_id = (_version),                                        \
+    .num        = (_num),                                            \
+    .info       = &(_info),                                          \
+    .size       = sizeof(_type),                                     \
+    .flags      = VMS_ARRAY,                                         \
+    .offset     = vmstate_offset_sub_array(_state, _field, _type, _start), \
+}
+
+#define VMSTATE_ARRAY_INT32_UNSAFE(_field, _state, _field_num, _info, _type) {\
+    .name       = (stringify(_field)),                               \
+    .num_offset = vmstate_offset_value(_state, _field_num, int32_t), \
+    .info       = &(_info),                                          \
+    .size       = sizeof(_type),                                     \
+    .flags      = VMS_VARRAY_INT32,                                  \
+    .offset     = offsetof(_state, _field),                          \
+}
+
+#define VMSTATE_VARRAY_INT32(_field, _state, _field_num, _version, _info, _type) {\
+    .name       = (stringify(_field)),                               \
+    .version_id = (_version),                                        \
+    .num_offset = vmstate_offset_value(_state, _field_num, int32_t), \
+    .info       = &(_info),                                          \
+    .size       = sizeof(_type),                                     \
+    .flags      = VMS_VARR