Imported Upstream version 2.5.1.1upstream/2.5.1.1

Change-Id: Ie290b0e68882590d8a64fab165a943940b7c98ed

46 files changed, 2605 insertions, 1480 deletions

diff --git a/block/accounting.c b/block/accounting.c
index 01d594ffd..185025ec1 100644
--- a/block/accounting.c
+++ b/block/accounting.c
@@ -2,6 +2,7 @@
  * QEMU System Emulator block accounting
  *
  * Copyright (c) 2011 Christoph Hellwig
+ * Copyright (c) 2015 Igalia, S.L.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
@@ -25,6 +26,54 @@
 #include "block/accounting.h"
 #include "block/block_int.h"
 #include "qemu/timer.h"
+#include "sysemu/qtest.h"
+
+static QEMUClockType clock_type = QEMU_CLOCK_REALTIME;
+static const int qtest_latency_ns = NANOSECONDS_PER_SECOND / 1000;
+
+void block_acct_init(BlockAcctStats *stats, bool account_invalid,
+                     bool account_failed)
+{
+    stats->account_invalid = account_invalid;
+    stats->account_failed = account_failed;
+
+    if (qtest_enabled()) {
+        clock_type = QEMU_CLOCK_VIRTUAL;
+    }
+}
+
+void block_acct_cleanup(BlockAcctStats *stats)
+{
+    BlockAcctTimedStats *s, *next;
+    QSLIST_FOREACH_SAFE(s, &stats->intervals, entries, next) {
+        g_free(s);
+    }
+}
+
+void block_acct_add_interval(BlockAcctStats *stats, unsigned interval_length)
+{
+    BlockAcctTimedStats *s;
+    unsigned i;
+
+    s = g_new0(BlockAcctTimedStats, 1);
+    s->interval_length = interval_length;
+    QSLIST_INSERT_HEAD(&stats->intervals, s, entries);
+
+    for (i = 0; i < BLOCK_MAX_IOTYPE; i++) {
+        timed_average_init(&s->latency[i], clock_type,
+                           (uint64_t) interval_length * NANOSECONDS_PER_SECOND);
+    }
+}
+
+BlockAcctTimedStats *block_acct_interval_next(BlockAcctStats *stats,
+                                              BlockAcctTimedStats *s)
+{
+    if (s == NULL) {
+        return QSLIST_FIRST(&stats->intervals);
+    } else {
+        return QSLIST_NEXT(s, entries);
+    }
+}
 
 void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
                       int64_t bytes, enum BlockAcctType type)
@@ -32,26 +81,69 @@ void block_acct_start(BlockAcctStats *stats, BlockAcctCookie *cookie,
     assert(type < BLOCK_MAX_IOTYPE);
 
     cookie->bytes = bytes;
-    cookie->start_time_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
+    cookie->start_time_ns = qemu_clock_get_ns(clock_type);
     cookie->type = type;
 }
 
 void block_acct_done(BlockAcctStats *stats, BlockAcctCookie *cookie)
 {
+    BlockAcctTimedStats *s;
+    int64_t time_ns = qemu_clock_get_ns(clock_type);
+    int64_t latency_ns = time_ns - cookie->start_time_ns;
+
+    if (qtest_enabled()) {
+        latency_ns = qtest_latency_ns;
+    }
+
     assert(cookie->type < BLOCK_MAX_IOTYPE);
 
     stats->nr_bytes[cookie->type] += cookie->bytes;
     stats->nr_ops[cookie->type]++;
-    stats->total_time_ns[cookie->type] +=
-        qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - cookie->start_time_ns;
+    stats->total_time_ns[cookie->type] += latency_ns;
+    stats->last_access_time_ns = time_ns;
+
+    QSLIST_FOREACH(s, &stats->intervals, entries) {
+        timed_average_account(&s->latency[cookie->type], latency_ns);
+    }
 }
 
+void block_acct_failed(BlockAcctStats *stats, BlockAcctCookie *cookie)
+{
+    assert(cookie->type < BLOCK_MAX_IOTYPE);
+
+    stats->failed_ops[cookie->type]++;
+
+    if (stats->account_failed) {
+        BlockAcctTimedStats *s;
+        int64_t time_ns = qemu_clock_get_ns(clock_type);
+        int64_t latency_ns = time_ns - cookie->start_time_ns;
+
+        if (qtest_enabled()) {
+            latency_ns = qtest_latency_ns;
+        }
 
-void block_acct_highest_sector(BlockAcctStats *stats, int64_t sector_num,
-                               unsigned int nb_sectors)
+        stats->total_time_ns[cookie->type] += latency_ns;
+        stats->last_access_time_ns = time_ns;
+
+        QSLIST_FOREACH(s, &stats->intervals, entries) {
+            timed_average_account(&s->latency[cookie->type], latency_ns);
+        }
+    }
+}
+
+void block_acct_invalid(BlockAcctStats *stats, enum BlockAcctType type)
 {
-    if (stats->wr_highest_sector < sector_num + nb_sectors - 1) {
-        stats->wr_highest_sector = sector_num + nb_sectors - 1;
+    assert(type < BLOCK_MAX_IOTYPE);
+
+    /* block_acct_done() and block_acct_failed() update
+     * total_time_ns[], but this one does not. The reason is that
+     * invalid requests are accounted during their submission,
+     * therefore there's no actual I/O involved. */
+
+    stats->invalid_ops[type]++;
+
+    if (stats->account_invalid) {
+        stats->last_access_time_ns = qemu_clock_get_ns(clock_type);
     }
 }
 
@@ -61,3 +153,20 @@ void block_acct_merge_done(BlockAcctStats *stats, enum BlockAcctType type,
     assert(type < BLOCK_MAX_IOTYPE);
     stats->merged[type] += num_requests;
 }
+
+int64_t block_acct_idle_time_ns(BlockAcctStats *stats)
+{
+    return qemu_clock_get_ns(clock_type) - stats->last_access_time_ns;
+}
+
+double block_acct_queue_depth(BlockAcctTimedStats *stats,
+                              enum BlockAcctType type)
+{
+    uint64_t sum, elapsed;
+
+    assert(type < BLOCK_MAX_IOTYPE);
+
+    sum = timed_average_sum(&stats->latency[type], &elapsed);
+
+    return (double) sum / elapsed;
+}
diff --git a/block/backup.c b/block/backup.c
index 965654d52..705bb7766 100644
--- a/block/backup.c
+++ b/block/backup.c
@@ -21,6 +21,7 @@
 #include "block/blockjob.h"
 #include "qapi/qmp/qerror.h"
 #include "qemu/ratelimit.h"
+#include "sysemu/block-backend.h"
 
 #define BACKUP_CLUSTER_BITS 16
 #define BACKUP_CLUSTER_SIZE (1 << BACKUP_CLUSTER_BITS)
@@ -89,7 +90,8 @@ static void cow_request_end(CowRequest *req)
 
 static int coroutine_fn backup_do_cow(BlockDriverState *bs,
                                       int64_t sector_num, int nb_sectors,
-                                      bool *error_is_read)
+                                      bool *error_is_read,
+                                      bool is_write_notifier)
 {
     BackupBlockJob *job = (BackupBlockJob *)bs->job;
     CowRequest cow_request;
@@ -129,8 +131,14 @@ static int coroutine_fn backup_do_cow(BlockDriverState *bs,
         iov.iov_len = n * BDRV_SECTOR_SIZE;
         qemu_iovec_init_external(&bounce_qiov, &iov, 1);
 
-        ret = bdrv_co_readv(bs, start * BACKUP_SECTORS_PER_CLUSTER, n,
-                            &bounce_qiov);
+        if (is_write_notifier) {
+            ret = bdrv_co_readv_no_serialising(bs,
+                                           start * BACKUP_SECTORS_PER_CLUSTER,
+                                           n, &bounce_qiov);
+        } else {
+            ret = bdrv_co_readv(bs, start * BACKUP_SECTORS_PER_CLUSTER, n,
+                                &bounce_qiov);
+        }
         if (ret < 0) {
             trace_backup_do_cow_read_fail(job, start, ret);
             if (error_is_read) {
@@ -190,7 +198,7 @@ static int coroutine_fn backup_before_write_notify(
     assert((req->offset & (BDRV_SECTOR_SIZE - 1)) == 0);
     assert((req->bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
 
-    return backup_do_cow(req->bs, sector_num, nb_sectors, NULL);
+    return backup_do_cow(req->bs, sector_num, nb_sectors, NULL, true);
 }
 
 static void backup_set_speed(BlockJob *job, int64_t speed, Error **errp)
@@ -208,7 +216,41 @@ static void backup_iostatus_reset(BlockJob *job)
 {
     BackupBlockJob *s = container_of(job, BackupBlockJob, common);
 
-    bdrv_iostatus_reset(s->target);
+    if (s->target->blk) {
+        blk_iostatus_reset(s->target->blk);
+    }
+}
+
+static void backup_cleanup_sync_bitmap(BackupBlockJob *job, int ret)
+{
+    BdrvDirtyBitmap *bm;
+    BlockDriverState *bs = job->common.bs;
+
+    if (ret < 0 || block_job_is_cancelled(&job->common)) {
+        /* Merge the successor back into the parent, delete nothing. */
+        bm = bdrv_reclaim_dirty_bitmap(bs, job->sync_bitmap, NULL);
+        assert(bm);
+    } else {
+        /* Everything is fine, delete this bitmap and install the backup. */
+        bm = bdrv_dirty_bitmap_abdicate(bs, job->sync_bitmap, NULL);
+        assert(bm);
+    }
+}
+
+static void backup_commit(BlockJob *job)
+{
+    BackupBlockJob *s = container_of(job, BackupBlockJob, common);
+    if (s->sync_bitmap) {
+        backup_cleanup_sync_bitmap(s, 0);
+    }
+}
+
+static void backup_abort(BlockJob *job)
+{
+    BackupBlockJob *s = container_of(job, BackupBlockJob, common);
+    if (s->sync_bitmap) {
+        backup_cleanup_sync_bitmap(s, -1);
+    }
 }
 
 static const BlockJobDriver backup_job_driver = {
@@ -216,6 +258,8 @@ static const BlockJobDriver backup_job_driver = {
     .job_type       = BLOCK_JOB_TYPE_BACKUP,
     .set_speed      = backup_set_speed,
     .iostatus_reset = backup_iostatus_reset,
+    .commit         = backup_commit,
+    .abort          = backup_abort,
 };
 
 static BlockErrorAction backup_error_action(BackupBlockJob *job,
@@ -303,7 +347,8 @@ static int coroutine_fn backup_run_incremental(BackupBlockJob *job)
                     return ret;
                 }
                 ret = backup_do_cow(bs, cluster * BACKUP_SECTORS_PER_CLUSTER,
-                                    BACKUP_SECTORS_PER_CLUSTER, &error_is_read);
+                                    BACKUP_SECTORS_PER_CLUSTER, &error_is_read,
+                                    false);
                 if ((ret < 0) &&
                     backup_error_action(job, error_is_read, -ret) ==
                     BLOCK_ERROR_ACTION_REPORT) {
@@ -352,8 +397,10 @@ static void coroutine_fn backup_run(void *opaque)
     job->bitmap = hbitmap_alloc(end, 0);
 
     bdrv_set_enable_write_cache(target, true);
-    bdrv_set_on_error(target, on_target_error, on_target_error);
-    bdrv_iostatus_enable(target);
+    if (target->blk) {
+        blk_set_on_error(target->blk, on_target_error, on_target_error);
+        blk_iostatus_enable(target->blk);
+    }
 
     bdrv_add_before_write_notifier(bs, &before_write);
 
@@ -408,7 +455,7 @@ static void coroutine_fn backup_run(void *opaque)
             }
             /* FULL sync mode we copy the whole drive. */
             ret = backup_do_cow(bs, start * BACKUP_SECTORS_PER_CLUSTER,
-                    BACKUP_SECTORS_PER_CLUSTER, &error_is_read);
+                    BACKUP_SECTORS_PER_CLUSTER, &error_is_read, false);
             if (ret < 0) {
                 /* Depending on error action, fail now or retry cluster */
                 BlockErrorAction action =
@@ -428,22 +475,11 @@ static void coroutine_fn backup_run(void *opaque)
     /* wait until pending backup_do_cow() calls have completed */
     qemu_co_rwlock_wrlock(&job->flush_rwlock);
     qemu_co_rwlock_unlock(&job->flush_rwlock);
-
-    if (job->sync_bitmap) {
-        BdrvDirtyBitmap *bm;
-        if (ret < 0 || block_job_is_cancelled(&job->common)) {
-            /* Merge the successor back into the parent, delete nothing. */
-            bm = bdrv_reclaim_dirty_bitmap(bs, job->sync_bitmap, NULL);
-            assert(bm);
-        } else {
-            /* Everything is fine, delete this bitmap and install the backup. */
-            bm = bdrv_dirty_bitmap_abdicate(bs, job->sync_bitmap, NULL);
-            assert(bm);
-        }
-    }
     hbitmap_free(job->bitmap);
 
-    bdrv_iostatus_disable(target);
+    if (target->blk) {
+        blk_iostatus_disable(target->blk);
+    }
     bdrv_op_unblock_all(target, job->common.blocker);
 
     data = g_malloc(sizeof(*data));
@@ -457,7 +493,7 @@ void backup_start(BlockDriverState *bs, BlockDriverState *target,
                   BlockdevOnError on_source_error,
                   BlockdevOnError on_target_error,
                   BlockCompletionFunc *cb, void *opaque,
-                  Error **errp)
+                  BlockJobTxn *txn, Error **errp)
 {
     int64_t len;
 
@@ -472,7 +508,7 @@ void backup_start(BlockDriverState *bs, BlockDriverState *target,
 
     if ((on_source_error == BLOCKDEV_ON_ERROR_STOP ||
          on_source_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
-        !bdrv_iostatus_is_enabled(bs)) {
+        (!bs->blk || !blk_iostatus_is_enabled(bs->blk))) {
         error_setg(errp, QERR_INVALID_PARAMETER, "on-source-error");
         return;
     }
@@ -539,6 +575,7 @@ void backup_start(BlockDriverState *bs, BlockDriverState *target,
                        sync_bitmap : NULL;
     job->common.len = len;
     job->common.co = qemu_coroutine_create(backup_run);
+    block_job_txn_add_job(txn, &job->common);
     qemu_coroutine_enter(job->common.co, job);
     return;
 
diff --git a/block/blkdebug.c b/block/blkdebug.c
index bc247f46f..dee3a0edf 100644
--- a/block/blkdebug.c
+++ b/block/blkdebug.c
@@ -30,6 +30,7 @@
 #include "qapi/qmp/qdict.h"
 #include "qapi/qmp/qint.h"
 #include "qapi/qmp/qstring.h"
+#include "sysemu/qtest.h"
 
 typedef struct BDRVBlkdebugState {
     int state;
@@ -426,11 +427,11 @@ static int blkdebug_open(BlockDriverState *bs, QDict *options, int flags,
     /* Set initial state */
     s->state = 1;
 
-    /* Open the backing file */
-    assert(bs->file == NULL);
-    ret = bdrv_open_image(&bs->file, qemu_opt_get(opts, "x-image"), options, "image",
-                          bs, &child_file, false, &local_err);
-    if (ret < 0) {
+    /* Open the image file */
+    bs->file = bdrv_open_child(qemu_opt_get(opts, "x-image"), options, "image",
+                               bs, &child_file, false, &local_err);
+    if (local_err) {
+        ret = -EINVAL;
         error_propagate(errp, local_err);
         goto out;
     }
@@ -449,7 +450,7 @@ static int blkdebug_open(BlockDriverState *bs, QDict *options, int flags,
     goto out;
 
 fail_unref:
-    bdrv_unref(bs->file);
+    bdrv_unref_child(bs, bs->file);
 out:
     qemu_opts_del(opts);
     return ret;
@@ -510,7 +511,8 @@ static BlockAIOCB *blkdebug_aio_readv(BlockDriverState *bs,
         return inject_error(bs, cb, opaque, rule);
     }
 
-    return bdrv_aio_readv(bs->file, sector_num, qiov, nb_sectors, cb, opaque);
+    return bdrv_aio_readv(bs->file->bs, sector_num, qiov, nb_sectors,
+                          cb, opaque);
 }
 
 static BlockAIOCB *blkdebug_aio_writev(BlockDriverState *bs,
@@ -532,7 +534,8 @@ static BlockAIOCB *blkdebug_aio_writev(BlockDriverState *bs,
         return inject_error(bs, cb, opaque, rule);
     }
 
-    return bdrv_aio_writev(bs->file, sector_num, qiov, nb_sectors, cb, opaque);
+    return bdrv_aio_writev(bs->file->bs, sector_num, qiov, nb_sectors,
+                           cb, opaque);
 }
 
 static BlockAIOCB *blkdebug_aio_flush(BlockDriverState *bs,
@@ -551,7 +554,7 @@ static BlockAIOCB *blkdebug_aio_flush(BlockDriverState *bs,
         return inject_error(bs, cb, opaque, rule);
     }
 
-    return bdrv_aio_flush(bs->file, cb, opaque);
+    return bdrv_aio_flush(bs->file->bs, cb, opaque);
 }
 
 
@@ -581,9 +584,13 @@ static void suspend_request(BlockDriverState *bs, BlkdebugRule *rule)
     remove_rule(rule);
     QLIST_INSERT_HEAD(&s->suspended_reqs, &r, next);
 
-    printf("blkdebug: Suspended request '%s'\n", r.tag);
+    if (!qtest_enabled()) {
+        printf("blkdebug: Suspended request '%s'\n", r.tag);
+    }
     qemu_coroutine_yield();
-    printf("blkdebug: Resuming request '%s'\n", r.tag);
+    if (!qtest_enabled()) {
+        printf("blkdebug: Resuming request '%s'\n", r.tag);
+    }
 
     QLIST_REMOVE(&r, next);
     g_free(r.tag);
@@ -716,12 +723,12 @@ static bool blkdebug_debug_is_suspended(BlockDriverState *bs, const char *tag)
 
 static int64_t blkdebug_getlength(BlockDriverState *bs)
 {
-    return bdrv_getlength(bs->file);
+    return bdrv_getlength(bs->file->bs);
 }
 
 static int blkdebug_truncate(BlockDriverState *bs, int64_t offset)
 {
-    return bdrv_truncate(bs->file, offset);
+    return bdrv_truncate(bs->file->bs, offset);
 }
 
 static void blkdebug_refresh_filename(BlockDriverState *bs)
@@ -741,24 +748,24 @@ static void blkdebug_refresh_filename(BlockDriverState *bs)
         }
     }
 
-    if (force_json && !bs->file->full_open_options) {
+    if (force_json && !bs->file->bs->full_open_options) {
         /* The config file cannot be recreated, so creating a plain filename
          * is impossible */
         return;
     }
 
-    if (!force_json && bs->file->exact_filename[0]) {
+    if (!force_json && bs->file->bs->exact_filename[0]) {
         snprintf(bs->exact_filename, sizeof(bs->exact_filename),
                  "blkdebug:%s:%s",
                  qdict_get_try_str(bs->options, "config") ?: "",
-                 bs->file->exact_filename);
+                 bs->file->bs->exact_filename);
     }
 
     opts = qdict_new();
     qdict_put_obj(opts, "driver", QOBJECT(qstring_from_str("blkdebug")));
 
-    QINCREF(bs->file->full_open_options);
-    qdict_put_obj(opts, "image", QOBJECT(bs->file->full_open_options));
+    QINCREF(bs->file->bs->full_open_options);
+    qdict_put_obj(opts, "image", QOBJECT(bs->file->bs->full_open_options));
 
     for (e = qdict_first(bs->options); e; e = qdict_next(bs->options, e)) {
         if (strcmp(qdict_entry_key(e), "x-image") &&
diff --git a/block/blkverify.c b/block/blkverify.c
index d277e6322..c5f8e8dcb 100644
--- a/block/blkverify.c
+++ b/block/blkverify.c
@@ -14,7 +14,7 @@
 #include "qapi/qmp/qstring.h"
 
 typedef struct {
-    BlockDriverState *test_file;
+    BdrvChild *test_file;
 } BDRVBlkverifyState;
 
 typedef struct BlkverifyAIOCB BlkverifyAIOCB;
@@ -123,26 +123,29 @@ static int blkverify_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     /* Open the raw file */
-    assert(bs->file == NULL);
-    ret = bdrv_open_image(&bs->file, qemu_opt_get(opts, "x-raw"), options,
-                          "raw", bs, &child_file, false, &local_err);
-    if (ret < 0) {
+    bs->file = bdrv_open_child(qemu_opt_get(opts, "x-raw"), options, "raw",
+                               bs, &child_file, false, &local_err);
+    if (local_err) {
+        ret = -EINVAL;
         error_propagate(errp, local_err);
         goto fail;
     }
 
     /* Open the test file */
-    assert(s->test_file == NULL);
-    ret = bdrv_open_image(&s->test_file, qemu_opt_get(opts, "x-image"), options,
-                          "test", bs, &child_format, false, &local_err);
-    if (ret < 0) {
+    s->test_file = bdrv_open_child(qemu_opt_get(opts, "x-image"), options,
+                                   "test", bs, &child_format, false,
+                                   &local_err);
+    if (local_err) {
+        ret = -EINVAL;
         error_propagate(errp, local_err);
-        s->test_file = NULL;
         goto fail;
     }
 
     ret = 0;
 fail:
+    if (ret < 0) {
+        bdrv_unref_child(bs, bs->file);
+    }
     qemu_opts_del(opts);
     return ret;
 }
@@ -151,7 +154,7 @@ static void blkverify_close(BlockDriverState *bs)
 {
     BDRVBlkverifyState *s = bs->opaque;
 
-    bdrv_unref(s->test_file);
+    bdrv_unref_child(bs, s->test_file);
     s->test_file = NULL;
 }
 
@@ -159,7 +162,7 @@ static int64_t blkverify_getlength(BlockDriverState *bs)
 {
     BDRVBlkverifyState *s = bs->opaque;
 
-    return bdrv_getlength(s->test_file);
+    return bdrv_getlength(s->test_file->bs);
 }
 
 static BlkverifyAIOCB *blkverify_aio_get(BlockDriverState *bs, bool is_write,
@@ -238,13 +241,13 @@ static BlockAIOCB *blkverify_aio_readv(BlockDriverState *bs,
                                             nb_sectors, cb, opaque);
 
     acb->verify = blkverify_verify_readv;
-    acb->buf = qemu_blockalign(bs->file, qiov->size);
+    acb->buf = qemu_blockalign(bs->file->bs, qiov->size);
     qemu_iovec_init(&acb->raw_qiov, acb->qiov->niov);
     qemu_iovec_clone(&acb->raw_qiov, qiov, acb->buf);
 
-    bdrv_aio_readv(s->test_file, sector_num, qiov, nb_sectors,
+    bdrv_aio_readv(s->test_file->bs, sector_num, qiov, nb_sectors,
                    blkverify_aio_cb, acb);
-    bdrv_aio_readv(bs->file, sector_num, &acb->raw_qiov, nb_sectors,
+    bdrv_aio_readv(bs->file->bs, sector_num, &acb->raw_qiov, nb_sectors,
                    blkverify_aio_cb, acb);
     return &acb->common;
 }
@@ -257,9 +260,9 @@ static BlockAIOCB *blkverify_aio_writev(BlockDriverState *bs,
     BlkverifyAIOCB *acb = blkverify_aio_get(bs, true, sector_num, qiov,
                                             nb_sectors, cb, opaque);
 
-    bdrv_aio_writev(s->test_file, sector_num, qiov, nb_sectors,
+    bdrv_aio_writev(s->test_file->bs, sector_num, qiov, nb_sectors,
                     blkverify_aio_cb, acb);
-    bdrv_aio_writev(bs->file, sector_num, qiov, nb_sectors,
+    bdrv_aio_writev(bs->file->bs, sector_num, qiov, nb_sectors,
                     blkverify_aio_cb, acb);
     return &acb->common;
 }
@@ -271,7 +274,7 @@ static BlockAIOCB *blkverify_aio_flush(BlockDriverState *bs,
     BDRVBlkverifyState *s = bs->opaque;
 
     /* Only flush test file, the raw file is not important */
-    return bdrv_aio_flush(s->test_file, cb, opaque);
+    return bdrv_aio_flush(s->test_file->bs, cb, opaque);
 }
 
 static bool blkverify_recurse_is_first_non_filter(BlockDriverState *bs,
@@ -279,13 +282,13 @@ static bool blkverify_recurse_is_first_non_filter(BlockDriverState *bs,
 {
     BDRVBlkverifyState *s = bs->opaque;
 
-    bool perm = bdrv_recurse_is_first_non_filter(bs->file, candidate);
+    bool perm = bdrv_recurse_is_first_non_filter(bs->file->bs, candidate);
 
     if (perm) {
         return true;
     }
 
-    return bdrv_recurse_is_first_non_filter(s->test_file, candidate);
+    return bdrv_recurse_is_first_non_filter(s->test_file->bs, candidate);
 }
 
 /* Propagate AioContext changes to ->test_file */
@@ -293,7 +296,7 @@ static void blkverify_detach_aio_context(BlockDriverState *bs)
 {
     BDRVBlkverifyState *s = bs->opaque;
 
-    bdrv_detach_aio_context(s->test_file);
+    bdrv_detach_aio_context(s->test_file->bs);
 }
 
 static void blkverify_attach_aio_context(BlockDriverState *bs,
@@ -301,32 +304,38 @@ static void blkverify_attach_aio_context(BlockDriverState *bs,
 {
     BDRVBlkverifyState *s = bs->opaque;
 
-    bdrv_attach_aio_context(s->test_file, new_context);
+    bdrv_attach_aio_context(s->test_file->bs, new_context);
 }
 
 static void blkverify_refresh_filename(BlockDriverState *bs)
 {
     BDRVBlkverifyState *s = bs->opaque;
 
-    /* bs->file has already been refreshed */
-    bdrv_refresh_filename(s->test_file);
+    /* bs->file->bs has already been refreshed */
+    bdrv_refresh_filename(s->test_file->bs);
 
-    if (bs->file->full_open_options && s->test_file->full_open_options) {
+    if (bs->file->bs->full_open_options
+        && s->test_file->bs->full_open_options)
+    {
         QDict *opts = qdict_new();
         qdict_put_obj(opts, "driver", QOBJECT(qstring_from_str("blkverify")));
 
-        QINCREF(bs->file->full_open_options);
-        qdict_put_obj(opts, "raw", QOBJECT(bs->file->full_open_options));
-        QINCREF(s->test_file->full_open_options);
-        qdict_put_obj(opts, "test", QOBJECT(s->test_file->full_open_options));
+        QINCREF(bs->file->bs->full_open_options);
+        qdict_put_obj(opts, "raw", QOBJECT(bs->file->bs->full_open_options));
+        QINCREF(s->test_file->bs->full_open_options);
+        qdict_put_obj(opts, "test",
+                      QOBJECT(s->test_file->bs->full_open_options));
 
         bs->full_open_options = opts;
     }
 
-    if (bs->file->exact_filename[0] && s->test_file->exact_filename[0]) {
+    if (bs->file->bs->exact_filename[0]
+        && s->test_file->bs->exact_filename[0])
+    {
         snprintf(bs->exact_filename, sizeof(bs->exact_filename),
                  "blkverify:%s:%s",
-                 bs->file->exact_filename, s->test_file->exact_filename);
+                 bs->file->bs->exact_filename,
+                 s->test_file->bs->exact_filename);
     }
 }
 
diff --git a/block/block-backend.c b/block/block-backend.c
index aee8a1202..419591f26 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -12,12 +12,17 @@
 
 #include "sysemu/block-backend.h"
 #include "block/block_int.h"
+#include "block/blockjob.h"
+#include "block/throttle-groups.h"
 #include "sysemu/blockdev.h"
+#include "sysemu/sysemu.h"
 #include "qapi-event.h"
 
 /* Number of coroutines to reserve per attached device model */
 #define COROUTINE_POOL_RESERVATION 64
 
+static AioContext *blk_aiocb_get_aio_context(BlockAIOCB *acb);
+
 struct BlockBackend {
     char *name;
     int refcnt;
@@ -29,15 +34,31 @@ struct BlockBackend {
     /* TODO change to DeviceState when all users are qdevified */
     const BlockDevOps *dev_ops;
     void *dev_opaque;
+
+    /* the block size for which the guest device expects atomicity */
+    int guest_block_size;
+
+    /* If the BDS tree is removed, some of its options are stored here (which
+     * can be used to restore those options in the new BDS on insert) */
+    BlockBackendRootState root_state;
+
+    /* I/O stats (display with "info blockstats"). */
+    BlockAcctStats stats;
+
+    BlockdevOnError on_read_error, on_write_error;
+    bool iostatus_enabled;
+    BlockDeviceIoStatus iostatus;
 };
 
 typedef struct BlockBackendAIOCB {
     BlockAIOCB common;
     QEMUBH *bh;
+    BlockBackend *blk;
     int ret;
 } BlockBackendAIOCB;
 
 static const AIOCBInfo block_backend_aiocb_info = {
+    .get_aio_context = blk_aiocb_get_aio_context,
     .aiocb_size = sizeof(BlockBackendAIOCB),
 };
 
@@ -126,7 +147,7 @@ BlockBackend *blk_new_open(const char *name, const char *filename,
         return NULL;
     }
 
-    ret = bdrv_open(&blk->bs, filename, reference, options, flags, NULL, errp);
+    ret = bdrv_open(&blk->bs, filename, reference, options, flags, errp);
     if (ret < 0) {
         blk_unref(blk);
         return NULL;
@@ -145,12 +166,17 @@ static void blk_delete(BlockBackend *blk)
         bdrv_unref(blk->bs);
         blk->bs = NULL;
     }
+    if (blk->root_state.throttle_state) {
+        g_free(blk->root_state.throttle_group);
+        throttle_group_unref(blk->root_state.throttle_state);
+    }
     /* Avoid double-remove after blk_hide_on_behalf_of_hmp_drive_del() */
     if (blk->name[0]) {
         QTAILQ_REMOVE(&blk_backends, blk, link);
     }
     g_free(blk->name);
     drive_info_del(blk->legacy_dinfo);
+    block_acct_cleanup(&blk->stats);
     g_free(blk);
 }
 
@@ -164,6 +190,11 @@ static void drive_info_del(DriveInfo *dinfo)
     g_free(dinfo);
 }
 
+int blk_get_refcnt(BlockBackend *blk)
+{
+    return blk ? blk->refcnt : 0;
+}
+
 /*
  * Increment @blk's reference count.
  * @blk must not be null.
@@ -239,6 +270,23 @@ BlockDriverState *blk_bs(BlockBackend *blk)
 }
 
 /*
+ * Changes the BlockDriverState attached to @blk
+ */
+void blk_set_bs(BlockBackend *blk, BlockDriverState *bs)
+{
+    bdrv_ref(bs);
+
+    if (blk->bs) {
+        blk->bs->blk = NULL;
+        bdrv_unref(blk->bs);
+    }
+    assert(bs->blk == NULL);
+
+    blk->bs = bs;
+    bs->blk = blk;
+}
+
+/*
  * Return @blk's DriveInfo if any, else null.
  */
 DriveInfo *blk_legacy_dinfo(BlockBackend *blk)
@@ -292,6 +340,29 @@ void blk_hide_on_behalf_of_hmp_drive_del(BlockBackend *blk)
 }
 
 /*
+ * Disassociates the currently associated BlockDriverState from @blk.
+ */
+void blk_remove_bs(BlockBackend *blk)
+{
+    blk_update_root_state(blk);
+
+    blk->bs->blk = NULL;
+    bdrv_unref(blk->bs);
+    blk->bs = NULL;
+}
+
+/*
+ * Associates a new BlockDriverState with @blk.
+ */
+void blk_insert_bs(BlockBackend *blk, BlockDriverState *bs)
+{
+    assert(!blk->bs && !bs->blk);
+    bdrv_ref(bs);
+    blk->bs = bs;
+    bs->blk = blk;
+}
+
+/*
  * Attach device model @dev to @blk.
  * Return 0 on success, -EBUSY when a device model is attached already.
  */
@@ -303,7 +374,7 @@ int blk_attach_dev(BlockBackend *blk, void *dev)
     }
     blk_ref(blk);
     blk->dev = dev;
-    bdrv_iostatus_reset(blk->bs);
+    blk_iostatus_reset(blk);
     return 0;
 }
 
@@ -330,7 +401,7 @@ void blk_detach_dev(BlockBackend *blk, void *dev)
     blk->dev = NULL;
     blk->dev_ops = NULL;
     blk->dev_opaque = NULL;
-    bdrv_set_guest_block_size(blk->bs, 512);
+    blk->guest_block_size = 512;
     blk_unref(blk);
 }
 
@@ -364,18 +435,15 @@ void blk_set_dev_ops(BlockBackend *blk, const BlockDevOps *ops,
 void blk_dev_change_media_cb(BlockBackend *blk, bool load)
 {
     if (blk->dev_ops && blk->dev_ops->change_media_cb) {
-        bool tray_was_closed = !blk_dev_is_tray_open(blk);
+        bool tray_was_open, tray_is_open;
 
+        tray_was_open = blk_dev_is_tray_open(blk);
         blk->dev_ops->change_media_cb(blk->dev_opaque, load);
-        if (tray_was_closed) {
-            /* tray open */
-            qapi_event_send_device_tray_moved(blk_name(blk),
-                                              true, &error_abort);
-        }
-        if (load) {
-            /* tray close */
-            qapi_event_send_device_tray_moved(blk_name(blk),
-                                              false, &error_abort);
+        tray_is_open = blk_dev_is_tray_open(blk);
+
+        if (tray_was_open != tray_is_open) {
+            qapi_event_send_device_tray_moved(blk_name(blk), tray_is_open,
+                                              &error_abort);
         }
     }
 }
@@ -390,6 +458,14 @@ bool blk_dev_has_removable_media(BlockBackend *blk)
 }
 
 /*
+ * Does @blk's attached device model have a tray?
+ */
+bool blk_dev_has_tray(BlockBackend *blk)
+{
+    return blk->dev_ops && blk->dev_ops->is_tray_open;
+}
+
+/*
  * Notify @blk's attached device model of a media eject request.
  * If @force is true, the medium is about to be yanked out forcefully.
  */
@@ -405,7 +481,7 @@ void blk_dev_eject_request(BlockBackend *blk, bool force)
  */
 bool blk_dev_is_tray_open(BlockBackend *blk)
 {
-    if (blk->dev_ops && blk->dev_ops->is_tray_open) {
+    if (blk_dev_has_tray(blk)) {
         return blk->dev_ops->is_tray_open(blk->dev_opaque);
     }
     return false;
@@ -435,7 +511,47 @@ void blk_dev_resize_cb(BlockBackend *blk)
 
 void blk_iostatus_enable(BlockBackend *blk)
 {
-    bdrv_iostatus_enable(blk->bs);
+    blk->iostatus_enabled = true;
+    blk->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 blk_iostatus_is_enabled(const BlockBackend *blk)
+{
+    return (blk->iostatus_enabled &&
+           (blk->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
+            blk->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
+            blk->on_read_error == BLOCKDEV_ON_ERROR_STOP));
+}
+
+BlockDeviceIoStatus blk_iostatus(const BlockBackend *blk)
+{
+    return blk->iostatus;
+}
+
+void blk_iostatus_disable(BlockBackend *blk)
+{
+    blk->iostatus_enabled = false;
+}
+
+void blk_iostatus_reset(BlockBackend *blk)
+{
+    if (blk_iostatus_is_enabled(blk)) {
+        blk->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
+        if (blk->bs && blk->bs->job) {
+            block_job_iostatus_reset(blk->bs->job);
+        }
+    }
+}
+
+void blk_iostatus_set_err(BlockBackend *blk, int error)
+{
+    assert(blk_iostatus_is_enabled(blk));
+    if (blk->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
+        blk->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
+                                          BLOCK_DEVICE_IO_STATUS_FAILED;
+    }
 }
 
 static int blk_check_byte_request(BlockBackend *blk, int64_t offset,
@@ -447,7 +563,7 @@ static int blk_check_byte_request(BlockBackend *blk, int64_t offset,
         return -EIO;
     }
 
-    if (!blk_is_inserted(blk)) {
+    if (!blk_is_available(blk)) {
         return -ENOMEDIUM;
     }
 
@@ -534,13 +650,15 @@ static void error_callback_bh(void *opaque)
     qemu_aio_unref(acb);
 }
 
-static BlockAIOCB *abort_aio_request(BlockBackend *blk, BlockCompletionFunc *cb,
-                                     void *opaque, int ret)
+BlockAIOCB *blk_abort_aio_request(BlockBackend *blk,
+                                  BlockCompletionFunc *cb,
+                                  void *opaque, int ret)
 {
     struct BlockBackendAIOCB *acb;
     QEMUBH *bh;
 
     acb = blk_aio_get(&block_backend_aiocb_info, blk, cb, opaque);
+    acb->blk = blk;
     acb->ret = ret;
 
     bh = aio_bh_new(blk_get_aio_context(blk), error_callback_bh, acb);
@@ -556,7 +674,7 @@ BlockAIOCB *blk_aio_write_zeroes(BlockBackend *blk, int64_t sector_num,
 {
     int ret = blk_check_request(blk, sector_num, nb_sectors);
     if (ret < 0) {
-        return abort_aio_request(blk, cb, opaque, ret);
+        return blk_abort_aio_request(blk, cb, opaque, ret);
     }
 
     return bdrv_aio_write_zeroes(blk->bs, sector_num, nb_sectors, flags,
@@ -585,16 +703,28 @@ int blk_pwrite(BlockBackend *blk, int64_t offset, const void *buf, int count)
 
 int64_t blk_getlength(BlockBackend *blk)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_getlength(blk->bs);
 }
 
 void blk_get_geometry(BlockBackend *blk, uint64_t *nb_sectors_ptr)
 {
-    bdrv_get_geometry(blk->bs, nb_sectors_ptr);
+    if (!blk->bs) {
+        *nb_sectors_ptr = 0;
+    } else {
+        bdrv_get_geometry(blk->bs, nb_sectors_ptr);
+    }
 }
 
 int64_t blk_nb_sectors(BlockBackend *blk)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_nb_sectors(blk->bs);
 }
 
@@ -604,7 +734,7 @@ BlockAIOCB *blk_aio_readv(BlockBackend *blk, int64_t sector_num,
 {
     int ret = blk_check_request(blk, sector_num, nb_sectors);
     if (ret < 0) {
-        return abort_aio_request(blk, cb, opaque, ret);
+        return blk_abort_aio_request(blk, cb, opaque, ret);
     }
 
     return bdrv_aio_readv(blk->bs, sector_num, iov, nb_sectors, cb, opaque);
@@ -616,7 +746,7 @@ BlockAIOCB *blk_aio_writev(BlockBackend *blk, int64_t sector_num,
 {
     int ret = blk_check_request(blk, sector_num, nb_sectors);
     if (ret < 0) {
-        return abort_aio_request(blk, cb, opaque, ret);
+        return blk_abort_aio_request(blk, cb, opaque, ret);
     }
 
     return bdrv_aio_writev(blk->bs, sector_num, iov, nb_sectors, cb, opaque);
@@ -625,6 +755,10 @@ BlockAIOCB *blk_aio_writev(BlockBackend *blk, int64_t sector_num,
 BlockAIOCB *blk_aio_flush(BlockBackend *blk,
                           BlockCompletionFunc *cb, void *opaque)
 {
+    if (!blk_is_available(blk)) {
+        return blk_abort_aio_request(blk, cb, opaque, -ENOMEDIUM);
+    }
+
     return bdrv_aio_flush(blk->bs, cb, opaque);
 }
 
@@ -634,7 +768,7 @@ BlockAIOCB *blk_aio_discard(BlockBackend *blk,
 {
     int ret = blk_check_request(blk, sector_num, nb_sectors);
     if (ret < 0) {
-        return abort_aio_request(blk, cb, opaque, ret);
+        return blk_abort_aio_request(blk, cb, opaque, ret);
     }
 
     return bdrv_aio_discard(blk->bs, sector_num, nb_sectors, cb, opaque);
@@ -666,12 +800,20 @@ int blk_aio_multiwrite(BlockBackend *blk, BlockRequest *reqs, int num_reqs)
 
 int blk_ioctl(BlockBackend *blk, unsigned long int req, void *buf)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_ioctl(blk->bs, req, buf);
 }
 
 BlockAIOCB *blk_aio_ioctl(BlockBackend *blk, unsigned long int req, void *buf,
                           BlockCompletionFunc *cb, void *opaque)
 {
+    if (!blk_is_available(blk)) {
+        return blk_abort_aio_request(blk, cb, opaque, -ENOMEDIUM);
+    }
+
     return bdrv_aio_ioctl(blk->bs, req, buf, cb, opaque);
 }
 
@@ -687,11 +829,19 @@ int blk_co_discard(BlockBackend *blk, int64_t sector_num, int nb_sectors)
 
 int blk_co_flush(BlockBackend *blk)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_co_flush(blk->bs);
 }
 
 int blk_flush(BlockBackend *blk)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_flush(blk->bs);
 }
 
@@ -702,7 +852,9 @@ int blk_flush_all(void)
 
 void blk_drain(BlockBackend *blk)
 {
-    bdrv_drain(blk->bs);
+    if (blk->bs) {
+        bdrv_drain(blk->bs);
+    }
 }
 
 void blk_drain_all(void)
@@ -710,76 +862,178 @@ void blk_drain_all(void)
     bdrv_drain_all();
 }
 
+void blk_set_on_error(BlockBackend *blk, BlockdevOnError on_read_error,
+                      BlockdevOnError on_write_error)
+{
+    blk->on_read_error = on_read_error;
+    blk->on_write_error = on_write_error;
+}
+
 BlockdevOnError blk_get_on_error(BlockBackend *blk, bool is_read)
 {
-    return bdrv_get_on_error(blk->bs, is_read);
+    return is_read ? blk->on_read_error : blk->on_write_error;
 }
 
 BlockErrorAction blk_get_error_action(BlockBackend *blk, bool is_read,
                                       int error)
 {
-    return bdrv_get_error_action(blk->bs, is_read, error);
+    BlockdevOnError on_err = blk_get_on_error(blk, is_read);
+
+    switch (on_err) {
+    case BLOCKDEV_ON_ERROR_ENOSPC:
+        return (error == ENOSPC) ?
+               BLOCK_ERROR_ACTION_STOP : BLOCK_ERROR_ACTION_REPORT;
+    case BLOCKDEV_ON_ERROR_STOP:
+        return BLOCK_ERROR_ACTION_STOP;
+    case BLOCKDEV_ON_ERROR_REPORT:
+        return BLOCK_ERROR_ACTION_REPORT;
+    case BLOCKDEV_ON_ERROR_IGNORE:
+        return BLOCK_ERROR_ACTION_IGNORE;
+    default:
+        abort();
+    }
+}
+
+static void send_qmp_error_event(BlockBackend *blk,
+                                 BlockErrorAction action,
+                                 bool is_read, int error)
+{
+    IoOperationType optype;
+
+    optype = is_read ? IO_OPERATION_TYPE_READ : IO_OPERATION_TYPE_WRITE;
+    qapi_event_send_block_io_error(blk_name(blk), optype, action,
+                                   blk_iostatus_is_enabled(blk),
+                                   error == ENOSPC, strerror(error),
+                                   &error_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 blk_error_action(BlockBackend *blk, BlockErrorAction action,
                       bool is_read, int error)
 {
-    bdrv_error_action(blk->bs, action, is_read, error);
+    assert(error >= 0);
+
+    if (action == BLOCK_ERROR_ACTION_STOP) {
+        /* First set the iostatus, so that "info block" returns an iostatus
+         * that matches the events raised so far (an additional error iostatus
+         * is fine, but not a lost one).
+         */
+        blk_iostatus_set_err(blk, error);
+
+        /* Then raise the request to stop the VM and the event.
+         * qemu_system_vmstop_request_prepare has two effects.  First,
+         * it ensures that the STOP event always comes after the
+         * BLOCK_IO_ERROR event.  Second, it ensures that even if management
+         * can observe the STOP event and do a "cont" before the STOP
+         * event is issued, the VM will not stop.  In this case, vm_start()
+         * also ensures that the STOP/RESUME pair of events is emitted.
+         */
+        qemu_system_vmstop_request_prepare();
+        send_qmp_error_event(blk, action, is_read, error);
+        qemu_system_vmstop_request(RUN_STATE_IO_ERROR);
+    } else {
+        send_qmp_error_event(blk, action, is_read, error);
+    }
 }
 
 int blk_is_read_only(BlockBackend *blk)
 {
-    return bdrv_is_read_only(blk->bs);
+    if (blk->bs) {
+        return bdrv_is_read_only(blk->bs);
+    } else {
+        return blk->root_state.read_only;
+    }
 }
 
 int blk_is_sg(BlockBackend *blk)
 {
+    if (!blk->bs) {
+        return 0;
+    }
+
     return bdrv_is_sg(blk->bs);
 }
 
 int blk_enable_write_cache(BlockBackend *blk)
 {
-    return bdrv_enable_write_cache(blk->bs);
+    if (blk->bs) {
+        return bdrv_enable_write_cache(blk->bs);
+    } else {
+        return !!(blk->root_state.open_flags & BDRV_O_CACHE_WB);
+    }
 }
 
 void blk_set_enable_write_cache(BlockBackend *blk, bool wce)
 {
-    bdrv_set_enable_write_cache(blk->bs, wce);
+    if (blk->bs) {
+        bdrv_set_enable_write_cache(blk->bs, wce);
+    } else {
+        if (wce) {
+            blk->root_state.open_flags |= BDRV_O_CACHE_WB;
+        } else {
+            blk->root_state.open_flags &= ~BDRV_O_CACHE_WB;
+        }
+    }
 }
 
 void blk_invalidate_cache(BlockBackend *blk, Error **errp)
 {
+    if (!blk->bs) {
+        error_setg(errp, "Device '%s' has no medium", blk->name);
+        return;
+    }
+
     bdrv_invalidate_cache(blk->bs, errp);
 }
 
-int blk_is_inserted(BlockBackend *blk)
+bool blk_is_inserted(BlockBackend *blk)
+{
+    return blk->bs && bdrv_is_inserted(blk->bs);
+}
+
+bool blk_is_available(BlockBackend *blk)
 {
-    return bdrv_is_inserted(blk->bs);
+    return blk_is_inserted(blk) && !blk_dev_is_tray_open(blk);
 }
 
 void blk_lock_medium(BlockBackend *blk, bool locked)
 {
-    bdrv_lock_medium(blk->bs, locked);
+    if (blk->bs) {
+        bdrv_lock_medium(blk->bs, locked);
+    }
 }
 
 void blk_eject(BlockBackend *blk, bool eject_flag)
 {
-    bdrv_eject(blk->bs, eject_flag);
+    if (blk->bs) {
+        bdrv_eject(blk->bs, eject_flag);
+    }
 }
 
 int blk_get_flags(BlockBackend *blk)
 {
-    return bdrv_get_flags(blk->bs);
+    if (blk->bs) {
+        return bdrv_get_flags(blk->bs);
+    } else {
+        return blk->root_state.open_flags;
+    }
 }
 
 int blk_get_max_transfer_length(BlockBackend *blk)
 {
-    return blk->bs->bl.max_transfer_length;
+    if (blk->bs) {
+        return blk->bs->bl.max_transfer_length;
+    } else {
+        return 0;
+    }
 }
 
 void blk_set_guest_block_size(BlockBackend *blk, int align)
 {
-    bdrv_set_guest_block_size(blk->bs, align);
+    blk->guest_block_size = align;
 }
 
 void *blk_blockalign(BlockBackend *blk, size_t size)
@@ -789,40 +1043,64 @@ void *blk_blockalign(BlockBackend *blk, size_t size)
 
 bool blk_op_is_blocked(BlockBackend *blk, BlockOpType op, Error **errp)
 {
+    if (!blk->bs) {
+        return false;
+    }
+
     return bdrv_op_is_blocked(blk->bs, op, errp);
 }
 
 void blk_op_unblock(BlockBackend *blk, BlockOpType op, Error *reason)
 {
-    bdrv_op_unblock(blk->bs, op, reason);
+    if (blk->bs) {
+        bdrv_op_unblock(blk->bs, op, reason);
+    }
 }
 
 void blk_op_block_all(BlockBackend *blk, Error *reason)
 {
-    bdrv_op_block_all(blk->bs, reason);
+    if (blk->bs) {
+        bdrv_op_block_all(blk->bs, reason);
+    }
 }
 
 void blk_op_unblock_all(BlockBackend *blk, Error *reason)
 {
-    bdrv_op_unblock_all(blk->bs, reason);
+    if (blk->bs) {
+        bdrv_op_unblock_all(blk->bs, reason);
+    }
 }
 
 AioContext *blk_get_aio_context(BlockBackend *blk)
 {
-    return bdrv_get_aio_context(blk->bs);
+    if (blk->bs) {
+        return bdrv_get_aio_context(blk->bs);
+    } else {
+        return qemu_get_aio_context();
+    }
+}
+
+static AioContext *blk_aiocb_get_aio_context(BlockAIOCB *acb)
+{
+    BlockBackendAIOCB *blk_acb = DO_UPCAST(BlockBackendAIOCB, common, acb);
+    return blk_get_aio_context(blk_acb->blk);
 }
 
 void blk_set_aio_context(BlockBackend *blk, AioContext *new_context)
 {
-    bdrv_set_aio_context(blk->bs, new_context);
+    if (blk->bs) {
+        bdrv_set_aio_context(blk->bs, new_context);
+    }
 }
 
 void blk_add_aio_context_notifier(BlockBackend *blk,
         void (*attached_aio_context)(AioContext *new_context, void *opaque),
         void (*detach_aio_context)(void *opaque), void *opaque)
 {
-    bdrv_add_aio_context_notifier(blk->bs, attached_aio_context,
-                                  detach_aio_context, opaque);
+    if (blk->bs) {
+        bdrv_add_aio_context_notifier(blk->bs, attached_aio_context,
+                                      detach_aio_context, opaque);
+    }
 }
 
 void blk_remove_aio_context_notifier(BlockBackend *blk,
@@ -831,28 +1109,36 @@ void blk_remove_aio_context_notifier(BlockBackend *blk,
                                      void (*detach_aio_context)(void *),
                                      void *opaque)
 {
-    bdrv_remove_aio_context_notifier(blk->bs, attached_aio_context,
-                                     detach_aio_context, opaque);
+    if (blk->bs) {
+        bdrv_remove_aio_context_notifier(blk->bs, attached_aio_context,
+                                         detach_aio_context, opaque);
+    }
 }
 
 void blk_add_close_notifier(BlockBackend *blk, Notifier *notify)
 {
-    bdrv_add_close_notifier(blk->bs, notify);
+    if (blk->bs) {
+        bdrv_add_close_notifier(blk->bs, notify);
+    }
 }
 
 void blk_io_plug(BlockBackend *blk)
 {
-    bdrv_io_plug(blk->bs);
+    if (blk->bs) {
+        bdrv_io_plug(blk->bs);
+    }
 }
 
 void blk_io_unplug(BlockBackend *blk)
 {
-    bdrv_io_unplug(blk->bs);
+    if (blk->bs) {
+        bdrv_io_unplug(blk->bs);
+    }
 }
 
 BlockAcctStats *blk_get_stats(BlockBackend *blk)
 {
-    return bdrv_get_stats(blk->bs);
+    return &blk->stats;
 }
 
 void *blk_aio_get(const AIOCBInfo *aiocb_info, BlockBackend *blk,
@@ -885,6 +1171,10 @@ int blk_write_compressed(BlockBackend *blk, int64_t sector_num,
 
 int blk_truncate(BlockBackend *blk, int64_t offset)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_truncate(blk->bs, offset);
 }
 
@@ -901,20 +1191,94 @@ int blk_discard(BlockBackend *blk, int64_t sector_num, int nb_sectors)
 int blk_save_vmstate(BlockBackend *blk, const uint8_t *buf,
                      int64_t pos, int size)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_save_vmstate(blk->bs, buf, pos, size);
 }
 
 int blk_load_vmstate(BlockBackend *blk, uint8_t *buf, int64_t pos, int size)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_load_vmstate(blk->bs, buf, pos, size);
 }
 
 int blk_probe_blocksizes(BlockBackend *blk, BlockSizes *bsz)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_probe_blocksizes(blk->bs, bsz);
 }
 
 int blk_probe_geometry(BlockBackend *blk, HDGeometry *geo)
 {
+    if (!blk_is_available(blk)) {
+        return -ENOMEDIUM;
+    }
+
     return bdrv_probe_geometry(blk->bs, geo);
 }
+
+/*
+ * Updates the BlockBackendRootState object with data from the currently
+ * attached BlockDriverState.
+ */
+void blk_update_root_state(BlockBackend *blk)
+{
+    assert(blk->bs);
+
+    blk->root_state.open_flags    = blk->bs->open_flags;
+    blk->root_state.read_only     = blk->bs->read_only;
+    blk->root_state.detect_zeroes = blk->bs->detect_zeroes;
+
+    if (blk->root_state.throttle_group) {
+        g_free(blk->root_state.throttle_group);
+        throttle_group_unref(blk->root_state.throttle_state);
+    }
+    if (blk->bs->throttle_state) {
+        const char *name = throttle_group_get_name(blk->bs);
+        blk->root_state.throttle_group = g_strdup(name);
+        blk->root_state.throttle_state = throttle_group_incref(name);
+    } else {
+        blk->root_state.throttle_group = NULL;
+        blk->root_state.throttle_state = NULL;
+    }
+}
+
+/*
+ * Applies the information in the root state to the given BlockDriverState. This
+ * does not include the flags which have to be specified for bdrv_open(), use
+ * blk_get_open_flags_from_root_state() to inquire them.
+ */
+void blk_apply_root_state(BlockBackend *blk, BlockDriverState *bs)
+{
+    bs->detect_zeroes = blk->root_state.detect_zeroes;
+    if (blk->root_state.throttle_group) {
+        bdrv_io_limits_enable(bs, blk->root_state.throttle_group);
+    }
+}
+
+/*
+ * Returns the flags to be used for bdrv_open() of a BlockDriverState which is
+ * supposed to inherit the root state.
+ */
+int blk_get_open_flags_from_root_state(BlockBackend *blk)
+{
+    int bs_flags;
+
+    bs_flags = blk->root_state.read_only ? 0 : BDRV_O_RDWR;
+    bs_flags |= blk->root_state.open_flags & ~BDRV_O_RDWR;
+
+    return bs_flags;
+}
+
+BlockBackendRootState *blk_get_root_state(BlockBackend *blk)
+{
+    return &blk->root_state;
+}
diff --git a/block/bochs.c b/block/bochs.c
index 199ac2b9a..18949b9d4 100644
--- a/block/bochs.c
+++ b/block/bochs.c
@@ -103,7 +103,7 @@ static int bochs_open(BlockDriverState *bs, QDict *options, int flags,
 
     bs->read_only = 1; // no write support yet
 
-    ret = bdrv_pread(bs->file, 0, &bochs, sizeof(bochs));
+    ret = bdrv_pread(bs->file->bs, 0, &bochs, sizeof(bochs));
     if (ret < 0) {
         return ret;
     }
@@ -137,7 +137,7 @@ static int bochs_open(BlockDriverState *bs, QDict *options, int flags,
         return -ENOMEM;
     }
 
-    ret = bdrv_pread(bs->file, le32_to_cpu(bochs.header), s->catalog_bitmap,
+    ret = bdrv_pread(bs->file->bs, le32_to_cpu(bochs.header), s->catalog_bitmap,
                      s->catalog_size * 4);
     if (ret < 0) {
         goto fail;
@@ -206,7 +206,7 @@ static int64_t seek_to_sector(BlockDriverState *bs, int64_t sector_num)
         (s->extent_blocks + s->bitmap_blocks));
 
     /* read in bitmap for current extent */
-    ret = bdrv_pread(bs->file, bitmap_offset + (extent_offset / 8),
+    ret = bdrv_pread(bs->file->bs, bitmap_offset + (extent_offset / 8),
                      &bitmap_entry, 1);
     if (ret < 0) {
         return ret;
@@ -229,7 +229,7 @@ static int bochs_read(BlockDriverState *bs, int64_t sector_num,
         if (block_offset < 0) {
             return block_offset;
         } else if (block_offset > 0) {
-            ret = bdrv_pread(bs->file, block_offset, buf, 512);
+            ret = bdrv_pread(bs->file->bs, block_offset, buf, 512);
             if (ret < 0) {
                 return ret;
             }
diff --git a/block/cloop.c b/block/cloop.c
index f328be06f..4190ae06d 100644
--- a/block/cloop.c
+++ b/block/cloop.c
@@ -66,7 +66,7 @@ static int cloop_open(BlockDriverState *bs, QDict *options, int flags,
     bs->read_only = 1;
 
     /* read header */
-    ret = bdrv_pread(bs->file, 128, &s->block_size, 4);
+    ret = bdrv_pread(bs->file->bs, 128, &s->block_size, 4);
     if (ret < 0) {
         return ret;
     }
@@ -92,7 +92,7 @@ static int cloop_open(BlockDriverState *bs, QDict *options, int flags,
         return -EINVAL;
     }
 
-    ret = bdrv_pread(bs->file, 128 + 4, &s->n_blocks, 4);
+    ret = bdrv_pread(bs->file->bs, 128 + 4, &s->n_blocks, 4);
     if (ret < 0) {
         return ret;
     }
@@ -123,7 +123,7 @@ static int cloop_open(BlockDriverState *bs, QDict *options, int flags,
         return -ENOMEM;
     }
 
-    ret = bdrv_pread(bs->file, 128 + 4 + 4, s->offsets, offsets_size);
+    ret = bdrv_pread(bs->file->bs, 128 + 4 + 4, s->offsets, offsets_size);
     if (ret < 0) {
         goto fail;
     }
@@ -203,8 +203,8 @@ static inline int cloop_read_block(BlockDriverState *bs, int block_num)
         int ret;
         uint32_t bytes = s->offsets[block_num + 1] - s->offsets[block_num];
 
-        ret = bdrv_pread(bs->file, s->offsets[block_num], s->compressed_block,
-                         bytes);
+        ret = bdrv_pread(bs->file->bs, s->offsets[block_num],
+                         s->compressed_block, bytes);
         if (ret != bytes) {
             return -1;
         }
diff --git a/block/commit.c b/block/commit.c
index 7312a5bdc..a5d02aa56 100644
--- a/block/commit.c
+++ b/block/commit.c
@@ -17,6 +17,7 @@
 #include "block/blockjob.h"
 #include "qapi/qmp/qerror.h"
 #include "qemu/ratelimit.h"
+#include "sysemu/block-backend.h"
 
 enum {
     /*
@@ -213,7 +214,7 @@ void commit_start(BlockDriverState *bs, BlockDriverState *base,
 
     if ((on_error == BLOCKDEV_ON_ERROR_STOP ||
          on_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
-        !bdrv_iostatus_is_enabled(bs)) {
+        (!bs->blk || !blk_iostatus_is_enabled(bs->blk))) {
         error_setg(errp, "Invalid parameter combination");
         return;
     }
@@ -235,14 +236,14 @@ void commit_start(BlockDriverState *bs, BlockDriverState *base,
     orig_overlay_flags = bdrv_get_flags(overlay_bs);
 
     /* convert base & overlay_bs to r/w, if necessary */
-    if (!(orig_base_flags & BDRV_O_RDWR)) {
-        reopen_queue = bdrv_reopen_queue(reopen_queue, base,
-                                         orig_base_flags | BDRV_O_RDWR);
-    }
     if (!(orig_overlay_flags & BDRV_O_RDWR)) {
-        reopen_queue = bdrv_reopen_queue(reopen_queue, overlay_bs,
+        reopen_queue = bdrv_reopen_queue(reopen_queue, overlay_bs, NULL,
                                          orig_overlay_flags | BDRV_O_RDWR);
     }
+    if (!(orig_base_flags & BDRV_O_RDWR)) {
+        reopen_queue = bdrv_reopen_queue(reopen_queue, base, NULL,
+                                         orig_base_flags | BDRV_O_RDWR);
+    }
     if (reopen_queue) {
         bdrv_reopen_multiple(reopen_queue, &local_err);
         if (local_err != NULL) {
diff --git a/block/curl.c b/block/curl.c
index 032cc8ae2..89941826e 100644
--- a/block/curl.c
+++ b/block/curl.c
@@ -154,18 +154,20 @@ static int curl_sock_cb(CURL *curl, curl_socket_t fd, int action,
     DPRINTF("CURL (AIO): Sock action %d on fd %d\n", action, fd);
     switch (action) {
         case CURL_POLL_IN:
-            aio_set_fd_handler(s->aio_context, fd, curl_multi_read,
-                               NULL, state);
+            aio_set_fd_handler(s->aio_context, fd, false,
+                               curl_multi_read, NULL, state);
             break;
         case CURL_POLL_OUT:
-            aio_set_fd_handler(s->aio_context, fd, NULL, curl_multi_do, state);
+            aio_set_fd_handler(s->aio_context, fd, false,
+                               NULL, curl_multi_do, state);
             break;
         case CURL_POLL_INOUT:
-            aio_set_fd_handler(s->aio_context, fd, curl_multi_read,
-                               curl_multi_do, state);
+            aio_set_fd_handler(s->aio_context, fd, false,
+                               curl_multi_read, curl_multi_do, state);
             break;
         case CURL_POLL_REMOVE:
-            aio_set_fd_handler(s->aio_context, fd, NULL, NULL, NULL);
+            aio_set_fd_handler(s->aio_context, fd, false,
+                               NULL, NULL, NULL);
             break;
     }
 
diff --git a/block/dmg.c b/block/dmg.c
index 9f2528169..546a6f533 100644
--- a/block/dmg.c
+++ b/block/dmg.c
@@ -85,7 +85,7 @@ static int read_uint64(BlockDriverState *bs, int64_t offset, uint64_t *result)
     uint64_t buffer;
     int ret;
 
-    ret = bdrv_pread(bs->file, offset, &buffer, 8);
+    ret = bdrv_pread(bs->file->bs, offset, &buffer, 8);
     if (ret < 0) {
         return ret;
     }
@@ -99,7 +99,7 @@ static int read_uint32(BlockDriverState *bs, int64_t offset, uint32_t *result)
     uint32_t buffer;
     int ret;
 
-    ret = bdrv_pread(bs->file, offset, &buffer, 4);
+    ret = bdrv_pread(bs->file->bs, offset, &buffer, 4);
     if (ret < 0) {
         return ret;
     }
@@ -354,7 +354,7 @@ static int dmg_read_resource_fork(BlockDriverState *bs, DmgHeaderState *ds,
         offset += 4;
 
         buffer = g_realloc(buffer, count);
-        ret = bdrv_pread(bs->file, offset, buffer, count);
+        ret = bdrv_pread(bs->file->bs, offset, buffer, count);
         if (ret < 0) {
             goto fail;
         }
@@ -391,7 +391,7 @@ static int dmg_read_plist_xml(BlockDriverState *bs, DmgHeaderState *ds,
 
     buffer = g_malloc(info_length + 1);
     buffer[info_length] = '\0';
-    ret = bdrv_pread(bs->file, info_begin, buffer, info_length);
+    ret = bdrv_pread(bs->file->bs, info_begin, buffer, info_length);
     if (ret != info_length) {
         ret = -EINVAL;
         goto fail;
@@ -446,7 +446,7 @@ static int dmg_open(BlockDriverState *bs, QDict *options, int flags,
     ds.max_sectors_per_chunk = 1;
 
     /* locate the UDIF trailer */
-    offset = dmg_find_koly_offset(bs->file, errp);
+    offset = dmg_find_koly_offset(bs->file->bs, errp);
     if (offset < 0) {
         ret = offset;
         goto fail;
@@ -514,9 +514,9 @@ static int dmg_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     /* initialize zlib engine */
-    s->compressed_chunk = qemu_try_blockalign(bs->file,
+    s->compressed_chunk = qemu_try_blockalign(bs->file->bs,
                                               ds.max_compressed_size + 1);
-    s->uncompressed_chunk = qemu_try_blockalign(bs->file,
+    s->uncompressed_chunk = qemu_try_blockalign(bs->file->bs,
                                                 512 * ds.max_sectors_per_chunk);
     if (s->compressed_chunk == NULL || s->uncompressed_chunk == NULL) {
         ret = -ENOMEM;
@@ -592,7 +592,7 @@ static inline int dmg_read_chunk(BlockDriverState *bs, uint64_t sector_num)
         case 0x80000005: { /* zlib compressed */
             /* we need to buffer, because only the chunk as whole can be
              * inflated. */
-            ret = bdrv_pread(bs->file, s->offsets[chunk],
+            ret = bdrv_pread(bs->file->bs, s->offsets[chunk],
                              s->compressed_chunk, s->lengths[chunk]);
             if (ret != s->lengths[chunk]) {
                 return -1;
@@ -616,7 +616,7 @@ static inline int dmg_read_chunk(BlockDriverState *bs, uint64_t sector_num)
         case 0x80000006: /* bzip2 compressed */
             /* we need to buffer, because only the chunk as whole can be
              * inflated. */
-            ret = bdrv_pread(bs->file, s->offsets[chunk],
+            ret = bdrv_pread(bs->file->bs, s->offsets[chunk],
                              s->compressed_chunk, s->lengths[chunk]);
             if (ret != s->lengths[chunk]) {
                 return -1;
@@ -641,7 +641,7 @@ static inline int dmg_read_chunk(BlockDriverState *bs, uint64_t sector_num)
             break;
 #endif /* CONFIG_BZIP2 */
         case 1: /* copy */
-            ret = bdrv_pread(bs->file, s->offsets[chunk],
+            ret = bdrv_pread(bs->file->bs, s->offsets[chunk],
                              s->uncompressed_chunk, s->lengths[chunk]);
             if (ret != s->lengths[chunk]) {
                 return -1;
diff --git a/block/gluster.c b/block/gluster.c
index 1eb3a8c39..0857c1464 100644
--- a/block/gluster.c
+++ b/block/gluster.c
@@ -429,28 +429,23 @@ static coroutine_fn int qemu_gluster_co_write_zeroes(BlockDriverState *bs,
         int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
 {
     int ret;
-    GlusterAIOCB *acb = g_slice_new(GlusterAIOCB);
+    GlusterAIOCB acb;
     BDRVGlusterState *s = bs->opaque;
     off_t size = nb_sectors * BDRV_SECTOR_SIZE;
     off_t offset = sector_num * BDRV_SECTOR_SIZE;
 
-    acb->size = size;
-    acb->ret = 0;
-    acb->coroutine = qemu_coroutine_self();
-    acb->aio_context = bdrv_get_aio_context(bs);
+    acb.size = size;
+    acb.ret = 0;
+    acb.coroutine = qemu_coroutine_self();
+    acb.aio_context = bdrv_get_aio_context(bs);
 
-    ret = glfs_zerofill_async(s->fd, offset, size, &gluster_finish_aiocb, acb);
+    ret = glfs_zerofill_async(s->fd, offset, size, gluster_finish_aiocb, &acb);
     if (ret < 0) {
-        ret = -errno;
-        goto out;
+        return -errno;
     }
 
     qemu_coroutine_yield();
-    ret = acb->ret;
-
-out:
-    g_slice_free(GlusterAIOCB, acb);
-    return ret;
+    return acb.ret;
 }
 
 static inline bool gluster_supports_zerofill(void)
@@ -541,35 +536,30 @@ static coroutine_fn int qemu_gluster_co_rw(BlockDriverState *bs,
         int64_t sector_num, int nb_sectors, QEMUIOVector *qiov, int write)
 {
     int ret;
-    GlusterAIOCB *acb = g_slice_new(GlusterAIOCB);
+    GlusterAIOCB acb;
     BDRVGlusterState *s = bs->opaque;
     size_t size = nb_sectors * BDRV_SECTOR_SIZE;
     off_t offset = sector_num * BDRV_SECTOR_SIZE;
 
-    acb->size = size;
-    acb->ret = 0;
-    acb->coroutine = qemu_coroutine_self();
-    acb->aio_context = bdrv_get_aio_context(bs);
+    acb.size = size;
+    acb.ret = 0;
+    acb.coroutine = qemu_coroutine_self();
+    acb.aio_context = bdrv_get_aio_context(bs);
 
     if (write) {
         ret = glfs_pwritev_async(s->fd, qiov->iov, qiov->niov, offset, 0,
-            &gluster_finish_aiocb, acb);
+            gluster_finish_aiocb, &acb);
     } else {
         ret = glfs_preadv_async(s->fd, qiov->iov, qiov->niov, offset, 0,
-            &gluster_finish_aiocb, acb);
+            gluster_finish_aiocb, &acb);
     }
 
     if (ret < 0) {
-        ret = -errno;
-        goto out;
+        return -errno;
     }
 
     qemu_coroutine_yield();
-    ret = acb->ret;
-
-out:
-    g_slice_free(GlusterAIOCB, acb);
-    return ret;
+    return acb.ret;
 }
 
 static int qemu_gluster_truncate(BlockDriverState *bs, int64_t offset)
@@ -600,26 +590,21 @@ static coroutine_fn int qemu_gluster_co_writev(BlockDriverState *bs,
 static coroutine_fn int qemu_gluster_co_flush_to_disk(BlockDriverState *bs)
 {
     int ret;
-    GlusterAIOCB *acb = g_slice_new(GlusterAIOCB);
+    GlusterAIOCB acb;
     BDRVGlusterState *s = bs->opaque;
 
-    acb->size = 0;
-    acb->ret = 0;
-    acb->coroutine = qemu_coroutine_self();
-    acb->aio_context = bdrv_get_aio_context(bs);
+    acb.size = 0;
+    acb.ret = 0;
+    acb.coroutine = qemu_coroutine_self();
+    acb.aio_context = bdrv_get_aio_context(bs);
 
-    ret = glfs_fsync_async(s->fd, &gluster_finish_aiocb, acb);
+    ret = glfs_fsync_async(s->fd, gluster_finish_aiocb, &acb);
     if (ret < 0) {
-        ret = -errno;
-        goto out;
+        return -errno;
     }
 
     qemu_coroutine_yield();
-    ret = acb->ret;
-
-out:
-    g_slice_free(GlusterAIOCB, acb);
-    return ret;
+    return acb.ret;
 }
 
 #ifdef CONFIG_GLUSTERFS_DISCARD
@@ -627,28 +612,23 @@ static coroutine_fn int qemu_gluster_co_discard(BlockDriverState *bs,
         int64_t sector_num, int nb_sectors)
 {
     int ret;
-    GlusterAIOCB *acb = g_slice_new(GlusterAIOCB);
+    GlusterAIOCB acb;
     BDRVGlusterState *s = bs->opaque;
     size_t size = nb_sectors * BDRV_SECTOR_SIZE;
     off_t offset = sector_num * BDRV_SECTOR_SIZE;
 
-    acb->size = 0;
-    acb->ret = 0;
-    acb->coroutine = qemu_coroutine_self();
-    acb->aio_context = bdrv_get_aio_context(bs);
+    acb.size = 0;
+    acb.ret = 0;
+    acb.coroutine = qemu_coroutine_self();
+    acb.aio_context = bdrv_get_aio_context(bs);
 
-    ret = glfs_discard_async(s->fd, offset, size, &gluster_finish_aiocb, acb);
+    ret = glfs_discard_async(s->fd, offset, size, gluster_finish_aiocb, &acb);
     if (ret < 0) {
-        ret = -errno;
-        goto out;
+        return -errno;
     }
 
     qemu_coroutine_yield();
-    ret = acb->ret;
-
-out:
-    g_slice_free(GlusterAIOCB, acb);
-    return ret;
+    return acb.ret;
 }
 #endif
 
diff --git a/block/io.c b/block/io.c
index d4bc83b33..e00fb5d69 100644
--- a/block/io.c
+++ b/block/io.c
@@ -23,6 +23,7 @@
  */
 
 #include "trace.h"
+#include "sysemu/block-backend.h"
 #include "block/blockjob.h"
 #include "block/block_int.h"
 #include "block/throttle-groups.h"
@@ -156,38 +157,38 @@ void bdrv_refresh_limits(BlockDriverState *bs, Error **errp)
 
     /* Take some limits from the children as a default */
     if (bs->file) {
-        bdrv_refresh_limits(bs->file, &local_err);
+        bdrv_refresh_limits(bs->file->bs, &local_err);
         if (local_err) {
             error_propagate(errp, local_err);
             return;
         }
-        bs->bl.opt_transfer_length = bs->file->bl.opt_transfer_length;
-        bs->bl.max_transfer_length = bs->file->bl.max_transfer_length;
-        bs->bl.min_mem_alignment = bs->file->bl.min_mem_alignment;
-        bs->bl.opt_mem_alignment = bs->file->bl.opt_mem_alignment;
+        bs->bl.opt_transfer_length = bs->file->bs->bl.opt_transfer_length;
+        bs->bl.max_transfer_length = bs->file->bs->bl.max_transfer_length;
+        bs->bl.min_mem_alignment = bs->file->bs->bl.min_mem_alignment;
+        bs->bl.opt_mem_alignment = bs->file->bs->bl.opt_mem_alignment;
     } else {
         bs->bl.min_mem_alignment = 512;
         bs->bl.opt_mem_alignment = getpagesize();
     }
 
-    if (bs->backing_hd) {
-        bdrv_refresh_limits(bs->backing_hd, &local_err);
+    if (bs->backing) {
+        bdrv_refresh_limits(bs->backing->bs, &local_err);
         if (local_err) {
             error_propagate(errp, local_err);
             return;
         }
         bs->bl.opt_transfer_length =
             MAX(bs->bl.opt_transfer_length,
-                bs->backing_hd->bl.opt_transfer_length);
+                bs->backing->bs->bl.opt_transfer_length);
         bs->bl.max_transfer_length =
             MIN_NON_ZERO(bs->bl.max_transfer_length,
-                         bs->backing_hd->bl.max_transfer_length);
+                         bs->backing->bs->bl.max_transfer_length);
         bs->bl.opt_mem_alignment =
             MAX(bs->bl.opt_mem_alignment,
-                bs->backing_hd->bl.opt_mem_alignment);
+                bs->backing->bs->bl.opt_mem_alignment);
         bs->bl.min_mem_alignment =
             MAX(bs->bl.min_mem_alignment,
-                bs->backing_hd->bl.min_mem_alignment);
+                bs->backing->bs->bl.min_mem_alignment);
     }
 
     /* Then let the driver override it */
@@ -213,8 +214,10 @@ void bdrv_disable_copy_on_read(BlockDriverState *bs)
 }
 
 /* Check if any requests are in-flight (including throttled requests) */
-static bool bdrv_requests_pending(BlockDriverState *bs)
+bool bdrv_requests_pending(BlockDriverState *bs)
 {
+    BdrvChild *child;
+
     if (!QLIST_EMPTY(&bs->tracked_requests)) {
         return true;
     }
@@ -224,17 +227,31 @@ static bool bdrv_requests_pending(BlockDriverState *bs)
     if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) {
         return true;
     }
-    if (bs->file && bdrv_requests_pending(bs->file)) {
-        return true;
-    }
-    if (bs->backing_hd && bdrv_requests_pending(bs->backing_hd)) {
-        return true;
+
+    QLIST_FOREACH(child, &bs->children, next) {
+        if (bdrv_requests_pending(child->bs)) {
+            return true;
+        }
     }
+
     return false;
 }
 
+static void bdrv_drain_recurse(BlockDriverState *bs)
+{
+    BdrvChild *child;
+
+    if (bs->drv && bs->drv->bdrv_drain) {
+        bs->drv->bdrv_drain(bs);
+    }
+    QLIST_FOREACH(child, &bs->children, next) {
+        bdrv_drain_recurse(child->bs);
+    }
+}
+
 /*
- * Wait for pending requests to complete on a single BlockDriverState subtree
+ * Wait for pending requests to complete on a single BlockDriverState subtree,
+ * and suspend block driver's internal I/O until next request arrives.
  *
  * Note that unlike bdrv_drain_all(), the caller must hold the BlockDriverState
  * AioContext.
@@ -247,6 +264,7 @@ void bdrv_drain(BlockDriverState *bs)
 {
     bool busy = true;
 
+    bdrv_drain_recurse(bs);
     while (busy) {
         /* Keep iterating */
          bdrv_flush_io_queue(bs);
@@ -344,13 +362,14 @@ static void tracked_request_end(BdrvTrackedRequest *req)
 static void tracked_request_begin(BdrvTrackedRequest *req,
                                   BlockDriverState *bs,
                                   int64_t offset,
-                                  unsigned int bytes, bool is_write)
+                                  unsigned int bytes,
+                                  enum BdrvTrackedRequestType type)
 {
     *req = (BdrvTrackedRequest){
         .bs = bs,
         .offset         = offset,
         .bytes          = bytes,
-        .is_write       = is_write,
+        .type           = type,
         .co             = qemu_coroutine_self(),
         .serialising    = false,
         .overlap_offset = offset,
@@ -844,7 +863,9 @@ static int coroutine_fn bdrv_aligned_preadv(BlockDriverState *bs,
         mark_request_serialising(req, bdrv_get_cluster_size(bs));
     }
 
-    wait_serialising_requests(req);
+    if (!(flags & BDRV_REQ_NO_SERIALISING)) {
+        wait_serialising_requests(req);
+    }
 
     if (flags & BDRV_REQ_COPY_ON_READ) {
         int pnum;
@@ -932,7 +953,8 @@ static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
         return ret;
     }
 
-    if (bs->copy_on_read) {
+    /* Don't do copy-on-read if we read data before write operation */
+    if (bs->copy_on_read && !(flags & BDRV_REQ_NO_SERIALISING)) {
         flags |= BDRV_REQ_COPY_ON_READ;
     }
 
@@ -966,7 +988,7 @@ static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
         bytes = ROUND_UP(bytes, align);
     }
 
-    tracked_request_begin(&req, bs, offset, bytes, false);
+    tracked_request_begin(&req, bs, offset, bytes, BDRV_TRACKED_READ);
     ret = bdrv_aligned_preadv(bs, &req, offset, bytes, align,
                               use_local_qiov ? &local_qiov : qiov,
                               flags);
@@ -1001,6 +1023,15 @@ int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
     return bdrv_co_do_readv(bs, sector_num, nb_sectors, qiov, 0);
 }
 
+int coroutine_fn bdrv_co_readv_no_serialising(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov)
+{
+    trace_bdrv_co_readv_no_serialising(bs, sector_num, nb_sectors);
+
+    return bdrv_co_do_readv(bs, sector_num, nb_sectors, qiov,
+                            BDRV_REQ_NO_SERIALISING);
+}
+
 int coroutine_fn bdrv_co_copy_on_readv(BlockDriverState *bs,
     int64_t sector_num, int nb_sectors, QEMUIOVector *qiov)
 {
@@ -1127,13 +1158,13 @@ static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
     if (ret < 0) {
         /* Do nothing, write notifier decided to fail this request */
     } else if (flags & BDRV_REQ_ZERO_WRITE) {
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_ZERO);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_ZERO);
         ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
     } else {
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV);
         ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
     }
-    BLKDBG_EVENT(bs, BLKDBG_PWRITEV_DONE);
+    bdrv_debug_event(bs, BLKDBG_PWRITEV_DONE);
 
     if (ret == 0 && !bs->enable_write_cache) {
         ret = bdrv_co_flush(bs);
@@ -1141,7 +1172,9 @@ static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
 
     bdrv_set_dirty(bs, sector_num, nb_sectors);
 
-    block_acct_highest_sector(&bs->stats, sector_num, nb_sectors);
+    if (bs->wr_highest_offset < offset + bytes) {
+        bs->wr_highest_offset = offset + bytes;
+    }
 
     if (ret >= 0) {
         bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
@@ -1182,13 +1215,13 @@ static int coroutine_fn bdrv_co_do_zero_pwritev(BlockDriverState *bs,
         /* RMW the unaligned part before head. */
         mark_request_serialising(req, align);
         wait_serialising_requests(req);
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_HEAD);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_HEAD);
         ret = bdrv_aligned_preadv(bs, req, offset & ~(align - 1), align,
                                   align, &local_qiov, 0);
         if (ret < 0) {
             goto fail;
         }
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
 
         memset(buf + head_padding_bytes, 0, zero_bytes);
         ret = bdrv_aligned_pwritev(bs, req, offset & ~(align - 1), align,
@@ -1220,13 +1253,13 @@ static int coroutine_fn bdrv_co_do_zero_pwritev(BlockDriverState *bs,
         /* RMW the unaligned part after tail. */
         mark_request_serialising(req, align);
         wait_serialising_requests(req);
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_TAIL);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_TAIL);
         ret = bdrv_aligned_preadv(bs, req, offset, align,
                                   align, &local_qiov, 0);
         if (ret < 0) {
             goto fail;
         }
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
 
         memset(buf, 0, bytes);
         ret = bdrv_aligned_pwritev(bs, req, offset, align,
@@ -1276,7 +1309,7 @@ static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
      * Pad qiov with the read parts and be sure to have a tracked request not
      * only for bdrv_aligned_pwritev, but also for the reads of the RMW cycle.
      */
-    tracked_request_begin(&req, bs, offset, bytes, true);
+    tracked_request_begin(&req, bs, offset, bytes, BDRV_TRACKED_WRITE);
 
     if (!qiov) {
         ret = bdrv_co_do_zero_pwritev(bs, offset, bytes, flags, &req);
@@ -1297,13 +1330,13 @@ static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
         };
         qemu_iovec_init_external(&head_qiov, &head_iov, 1);
 
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_HEAD);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_HEAD);
         ret = bdrv_aligned_preadv(bs, &req, offset & ~(align - 1), align,
                                   align, &head_qiov, 0);
         if (ret < 0) {
             goto fail;
         }
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
 
         qemu_iovec_init(&local_qiov, qiov->niov + 2);
         qemu_iovec_add(&local_qiov, head_buf, offset & (align - 1));
@@ -1331,13 +1364,13 @@ static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
         };
         qemu_iovec_init_external(&tail_qiov, &tail_iov, 1);
 
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_TAIL);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_TAIL);
         ret = bdrv_aligned_preadv(bs, &req, (offset + bytes) & ~(align - 1), align,
                                   align, &tail_qiov, 0);
         if (ret < 0) {
             goto fail;
         }
-        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
+        bdrv_debug_event(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
 
         if (!use_local_qiov) {
             qemu_iovec_init(&local_qiov, qiov->niov + 1);
@@ -1486,7 +1519,7 @@ static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
 
     if (ret & BDRV_BLOCK_RAW) {
         assert(ret & BDRV_BLOCK_OFFSET_VALID);
-        return bdrv_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
+        return bdrv_get_block_status(bs->file->bs, ret >> BDRV_SECTOR_BITS,
                                      *pnum, pnum);
     }
 
@@ -1495,8 +1528,8 @@ static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
     } else {
         if (bdrv_unallocated_blocks_are_zero(bs)) {
             ret |= BDRV_BLOCK_ZERO;
-        } else if (bs->backing_hd) {
-            BlockDriverState *bs2 = bs->backing_hd;
+        } else if (bs->backing) {
+            BlockDriverState *bs2 = bs->backing->bs;
             int64_t nb_sectors2 = bdrv_nb_sectors(bs2);
             if (nb_sectors2 >= 0 && sector_num >= nb_sectors2) {
                 ret |= BDRV_BLOCK_ZERO;
@@ -1509,7 +1542,7 @@ static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
         (ret & BDRV_BLOCK_OFFSET_VALID)) {
         int file_pnum;
 
-        ret2 = bdrv_co_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
+        ret2 = bdrv_co_get_block_status(bs->file->bs, ret >> BDRV_SECTOR_BITS,
                                         *pnum, &file_pnum);
         if (ret2 >= 0) {
             /* Ignore errors.  This is just providing extra information, it
@@ -1541,7 +1574,7 @@ static int64_t coroutine_fn bdrv_co_get_block_status_above(BlockDriverState *bs,
     int64_t ret = 0;
 
     assert(bs != base);
-    for (p = bs; p != base; p = p->backing_hd) {
+    for (p = bs; p != base; p = backing_bs(p)) {
         ret = bdrv_co_get_block_status(p, sector_num, nb_sectors, pnum);
         if (ret < 0 || ret & BDRV_BLOCK_ALLOCATED) {
             break;
@@ -1604,7 +1637,7 @@ int64_t bdrv_get_block_status(BlockDriverState *bs,
                               int64_t sector_num,
                               int nb_sectors, int *pnum)
 {
-    return bdrv_get_block_status_above(bs, bs->backing_hd,
+    return bdrv_get_block_status_above(bs, backing_bs(bs),
                                        sector_num, nb_sectors, pnum);
 }
 
@@ -1662,7 +1695,7 @@ int bdrv_is_allocated_above(BlockDriverState *top,
             n = pnum_inter;
         }
 
-        intermediate = intermediate->backing_hd;
+        intermediate = backing_bs(intermediate);
     }
 
     *pnum = n;
@@ -1713,7 +1746,7 @@ int bdrv_writev_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos)
     } 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 bdrv_writev_vmstate(bs->file->bs, qiov, pos);
     }
 
     return -ENOTSUP;
@@ -1728,7 +1761,7 @@ int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
     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 bdrv_load_vmstate(bs->file->bs, buf, pos, size);
     return -ENOTSUP;
 }
 
@@ -1893,7 +1926,10 @@ static int multiwrite_merge(BlockDriverState *bs, BlockRequest *reqs,
         }
     }
 
-    block_acct_merge_done(&bs->stats, BLOCK_ACCT_WRITE, num_reqs - outidx - 1);
+    if (bs->blk) {
+        block_acct_merge_done(blk_get_stats(bs->blk), BLOCK_ACCT_WRITE,
+                              num_reqs - outidx - 1);
+    }
 
     return outidx + 1;
 }
@@ -2208,7 +2244,7 @@ void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
 {
     BlockAIOCB *acb;
 
-    acb = g_slice_alloc(aiocb_info->aiocb_size);
+    acb = g_malloc(aiocb_info->aiocb_size);
     acb->aiocb_info = aiocb_info;
     acb->bs = bs;
     acb->cb = cb;
@@ -2228,7 +2264,7 @@ void qemu_aio_unref(void *p)
     BlockAIOCB *acb = p;
     assert(acb->refcnt > 0);
     if (--acb->refcnt == 0) {
-        g_slice_free1(acb->aiocb_info->aiocb_size, acb);
+        g_free(acb);
     }
 }
 
@@ -2298,18 +2334,20 @@ static void coroutine_fn bdrv_flush_co_entry(void *opaque)
 int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
 {
     int ret;
+    BdrvTrackedRequest req;
 
     if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs) ||
         bdrv_is_sg(bs)) {
         return 0;
     }
 
+    tracked_request_begin(&req, bs, 0, 0, BDRV_TRACKED_FLUSH);
     /* 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;
+            goto out;
         }
     }
 
@@ -2349,14 +2387,17 @@ int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
         ret = 0;
     }
     if (ret < 0) {
-        return ret;
+        goto out;
     }
 
     /* 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);
+    ret = bs->file ? bdrv_co_flush(bs->file->bs) : 0;
+out:
+    tracked_request_end(&req);
+    return ret;
 }
 
 int bdrv_flush(BlockDriverState *bs)
@@ -2399,6 +2440,7 @@ static void coroutine_fn bdrv_discard_co_entry(void *opaque)
 int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                  int nb_sectors)
 {
+    BdrvTrackedRequest req;
     int max_discard, ret;
 
     if (!bs->drv) {
@@ -2421,6 +2463,8 @@ int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
         return 0;
     }
 
+    tracked_request_begin(&req, bs, sector_num, nb_sectors,
+                          BDRV_TRACKED_DISCARD);
     bdrv_set_dirty(bs, sector_num, nb_sectors);
 
     max_discard = MIN_NON_ZERO(bs->bl.max_discard, BDRV_REQUEST_MAX_SECTORS);
@@ -2454,20 +2498,24 @@ int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
             acb = bs->drv->bdrv_aio_discard(bs, sector_num, nb_sectors,
                                             bdrv_co_io_em_complete, &co);
             if (acb == NULL) {
-                return -EIO;
+                ret = -EIO;
+                goto out;
             } else {
                 qemu_coroutine_yield();
                 ret = co.ret;
             }
         }
         if (ret && ret != -ENOTSUP) {
-            return ret;
+            goto out;
         }
 
         sector_num += num;
         nb_sectors -= num;
     }
-    return 0;
+    ret = 0;
+out:
+    tracked_request_end(&req);
+    return ret;
 }
 
 int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
@@ -2496,26 +2544,109 @@ int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
     return rwco.ret;
 }
 
-/* needed for generic scsi interface */
+typedef struct {
+    CoroutineIOCompletion *co;
+    QEMUBH *bh;
+} BdrvIoctlCompletionData;
 
-int bdrv_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
+static void bdrv_ioctl_bh_cb(void *opaque)
+{
+    BdrvIoctlCompletionData *data = opaque;
+
+    bdrv_co_io_em_complete(data->co, -ENOTSUP);
+    qemu_bh_delete(data->bh);
+}
+
+static int bdrv_co_do_ioctl(BlockDriverState *bs, int req, void *buf)
 {
     BlockDriver *drv = bs->drv;
+    BdrvTrackedRequest tracked_req;
+    CoroutineIOCompletion co = {
+        .coroutine = qemu_coroutine_self(),
+    };
+    BlockAIOCB *acb;
 
-    if (drv && drv->bdrv_ioctl)
-        return drv->bdrv_ioctl(bs, req, buf);
-    return -ENOTSUP;
+    tracked_request_begin(&tracked_req, bs, 0, 0, BDRV_TRACKED_IOCTL);
+    if (!drv || !drv->bdrv_aio_ioctl) {
+        co.ret = -ENOTSUP;
+        goto out;
+    }
+
+    acb = drv->bdrv_aio_ioctl(bs, req, buf, bdrv_co_io_em_complete, &co);
+    if (!acb) {
+        BdrvIoctlCompletionData *data = g_new(BdrvIoctlCompletionData, 1);
+        data->bh = aio_bh_new(bdrv_get_aio_context(bs),
+                                bdrv_ioctl_bh_cb, data);
+        data->co = &co;
+        qemu_bh_schedule(data->bh);
+    }
+    qemu_coroutine_yield();
+out:
+    tracked_request_end(&tracked_req);
+    return co.ret;
+}
+
+typedef struct {
+    BlockDriverState *bs;
+    int req;
+    void *buf;
+    int ret;
+} BdrvIoctlCoData;
+
+static void coroutine_fn bdrv_co_ioctl_entry(void *opaque)
+{
+    BdrvIoctlCoData *data = opaque;
+    data->ret = bdrv_co_do_ioctl(data->bs, data->req, data->buf);
+}
+
+/* needed for generic scsi interface */
+int bdrv_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
+{
+    BdrvIoctlCoData data = {
+        .bs = bs,
+        .req = req,
+        .buf = buf,
+        .ret = -EINPROGRESS,
+    };
+
+    if (qemu_in_coroutine()) {
+        /* Fast-path if already in coroutine context */
+        bdrv_co_ioctl_entry(&data);
+    } else {
+        Coroutine *co = qemu_coroutine_create(bdrv_co_ioctl_entry);
+        qemu_coroutine_enter(co, &data);
+    }
+    while (data.ret == -EINPROGRESS) {
+        aio_poll(bdrv_get_aio_context(bs), true);
+    }
+    return data.ret;
+}
+
+static void coroutine_fn bdrv_co_aio_ioctl_entry(void *opaque)
+{
+    BlockAIOCBCoroutine *acb = opaque;
+    acb->req.error = bdrv_co_do_ioctl(acb->common.bs,
+                                      acb->req.req, acb->req.buf);
+    bdrv_co_complete(acb);
 }
 
 BlockAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
         unsigned long int req, void *buf,
         BlockCompletionFunc *cb, void *opaque)
 {
-    BlockDriver *drv = bs->drv;
+    BlockAIOCBCoroutine *acb = qemu_aio_get(&bdrv_em_co_aiocb_info,
+                                            bs, cb, opaque);
+    Coroutine *co;
 
-    if (drv && drv->bdrv_aio_ioctl)
-        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
-    return NULL;
+    acb->need_bh = true;
+    acb->req.error = -EINPROGRESS;
+    acb->req.req = req;
+    acb->req.buf = buf;
+    co = qemu_coroutine_create(bdrv_co_aio_ioctl_entry);
+    qemu_coroutine_enter(co, acb);
+
+    bdrv_co_maybe_schedule_bh(acb);
+    return &acb->common;
 }
 
 void *qemu_blockalign(BlockDriverState *bs, size_t size)
@@ -2584,7 +2715,7 @@ void bdrv_io_plug(BlockDriverState *bs)
     if (drv && drv->bdrv_io_plug) {
         drv->bdrv_io_plug(bs);
     } else if (bs->file) {
-        bdrv_io_plug(bs->file);
+        bdrv_io_plug(bs->file->bs);
     }
 }
 
@@ -2594,7 +2725,7 @@ void bdrv_io_unplug(BlockDriverState *bs)
     if (drv && drv->bdrv_io_unplug) {
         drv->bdrv_io_unplug(bs);
     } else if (bs->file) {
-        bdrv_io_unplug(bs->file);
+        bdrv_io_unplug(bs->file->bs);
     }
 }
 
@@ -2604,7 +2735,24 @@ void bdrv_flush_io_queue(BlockDriverState *bs)
     if (drv && drv->bdrv_flush_io_queue) {
         drv->bdrv_flush_io_queue(bs);
     } else if (bs->file) {
-        bdrv_flush_io_queue(bs->file);
+        bdrv_flush_io_queue(bs->file->bs);
     }
     bdrv_start_throttled_reqs(bs);
 }
+
+void bdrv_drained_begin(BlockDriverState *bs)
+{
+    if (!bs->quiesce_counter++) {
+        aio_disable_external(bdrv_get_aio_context(bs));
+    }
+    bdrv_drain(bs);
+}
+
+void bdrv_drained_end(BlockDriverState *bs)
+{
+    assert(bs->quiesce_counter > 0);
+    if (--bs->quiesce_counter > 0) {
+        return;
+    }
+    aio_enable_external(bdrv_get_aio_context(bs));
+}
diff --git a/block/iscsi.c b/block/iscsi.c
index 93f1ee4c6..bd1f1bfcd 100644
--- a/block/iscsi.c
+++ b/block/iscsi.c
@@ -84,6 +84,7 @@ typedef struct IscsiTask {
     IscsiLun *iscsilun;
     QEMUTimer retry_timer;
     bool force_next_flush;
+    int err_code;
 } IscsiTask;
 
 typedef struct IscsiAIOCB {
@@ -96,6 +97,7 @@ typedef struct IscsiAIOCB {
     int status;
     int64_t sector_num;
     int nb_sectors;
+    int ret;
 #ifdef __linux__
     sg_io_hdr_t *ioh;
 #endif
@@ -169,19 +171,70 @@ static inline unsigned exp_random(double mean)
     return -mean * log((double)rand() / RAND_MAX);
 }
 
-/* SCSI_STATUS_TASK_SET_FULL and SCSI_STATUS_TIMEOUT were introduced
- * in libiscsi 1.10.0 as part of an enum. The LIBISCSI_API_VERSION
- * macro was introduced in 1.11.0. So use the API_VERSION macro as
- * a hint that the macros are defined and define them ourselves
- * otherwise to keep the required libiscsi version at 1.9.0 */
-#if !defined(LIBISCSI_API_VERSION)
-#define QEMU_SCSI_STATUS_TASK_SET_FULL  0x28
-#define QEMU_SCSI_STATUS_TIMEOUT        0x0f000002
-#else
-#define QEMU_SCSI_STATUS_TASK_SET_FULL  SCSI_STATUS_TASK_SET_FULL
-#define QEMU_SCSI_STATUS_TIMEOUT        SCSI_STATUS_TIMEOUT
+/* SCSI_SENSE_ASCQ_INVALID_FIELD_IN_PARAMETER_LIST was introduced in
+ * libiscsi 1.10.0, together with other constants we need.  Use it as
+ * a hint that we have to define them ourselves if needed, to keep the
+ * minimum required libiscsi version at 1.9.0.  We use an ASCQ macro for
+ * the test because SCSI_STATUS_* is an enum.
+ *
+ * To guard against future changes where SCSI_SENSE_ASCQ_* also becomes
+ * an enum, check against the LIBISCSI_API_VERSION macro, which was
+ * introduced in 1.11.0.  If it is present, there is no need to define
+ * anything.
+ */
+#if !defined(SCSI_SENSE_ASCQ_INVALID_FIELD_IN_PARAMETER_LIST) && \
+    !defined(LIBISCSI_API_VERSION)
+#define SCSI_STATUS_TASK_SET_FULL                          0x28
+#define SCSI_STATUS_TIMEOUT                                0x0f000002
+#define SCSI_SENSE_ASCQ_INVALID_FIELD_IN_PARAMETER_LIST    0x2600
+#define SCSI_SENSE_ASCQ_PARAMETER_LIST_LENGTH_ERROR        0x1a00
 #endif
 
+static int iscsi_translate_sense(struct scsi_sense *sense)
+{
+    int ret;
+
+    switch (sense->key) {
+    case SCSI_SENSE_NOT_READY:
+        return -EBUSY;
+    case SCSI_SENSE_DATA_PROTECTION:
+        return -EACCES;
+    case SCSI_SENSE_COMMAND_ABORTED:
+        return -ECANCELED;
+    case SCSI_SENSE_ILLEGAL_REQUEST:
+        /* Parse ASCQ */
+        break;
+    default:
+        return -EIO;
+    }
+    switch (sense->ascq) {
+    case SCSI_SENSE_ASCQ_PARAMETER_LIST_LENGTH_ERROR:
+    case SCSI_SENSE_ASCQ_INVALID_OPERATION_CODE:
+    case SCSI_SENSE_ASCQ_INVALID_FIELD_IN_CDB:
+    case SCSI_SENSE_ASCQ_INVALID_FIELD_IN_PARAMETER_LIST:
+        ret = -EINVAL;
+        break;
+    case SCSI_SENSE_ASCQ_LBA_OUT_OF_RANGE:
+        ret = -ENOSPC;
+        break;
+    case SCSI_SENSE_ASCQ_LOGICAL_UNIT_NOT_SUPPORTED:
+        ret = -ENOTSUP;
+        break;
+    case SCSI_SENSE_ASCQ_MEDIUM_NOT_PRESENT:
+    case SCSI_SENSE_ASCQ_MEDIUM_NOT_PRESENT_TRAY_CLOSED:
+    case SCSI_SENSE_ASCQ_MEDIUM_NOT_PRESENT_TRAY_OPEN:
+        ret = -ENOMEDIUM;
+        break;
+    case SCSI_SENSE_ASCQ_WRITE_PROTECTED:
+        ret = -EACCES;
+        break;
+    default:
+        ret = -EIO;
+        break;
+    }
+    return ret;
+}
+
 static void
 iscsi_co_generic_cb(struct iscsi_context *iscsi, int status,
                         void *command_data, void *opaque)
@@ -203,11 +256,11 @@ iscsi_co_generic_cb(struct iscsi_context *iscsi, int status,
                 goto out;
             }
             if (status == SCSI_STATUS_BUSY ||
-                status == QEMU_SCSI_STATUS_TIMEOUT ||
-                status == QEMU_SCSI_STATUS_TASK_SET_FULL) {
+                status == SCSI_STATUS_TIMEOUT ||
+                status == SCSI_STATUS_TASK_SET_FULL) {
                 unsigned retry_time =
                     exp_random(iscsi_retry_times[iTask->retries - 1]);
-                if (status == QEMU_SCSI_STATUS_TIMEOUT) {
+                if (status == SCSI_STATUS_TIMEOUT) {
                     /* make sure the request is rescheduled AFTER the
                      * reconnect is initiated */
                     retry_time = EVENT_INTERVAL * 2;
@@ -226,6 +279,7 @@ iscsi_co_generic_cb(struct iscsi_context *iscsi, int status,
                 return;
             }
         }
+        iTask->err_code = iscsi_translate_sense(&task->sense);
         error_report("iSCSI Failure: %s", iscsi_get_error(iscsi));
     } else {
         iTask->iscsilun->force_next_flush |= iTask->force_next_flush;
@@ -291,8 +345,8 @@ iscsi_set_events(IscsiLun *iscsilun)
     int ev = iscsi_which_events(iscsi);
 
     if (ev != iscsilun->events) {
-        aio_set_fd_handler(iscsilun->aio_context,
-                           iscsi_get_fd(iscsi),
+        aio_set_fd_handler(iscsilun->aio_context, iscsi_get_fd(iscsi),
+                           false,
                            (ev & POLLIN) ? iscsi_process_read : NULL,
                            (ev & POLLOUT) ? iscsi_process_write : NULL,
                            iscsilun);
@@ -455,7 +509,7 @@ retry:
     }
 
     if (iTask.status != SCSI_STATUS_GOOD) {
-        return -EIO;
+        return iTask.err_code;
     }
 
     iscsi_allocationmap_set(iscsilun, sector_num, nb_sectors);
@@ -644,7 +698,7 @@ retry:
     }
 
     if (iTask.status != SCSI_STATUS_GOOD) {
-        return -EIO;
+        return iTask.err_code;
     }
 
     return 0;
@@ -683,7 +737,7 @@ retry:
     }
 
     if (iTask.status != SCSI_STATUS_GOOD) {
-        return -EIO;
+        return iTask.err_code;
     }
 
     return 0;
@@ -703,7 +757,7 @@ iscsi_aio_ioctl_cb(struct iscsi_context *iscsi, int status,
     if (status < 0) {
         error_report("Failed to ioctl(SG_IO) to iSCSI lun. %s",
                      iscsi_get_error(iscsi));
-        acb->status = -EIO;
+        acb->status = iscsi_translate_sense(&acb->task->sense);
     }
 
     acb->ioh->driver_status = 0;
@@ -726,6 +780,38 @@ iscsi_aio_ioctl_cb(struct iscsi_context *iscsi, int status,
     iscsi_schedule_bh(acb);
 }
 
+static void iscsi_ioctl_bh_completion(void *opaque)
+{
+    IscsiAIOCB *acb = opaque;
+
+    qemu_bh_delete(acb->bh);
+    acb->common.cb(acb->common.opaque, acb->ret);
+    qemu_aio_unref(acb);
+}
+
+static void iscsi_ioctl_handle_emulated(IscsiAIOCB *acb, int req, void *buf)
+{
+    BlockDriverState *bs = acb->common.bs;
+    IscsiLun *iscsilun = bs->opaque;
+    int ret = 0;
+
+    switch (req) {
+    case SG_GET_VERSION_NUM:
+        *(int *)buf = 30000;
+        break;
+    case SG_GET_SCSI_ID:
+        ((struct sg_scsi_id *)buf)->scsi_type = iscsilun->type;
+        break;
+    default:
+        ret = -EINVAL;
+    }
+    assert(!acb->bh);
+    acb->bh = aio_bh_new(bdrv_get_aio_context(bs),
+                         iscsi_ioctl_bh_completion, acb);
+    acb->ret = ret;
+    qemu_bh_schedule(acb->bh);
+}
+
 static BlockAIOCB *iscsi_aio_ioctl(BlockDriverState *bs,
         unsigned long int req, void *buf,
         BlockCompletionFunc *cb, void *opaque)
@@ -735,8 +821,6 @@ static BlockAIOCB *iscsi_aio_ioctl(BlockDriverState *bs,
     struct iscsi_data data;
     IscsiAIOCB *acb;
 
-    assert(req == SG_IO);
-
     acb = qemu_aio_get(&iscsi_aiocb_info, bs, cb, opaque);
 
     acb->iscsilun = iscsilun;
@@ -745,6 +829,11 @@ static BlockAIOCB *iscsi_aio_ioctl(BlockDriverState *bs,
     acb->buf         = NULL;
     acb->ioh         = buf;
 
+    if (req != SG_IO) {
+        iscsi_ioctl_handle_emulated(acb, req, buf);
+        return &acb->common;
+    }
+
     acb->task = malloc(sizeof(struct scsi_task));
     if (acb->task == NULL) {
         error_report("iSCSI: Failed to allocate task for scsi command. %s",
@@ -809,38 +898,6 @@ static BlockAIOCB *iscsi_aio_ioctl(BlockDriverState *bs,
     return &acb->common;
 }
 
-static void ioctl_cb(void *opaque, int status)
-{
-    int *p_status = opaque;
-    *p_status = status;
-}
-
-static int iscsi_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
-{
-    IscsiLun *iscsilun = bs->opaque;
-    int status;
-
-    switch (req) {
-    case SG_GET_VERSION_NUM:
-        *(int *)buf = 30000;
-        break;
-    case SG_GET_SCSI_ID:
-        ((struct sg_scsi_id *)buf)->scsi_type = iscsilun->type;
-        break;
-    case SG_IO:
-        status = -EINPROGRESS;
-        iscsi_aio_ioctl(bs, req, buf, ioctl_cb, &status);
-
-        while (status == -EINPROGRESS) {
-            aio_poll(iscsilun->aio_context, true);
-        }
-
-        return 0;
-    default:
-        return -1;
-    }
-    return 0;
-}
 #endif
 
 static int64_t
@@ -905,7 +962,7 @@ retry:
     }
 
     if (iTask.status != SCSI_STATUS_GOOD) {
-        return -EIO;
+        return iTask.err_code;
     }
 
     iscsi_allocationmap_clear(iscsilun, sector_num, nb_sectors);
@@ -999,7 +1056,7 @@ retry:
     }
 
     if (iTask.status != SCSI_STATUS_GOOD) {
-        return -EIO;
+        return iTask.err_code;
     }
 
     if (flags & BDRV_REQ_MAY_UNMAP) {
@@ -1280,9 +1337,8 @@ static void iscsi_detach_aio_context(BlockDriverState *bs)
 {
     IscsiLun *iscsilun = bs->opaque;
 
-    aio_set_fd_handler(iscsilun->aio_context,
-                       iscsi_get_fd(iscsilun->iscsi),
-                       NULL, NULL, NULL);
+    aio_set_fd_handler(iscsilun->aio_context, iscsi_get_fd(iscsilun->iscsi),
+                       false, NULL, NULL, NULL);
     iscsilun->events = 0;
 
     if (iscsilun->nop_timer) {
@@ -1772,7 +1828,6 @@ static BlockDriver bdrv_iscsi = {
     .bdrv_co_flush_to_disk = iscsi_co_flush,
 
 #ifdef __linux__
-    .bdrv_ioctl       = iscsi_ioctl,
     .bdrv_aio_ioctl   = iscsi_aio_ioctl,
 #endif
 
diff --git a/block/linux-aio.c b/block/linux-aio.c
index c991443c5..88b0520a8 100644
--- a/block/linux-aio.c
+++ b/block/linux-aio.c
@@ -287,7 +287,7 @@ void laio_detach_aio_context(void *s_, AioContext *old_context)
 {
     struct qemu_laio_state *s = s_;
 
-    aio_set_event_notifier(old_context, &s->e, NULL);
+    aio_set_event_notifier(old_context, &s->e, false, NULL);
     qemu_bh_delete(s->completion_bh);
 }
 
@@ -296,7 +296,8 @@ void laio_attach_aio_context(void *s_, AioContext *new_context)
     struct qemu_laio_state *s = s_;
 
     s->completion_bh = aio_bh_new(new_context, qemu_laio_completion_bh, s);
-    aio_set_event_notifier(new_context, &s->e, qemu_laio_completion_cb);
+    aio_set_event_notifier(new_context, &s->e, false,
+                           qemu_laio_completion_cb);
 }
 
 void *laio_init(void)
diff --git a/block/mirror.c b/block/mirror.c
index b2fb4b9b1..0e8f5565a 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -14,6 +14,7 @@
 #include "trace.h"
 #include "block/blockjob.h"
 #include "block/block_int.h"
+#include "sysemu/block-backend.h"
 #include "qapi/qmp/qerror.h"
 #include "qemu/ratelimit.h"
 #include "qemu/bitmap.h"
@@ -113,7 +114,7 @@ static void mirror_iteration_done(MirrorOp *op, int ret)
     }
 
     qemu_iovec_destroy(&op->qiov);
-    g_slice_free(MirrorOp, op);
+    g_free(op);
 
     if (s->waiting_for_io) {
         qemu_coroutine_enter(s->common.co, NULL);
@@ -246,6 +247,10 @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
             trace_mirror_break_buf_busy(s, nb_chunks, s->in_flight);
             break;
         }
+        if (IOV_MAX < nb_chunks + added_chunks) {
+            trace_mirror_break_iov_max(s, nb_chunks, added_chunks);
+            break;
+        }
 
         /* We have enough free space to copy these sectors.  */
         bitmap_set(s->in_flight_bitmap, next_chunk, added_chunks);
@@ -260,7 +265,7 @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
     } while (delay_ns == 0 && next_sector < end);
 
     /* Allocate a MirrorOp that is used as an AIO callback.  */
-    op = g_slice_new(MirrorOp);
+    op = g_new(MirrorOp, 1);
     op->s = s;
     op->sector_num = sector_num;
     op->nb_sectors = nb_sectors;
@@ -349,6 +354,11 @@ static void mirror_exit(BlockJob *job, void *opaque)
     MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
     MirrorExitData *data = opaque;
     AioContext *replace_aio_context = NULL;
+    BlockDriverState *src = s->common.bs;
+
+    /* Make sure that the source BDS doesn't go away before we called
+     * block_job_completed(). */
+    bdrv_ref(src);
 
     if (s->to_replace) {
         replace_aio_context = bdrv_get_aio_context(s->to_replace);
@@ -363,14 +373,7 @@ static void mirror_exit(BlockJob *job, void *opaque)
         if (bdrv_get_flags(s->target) != bdrv_get_flags(to_replace)) {
             bdrv_reopen(s->target, bdrv_get_flags(to_replace), NULL);
         }
-        bdrv_swap(s->target, to_replace);
-        if (s->common.driver->job_type == BLOCK_JOB_TYPE_COMMIT) {
-            /* drop the bs loop chain formed by the swap: break the loop then
-             * trigger the unref from the top one */
-            BlockDriverState *p = s->base->backing_hd;
-            bdrv_set_backing_hd(s->base, NULL);
-            bdrv_unref(p);
-        }
+        bdrv_replace_in_backing_chain(to_replace, s->target);
     }
     if (s->to_replace) {
         bdrv_op_unblock_all(s->to_replace, s->replace_blocker);
@@ -381,9 +384,12 @@ static void mirror_exit(BlockJob *job, void *opaque)
         aio_context_release(replace_aio_context);
     }
     g_free(s->replaces);
+    bdrv_op_unblock_all(s->target, s->common.blocker);
     bdrv_unref(s->target);
     block_job_completed(&s->common, data->ret);
     g_free(data);
+    bdrv_drained_end(src);
+    bdrv_unref(src);
 }
 
 static void coroutine_fn mirror_run(void *opaque)
@@ -427,7 +433,7 @@ static void coroutine_fn mirror_run(void *opaque)
      */
     bdrv_get_backing_filename(s->target, backing_filename,
                               sizeof(backing_filename));
-    if (backing_filename[0] && !s->target->backing_hd) {
+    if (backing_filename[0] && !s->target->backing) {
         ret = bdrv_get_info(s->target, &bdi);
         if (ret < 0) {
             goto immediate_exit;
@@ -451,6 +457,8 @@ static void coroutine_fn mirror_run(void *opaque)
     if (!s->is_none_mode) {
         /* First part, loop on the sectors and initialize the dirty bitmap.  */
         BlockDriverState *base = s->base;
+        bool mark_all_dirty = s->base == NULL && !bdrv_has_zero_init(s->target);
+
         for (sector_num = 0; sector_num < end; ) {
             /* Just to make sure we are not exceeding int limit. */
             int nb_sectors = MIN(INT_MAX >> BDRV_SECTOR_BITS,
@@ -473,7 +481,7 @@ static void coroutine_fn mirror_run(void *opaque)
             }
 
             assert(n > 0);
-            if (ret == 1) {
+            if (ret == 1 || mark_all_dirty) {
                 bdrv_set_dirty_bitmap(s->dirty_bitmap, sector_num, n);
             }
             sector_num += n;
@@ -594,10 +602,15 @@ immediate_exit:
     g_free(s->cow_bitmap);
     g_free(s->in_flight_bitmap);
     bdrv_release_dirty_bitmap(bs, s->dirty_bitmap);
-    bdrv_iostatus_disable(s->target);
+    if (s->target->blk) {
+        blk_iostatus_disable(s->target->blk);
+    }
 
     data = g_malloc(sizeof(*data));
     data->ret = ret;
+    /* Before we switch to target in mirror_exit, make sure data doesn't
+     * change. */
+    bdrv_drained_begin(s->common.bs);
     block_job_defer_to_main_loop(&s->common, mirror_exit, data);
 }
 
@@ -616,7 +629,9 @@ static void mirror_iostatus_reset(BlockJob *job)
 {
     MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
 
-    bdrv_iostatus_reset(s->target);
+    if (s->target->blk) {
+        blk_iostatus_reset(s->target->blk);
+    }
 }
 
 static void mirror_complete(BlockJob *job, Error **errp)
@@ -631,8 +646,7 @@ static void mirror_complete(BlockJob *job, Error **errp)
         return;
     }
     if (!s->synced) {
-        error_setg(errp, QERR_BLOCK_JOB_NOT_READY,
-                   bdrv_get_device_name(job->bs));
+        error_setg(errp, QERR_BLOCK_JOB_NOT_READY, job->id);
         return;
     }
 
@@ -640,9 +654,9 @@ static void mirror_complete(BlockJob *job, Error **errp)
     if (s->replaces) {
         AioContext *replace_aio_context;
 
-        s->to_replace = check_to_replace_node(s->replaces, &local_err);
+        s->to_replace = bdrv_find_node(s->replaces);
         if (!s->to_replace) {
-            error_propagate(errp, local_err);
+            error_setg(errp, "Node name '%s' not found", s->replaces);
             return;
         }
 
@@ -700,7 +714,7 @@ static void mirror_start_job(BlockDriverState *bs, BlockDriverState *target,
 
     if ((on_source_error == BLOCKDEV_ON_ERROR_STOP ||
          on_source_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
-        !bdrv_iostatus_is_enabled(bs)) {
+        (!bs->blk || !blk_iostatus_is_enabled(bs->blk))) {
         error_setg(errp, QERR_INVALID_PARAMETER, "on-source-error");
         return;
     }
@@ -732,12 +746,17 @@ static void mirror_start_job(BlockDriverState *bs, BlockDriverState *target,
     s->dirty_bitmap = bdrv_create_dirty_bitmap(bs, granularity, NULL, errp);
     if (!s->dirty_bitmap) {
         g_free(s->replaces);
-        block_job_release(bs);
+        block_job_unref(&s->common);
         return;
     }
+
+    bdrv_op_block_all(s->target, s->common.blocker);
+
     bdrv_set_enable_write_cache(s->target, true);
-    bdrv_set_on_error(s->target, on_target_error, on_target_error);
-    bdrv_iostatus_enable(s->target);
+    if (s->target->blk) {
+        blk_set_on_error(s->target->blk, on_target_error, on_target_error);
+        blk_iostatus_enable(s->target->blk);
+    }
     s->common.co = qemu_coroutine_create(mirror_run);
     trace_mirror_start(bs, s, s->common.co, opaque);
     qemu_coroutine_enter(s->common.co, s);
@@ -760,7 +779,7 @@ void mirror_start(BlockDriverState *bs, BlockDriverState *target,
         return;
     }
     is_none_mode = mode == MIRROR_SYNC_MODE_NONE;
-    base = mode == MIRROR_SYNC_MODE_TOP ? bs->backing_hd : NULL;
+    base = mode == MIRROR_SYNC_MODE_TOP ? backing_bs(bs) : NULL;
     mirror_start_job(bs, target, replaces,
                      speed, granularity, buf_size,
                      on_source_error, on_target_error, unmap, cb, opaque, errp,
diff --git a/block/nbd-client.c b/block/nbd-client.c
index e1bb9198c..b7fd17a11 100644
--- a/block/nbd-client.c
+++ b/block/nbd-client.c
@@ -124,7 +124,7 @@ static int nbd_co_send_request(BlockDriverState *bs,
     s->send_coroutine = qemu_coroutine_self();
     aio_context = bdrv_get_aio_context(bs);
 
-    aio_set_fd_handler(aio_context, s->sock,
+    aio_set_fd_handler(aio_context, s->sock, false,
                        nbd_reply_ready, nbd_restart_write, bs);
     if (qiov) {
         if (!s->is_unix) {
@@ -144,7 +144,8 @@ static int nbd_co_send_request(BlockDriverState *bs,
     } else {
         rc = nbd_send_request(s->sock, request);
     }
-    aio_set_fd_handler(aio_context, s->sock, nbd_reply_ready, NULL, bs);
+    aio_set_fd_handler(aio_context, s->sock, false,
+                       nbd_reply_ready, NULL, bs);
     s->send_coroutine = NULL;
     qemu_co_mutex_unlock(&s->send_mutex);
     return rc;
@@ -348,14 +349,15 @@ int nbd_client_co_discard(BlockDriverState *bs, int64_t sector_num,
 void nbd_client_detach_aio_context(BlockDriverState *bs)
 {
     aio_set_fd_handler(bdrv_get_aio_context(bs),
-                       nbd_get_client_session(bs)->sock, NULL, NULL, NULL);
+                       nbd_get_client_session(bs)->sock,
+                       false, NULL, NULL, NULL);
 }
 
 void nbd_client_attach_aio_context(BlockDriverState *bs,
                                    AioContext *new_context)
 {
     aio_set_fd_handler(new_context, nbd_get_client_session(bs)->sock,
-                       nbd_reply_ready, NULL, bs);
+                       false, nbd_reply_ready, NULL, bs);
 }
 
 void nbd_client_close(BlockDriverState *bs)
diff --git a/block/nbd.c b/block/nbd.c
index 217618612..cd6a58777 100644
--- a/block/nbd.c
+++ b/block/nbd.c
@@ -43,7 +43,6 @@
 
 typedef struct BDRVNBDState {
     NbdClientSession client;
-    QemuOpts *socket_opts;
 } BDRVNBDState;
 
 static int nbd_parse_uri(const char *filename, QDict *options)
@@ -190,10 +189,10 @@ out:
     g_free(file);
 }
 
-static void nbd_config(BDRVNBDState *s, QDict *options, char **export,
-                       Error **errp)
+static SocketAddress *nbd_config(BDRVNBDState *s, QDict *options, char **export,
+                                 Error **errp)
 {
-    Error *local_err = NULL;
+    SocketAddress *saddr;
 
     if (qdict_haskey(options, "path") == qdict_haskey(options, "host")) {
         if (qdict_haskey(options, "path")) {
@@ -201,28 +200,37 @@ static void nbd_config(BDRVNBDState *s, QDict *options, char **export,
         } else {
             error_setg(errp, "one of path and host must be specified.");
         }
-        return;
+        return NULL;
     }
 
-    s->client.is_unix = qdict_haskey(options, "path");
-    s->socket_opts = qemu_opts_create(&socket_optslist, NULL, 0,
-                                      &error_abort);
+    saddr = g_new0(SocketAddress, 1);
 
-    qemu_opts_absorb_qdict(s->socket_opts, options, &local_err);
-    if (local_err) {
-        error_propagate(errp, local_err);
-        return;
+    if (qdict_haskey(options, "path")) {
+        saddr->type = SOCKET_ADDRESS_KIND_UNIX;
+        saddr->u.q_unix = g_new0(UnixSocketAddress, 1);
+        saddr->u.q_unix->path = g_strdup(qdict_get_str(options, "path"));
+        qdict_del(options, "path");
+    } else {
+        saddr->type = SOCKET_ADDRESS_KIND_INET;
+        saddr->u.inet = g_new0(InetSocketAddress, 1);
+        saddr->u.inet->host = g_strdup(qdict_get_str(options, "host"));
+        if (!qdict_get_try_str(options, "port")) {
+            saddr->u.inet->port = g_strdup_printf("%d", NBD_DEFAULT_PORT);
+        } else {
+            saddr->u.inet->port = g_strdup(qdict_get_str(options, "port"));
+        }
+        qdict_del(options, "host");
+        qdict_del(options, "port");
     }
 
-    if (!qemu_opt_get(s->socket_opts, "port")) {
-        qemu_opt_set_number(s->socket_opts, "port", NBD_DEFAULT_PORT,
-                            &error_abort);
-    }
+    s->client.is_unix = saddr->type == SOCKET_ADDRESS_KIND_UNIX;
 
     *export = g_strdup(qdict_get_try_str(options, "export"));
     if (*export) {
         qdict_del(options, "export");
     }
+
+    return saddr;
 }
 
 NbdClientSession *nbd_get_client_session(BlockDriverState *bs)
@@ -231,26 +239,24 @@ NbdClientSession *nbd_get_client_session(BlockDriverState *bs)
     return &s->client;
 }
 
-static int nbd_establish_connection(BlockDriverState *bs, Error **errp)
+static int nbd_establish_connection(BlockDriverState *bs,
+                                    SocketAddress *saddr,
+                                    Error **errp)
 {
     BDRVNBDState *s = bs->opaque;
     int sock;
 
-    if (s->client.is_unix) {
-        sock = unix_connect_opts(s->socket_opts, errp, NULL, NULL);
-    } else {
-        sock = inet_connect_opts(s->socket_opts, errp, NULL, NULL);
-        if (sock >= 0) {
-            socket_set_nodelay(sock);
-        }
-    }
+    sock = socket_connect(saddr, errp, NULL, NULL);
 
-    /* Failed to establish connection */
     if (sock < 0) {
         logout("Failed to establish connection to NBD server\n");
         return -EIO;
     }
 
+    if (!s->client.is_unix) {
+        socket_set_nodelay(sock);
+    }
+
     return sock;
 }
 
@@ -260,19 +266,19 @@ static int nbd_open(BlockDriverState *bs, QDict *options, int flags,
     BDRVNBDState *s = bs->opaque;
     char *export = NULL;
     int result, sock;
-    Error *local_err = NULL;
+    SocketAddress *saddr;
 
     /* Pop the config into our state object. Exit if invalid. */
-    nbd_config(s, options, &export, &local_err);
-    if (local_err) {
-        error_propagate(errp, local_err);
+    saddr = nbd_config(s, options, &export, errp);
+    if (!saddr) {
         return -EINVAL;
     }
 
     /* establish TCP connection, return error if it fails
      * TODO: Configurable retry-until-timeout behaviour.
      */
-    sock = nbd_establish_connection(bs, errp);
+    sock = nbd_establish_connection(bs, saddr, errp);
+    qapi_free_SocketAddress(saddr);
     if (sock < 0) {
         g_free(export);
         return sock;
@@ -315,9 +321,6 @@ static int nbd_co_discard(BlockDriverState *bs, int64_t sector_num,
 
 static void nbd_close(BlockDriverState *bs)
 {
-    BDRVNBDState *s = bs->opaque;
-
-    qemu_opts_del(s->socket_opts);
     nbd_client_close(bs);
 }
 
diff --git a/block/nfs.c b/block/nfs.c
index 02eb4e464..fd79f8994 100644
--- a/block/nfs.c
+++ b/block/nfs.c
@@ -43,6 +43,7 @@ typedef struct NFSClient {
     int events;
     bool has_zero_init;
     AioContext *aio_context;
+    blkcnt_t st_blocks;
 } NFSClient;
 
 typedef struct NFSRPC {
@@ -62,11 +63,10 @@ static void nfs_set_events(NFSClient *client)
 {
     int ev = nfs_which_events(client->context);
     if (ev != client->events) {
-        aio_set_fd_handler(client->aio_context,
-                           nfs_get_fd(client->context),
+        aio_set_fd_handler(client->aio_context, nfs_get_fd(client->context),
+                           false,
                            (ev & POLLIN) ? nfs_process_read : NULL,
-                           (ev & POLLOUT) ? nfs_process_write : NULL,
-                           client);
+                           (ev & POLLOUT) ? nfs_process_write : NULL, client);
 
     }
     client->events = ev;
@@ -241,9 +241,8 @@ static void nfs_detach_aio_context(BlockDriverState *bs)
 {
     NFSClient *client = bs->opaque;
 
-    aio_set_fd_handler(client->aio_context,
-                       nfs_get_fd(client->context),
-                       NULL, NULL, NULL);
+    aio_set_fd_handler(client->aio_context, nfs_get_fd(client->context),
+                       false, NULL, NULL, NULL);
     client->events = 0;
 }
 
@@ -262,9 +261,8 @@ static void nfs_client_close(NFSClient *client)
         if (client->fh) {
             nfs_close(client->context, client->fh);
         }
-        aio_set_fd_handler(client->aio_context,
-                           nfs_get_fd(client->context),
-                           NULL, NULL, NULL);
+        aio_set_fd_handler(client->aio_context, nfs_get_fd(client->context),
+                           false, NULL, NULL, NULL);
         nfs_destroy_context(client->context);
     }
     memset(client, 0, sizeof(NFSClient));
@@ -374,6 +372,7 @@ static int64_t nfs_client_open(NFSClient *client, const char *filename,
     }
 
     ret = DIV_ROUND_UP(st.st_size, BDRV_SECTOR_SIZE);
+    client->st_blocks = st.st_blocks;
     client->has_zero_init = S_ISREG(st.st_mode);
     goto out;
 fail:
@@ -464,6 +463,11 @@ static int64_t nfs_get_allocated_file_size(BlockDriverState *bs)
     NFSRPC task = {0};
     struct stat st;
 
+    if (bdrv_is_read_only(bs) &&
+        !(bs->open_flags & BDRV_O_NOCACHE)) {
+        return client->st_blocks * 512;
+    }
+
     task.st = &st;
     if (nfs_fstat_async(client->context, client->fh, nfs_co_generic_cb,
                         &task) != 0) {
@@ -484,6 +488,34 @@ static int nfs_file_truncate(BlockDriverState *bs, int64_t offset)
     return nfs_ftruncate(client->context, client->fh, offset);
 }
 
+/* Note that this will not re-establish a connection with the NFS server
+ * - it is effectively a NOP.  */
+static int nfs_reopen_prepare(BDRVReopenState *state,
+                              BlockReopenQueue *queue, Error **errp)
+{
+    NFSClient *client = state->bs->opaque;
+    struct stat st;
+    int ret = 0;
+
+    if (state->flags & BDRV_O_RDWR && bdrv_is_read_only(state->bs)) {
+        error_setg(errp, "Cannot open a read-only mount as read-write");
+        return -EACCES;
+    }
+
+    /* Update cache for read-only reopens */
+    if (!(state->flags & BDRV_O_RDWR)) {
+        ret = nfs_fstat(client->context, client->fh, &st);
+        if (ret < 0) {
+            error_setg(errp, "Failed to fstat file: %s",
+                       nfs_get_error(client->context));
+            return ret;
+        }
+        client->st_blocks = st.st_blocks;
+    }
+
+    return 0;
+}
+
 static BlockDriver bdrv_nfs = {
     .format_name                    = "nfs",
     .protocol_name                  = "nfs",
@@ -499,6 +531,7 @@ static BlockDriver bdrv_nfs = {
     .bdrv_file_open                 = nfs_file_open,
     .bdrv_close                     = nfs_file_close,
     .bdrv_create                    = nfs_file_create,
+    .bdrv_reopen_prepare            = nfs_reopen_prepare,
 
     .bdrv_co_readv                  = nfs_co_readv,
     .bdrv_co_writev                 = nfs_co_writev,
diff --git a/block/parallels.c b/block/parallels.c
index 046b56844..f689fdeaf 100644
--- a/block/parallels.c
+++ b/block/parallels.c
@@ -202,13 +202,13 @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
 
     to_allocate = (sector_num + *pnum + s->tracks - 1) / s->tracks - idx;
     space = to_allocate * s->tracks;
-    if (s->data_end + space > bdrv_getlength(bs->file) >> BDRV_SECTOR_BITS) {
+    if (s->data_end + space > bdrv_getlength(bs->file->bs) >> BDRV_SECTOR_BITS) {
         int ret;
         space += s->prealloc_size;
         if (s->prealloc_mode == PRL_PREALLOC_MODE_FALLOCATE) {
-            ret = bdrv_write_zeroes(bs->file, s->data_end, space, 0);
+            ret = bdrv_write_zeroes(bs->file->bs, s->data_end, space, 0);
         } else {
-            ret = bdrv_truncate(bs->file,
+            ret = bdrv_truncate(bs->file->bs,
                                 (s->data_end + space) << BDRV_SECTOR_BITS);
         }
         if (ret < 0) {
@@ -220,7 +220,7 @@ static int64_t allocate_clusters(BlockDriverState *bs, int64_t sector_num,
         s->bat_bitmap[idx + i] = cpu_to_le32(s->data_end / s->off_multiplier);
         s->data_end += s->tracks;
         bitmap_set(s->bat_dirty_bmap,
-                   bat_entry_off(idx) / s->bat_dirty_block, 1);
+                   bat_entry_off(idx + i) / s->bat_dirty_block, 1);
     }
 
     return bat2sect(s, idx) + sector_num % s->tracks;
@@ -244,7 +244,8 @@ static coroutine_fn int parallels_co_flush_to_os(BlockDriverState *bs)
         if (off + to_write > s->header_size) {
             to_write = s->header_size - off;
         }
-        ret = bdrv_pwrite(bs->file, off, (uint8_t *)s->header + off, to_write);
+        ret = bdrv_pwrite(bs->file->bs, off, (uint8_t *)s->header + off,
+                          to_write);
         if (ret < 0) {
             qemu_co_mutex_unlock(&s->lock);
             return ret;
@@ -303,7 +304,7 @@ static coroutine_fn int parallels_co_writev(BlockDriverState *bs,
         qemu_iovec_reset(&hd_qiov);
         qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes);
 
-        ret = bdrv_co_writev(bs->file, position, n, &hd_qiov);
+        ret = bdrv_co_writev(bs->file->bs, position, n, &hd_qiov);
         if (ret < 0) {
             break;
         }
@@ -343,7 +344,7 @@ static coroutine_fn int parallels_co_readv(BlockDriverState *bs,
             qemu_iovec_reset(&hd_qiov);
             qemu_iovec_concat(&hd_qiov, qiov, bytes_done, nbytes);
 
-            ret = bdrv_co_readv(bs->file, position, n, &hd_qiov);
+            ret = bdrv_co_readv(bs->file->bs, position, n, &hd_qiov);
             if (ret < 0) {
                 break;
             }
@@ -369,7 +370,7 @@ static int parallels_check(BlockDriverState *bs, BdrvCheckResult *res,
     bool flush_bat = false;
     int cluster_size = s->tracks << BDRV_SECTOR_BITS;
 
-    size = bdrv_getlength(bs->file);
+    size = bdrv_getlength(bs->file->bs);
     if (size < 0) {
         res->check_errors++;
         return size;
@@ -424,7 +425,7 @@ static int parallels_check(BlockDriverState *bs, BdrvCheckResult *res,
     }
 
     if (flush_bat) {
-        ret = bdrv_pwrite_sync(bs->file, 0, s->header, s->header_size);
+        ret = bdrv_pwrite_sync(bs->file->bs, 0, s->header, s->header_size);
         if (ret < 0) {
             res->check_errors++;
             return ret;
@@ -440,7 +441,7 @@ static int parallels_check(BlockDriverState *bs, BdrvCheckResult *res,
                 size - res->image_end_offset);
         res->leaks += count;
         if (fix & BDRV_FIX_LEAKS) {
-            ret = bdrv_truncate(bs->file, res->image_end_offset);
+            ret = bdrv_truncate(bs->file->bs, res->image_end_offset);
             if (ret < 0) {
                 res->check_errors++;
                 return ret;
@@ -476,7 +477,7 @@ static int parallels_create(const char *filename, QemuOpts *opts, Error **errp)
 
     file = NULL;
     ret = bdrv_open(&file, filename, NULL, NULL,
-                    BDRV_O_RDWR | BDRV_O_PROTOCOL, NULL, &local_err);
+                    BDRV_O_RDWR | BDRV_O_PROTOCOL, &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         return ret;
@@ -546,12 +547,13 @@ static int parallels_probe(const uint8_t *buf, int buf_size,
 static int parallels_update_header(BlockDriverState *bs)
 {
     BDRVParallelsState *s = bs->opaque;
-    unsigned size = MAX(bdrv_opt_mem_align(bs->file), sizeof(ParallelsHeader));
+    unsigned size = MAX(bdrv_opt_mem_align(bs->file->bs),
+                        sizeof(ParallelsHeader));
 
     if (size > s->header_size) {
         size = s->header_size;
     }
-    return bdrv_pwrite_sync(bs->file, 0, s->header, size);
+    return bdrv_pwrite_sync(bs->file->bs, 0, s->header, size);
 }
 
 static int parallels_open(BlockDriverState *bs, QDict *options, int flags,
@@ -564,7 +566,7 @@ static int parallels_open(BlockDriverState *bs, QDict *options, int flags,
     Error *local_err = NULL;
     char *buf;
 
-    ret = bdrv_pread(bs->file, 0, &ph, sizeof(ph));
+    ret = bdrv_pread(bs->file->bs, 0, &ph, sizeof(ph));
     if (ret < 0) {
         goto fail;
     }
@@ -603,8 +605,8 @@ static int parallels_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     size = bat_entry_off(s->bat_size);
-    s->header_size = ROUND_UP(size, bdrv_opt_mem_align(bs->file));
-    s->header = qemu_try_blockalign(bs->file, s->header_size);
+    s->header_size = ROUND_UP(size, bdrv_opt_mem_align(bs->file->bs));
+    s->header = qemu_try_blockalign(bs->file->bs, s->header_size);
     if (s->header == NULL) {
         ret = -ENOMEM;
         goto fail;
@@ -619,7 +621,7 @@ static int parallels_open(BlockDriverState *bs, QDict *options, int flags,
         s->header_size = size;
     }
 
-    ret = bdrv_pread(bs->file, 0, s->header, s->header_size);
+    ret = bdrv_pread(bs->file->bs, 0, s->header, s->header_size);
     if (ret < 0) {
         goto fail;
     }
@@ -663,8 +665,8 @@ static int parallels_open(BlockDriverState *bs, QDict *options, int flags,
     if (local_err != NULL) {
         goto fail_options;
     }
-    if (!bdrv_has_zero_init(bs->file) ||
-            bdrv_truncate(bs->file, bdrv_getlength(bs->file)) != 0) {
+    if (!bdrv_has_zero_init(bs->file->bs) ||
+            bdrv_truncate(bs->file->bs, bdrv_getlength(bs->file->bs)) != 0) {
         s->prealloc_mode = PRL_PREALLOC_MODE_FALLOCATE;
     }
 
@@ -707,7 +709,7 @@ static void parallels_close(BlockDriverState *bs)
     }
 
     if (bs->open_flags & BDRV_O_RDWR) {
-        bdrv_truncate(bs->file, s->data_end << BDRV_SECTOR_BITS);
+        bdrv_truncate(bs->file->bs, s->data_end << BDRV_SECTOR_BITS);
     }
 
     g_free(s->bat_dirty_bmap);
diff --git a/block/qapi.c b/block/qapi.c
index 2ce509711..267f147fe 100644
--- a/block/qapi.c
+++ b/block/qapi.c
@@ -64,7 +64,7 @@ BlockDeviceInfo *bdrv_block_device_info(BlockDriverState *bs, Error **errp)
     info->backing_file_depth = bdrv_get_backing_file_depth(bs);
     info->detect_zeroes = bs->detect_zeroes;
 
-    if (bs->io_limits_enabled) {
+    if (bs->throttle_state) {
         ThrottleConfig cfg;
 
         throttle_group_get_config(bs, &cfg);
@@ -110,8 +110,8 @@ BlockDeviceInfo *bdrv_block_device_info(BlockDriverState *bs, Error **errp)
             qapi_free_BlockDeviceInfo(info);
             return NULL;
         }
-        if (bs0->drv && bs0->backing_hd) {
-            bs0 = bs0->backing_hd;
+        if (bs0->drv && bs0->backing) {
+            bs0 = bs0->backing->bs;
             (*p_image_info)->has_backing_image = true;
             p_image_info = &((*p_image_info)->backing_image);
         } else {
@@ -301,17 +301,17 @@ static void bdrv_query_info(BlockBackend *blk, BlockInfo **p_info,
         info->tray_open = blk_dev_is_tray_open(blk);
     }
 
-    if (bdrv_iostatus_is_enabled(bs)) {
+    if (blk_iostatus_is_enabled(blk)) {
         info->has_io_status = true;
-        info->io_status = bs->iostatus;
+        info->io_status = blk_iostatus(blk);
     }
 
-    if (!QLIST_EMPTY(&bs->dirty_bitmaps)) {
+    if (bs && !QLIST_EMPTY(&bs->dirty_bitmaps)) {
         info->has_dirty_bitmaps = true;
         info->dirty_bitmaps = bdrv_query_dirty_bitmaps(bs);
     }
 
-    if (bs->drv) {
+    if (bs && bs->drv) {
         info->has_inserted = true;
         info->inserted = bdrv_block_device_info(bs, errp);
         if (info->inserted == NULL) {
@@ -344,27 +344,82 @@ static BlockStats *bdrv_query_stats(const BlockDriverState *bs,
     }
 
     s->stats = g_malloc0(sizeof(*s->stats));
-    s->stats->rd_bytes = bs->stats.nr_bytes[BLOCK_ACCT_READ];
-    s->stats->wr_bytes = bs->stats.nr_bytes[BLOCK_ACCT_WRITE];
-    s->stats->rd_operations = bs->stats.nr_ops[BLOCK_ACCT_READ];
-    s->stats->wr_operations = bs->stats.nr_ops[BLOCK_ACCT_WRITE];
-    s->stats->rd_merged = bs->stats.merged[BLOCK_ACCT_READ];
-    s->stats->wr_merged = bs->stats.merged[BLOCK_ACCT_WRITE];
-    s->stats->wr_highest_offset =
-        bs->stats.wr_highest_sector * BDRV_SECTOR_SIZE;
-    s->stats->flush_operations = bs->stats.nr_ops[BLOCK_ACCT_FLUSH];
-    s->stats->wr_total_time_ns = bs->stats.total_time_ns[BLOCK_ACCT_WRITE];
-    s->stats->rd_total_time_ns = bs->stats.total_time_ns[BLOCK_ACCT_READ];
-    s->stats->flush_total_time_ns = bs->stats.total_time_ns[BLOCK_ACCT_FLUSH];
+    if (bs->blk) {
+        BlockAcctStats *stats = blk_get_stats(bs->blk);
+        BlockAcctTimedStats *ts = NULL;
+
+        s->stats->rd_bytes = stats->nr_bytes[BLOCK_ACCT_READ];
+        s->stats->wr_bytes = stats->nr_bytes[BLOCK_ACCT_WRITE];
+        s->stats->rd_operations = stats->nr_ops[BLOCK_ACCT_READ];
+        s->stats->wr_operations = stats->nr_ops[BLOCK_ACCT_WRITE];
+
+        s->stats->failed_rd_operations = stats->failed_ops[BLOCK_ACCT_READ];
+        s->stats->failed_wr_operations = stats->failed_ops[BLOCK_ACCT_WRITE];
+        s->stats->failed_flush_operations = stats->failed_ops[BLOCK_ACCT_FLUSH];
+
+        s->stats->invalid_rd_operations = stats->invalid_ops[BLOCK_ACCT_READ];
+        s->stats->invalid_wr_operations = stats->invalid_ops[BLOCK_ACCT_WRITE];
+        s->stats->invalid_flush_operations =
+            stats->invalid_ops[BLOCK_ACCT_FLUSH];
+
+        s->stats->rd_merged = stats->merged[BLOCK_ACCT_READ];
+        s->stats->wr_merged = stats->merged[BLOCK_ACCT_WRITE];
+        s->stats->flush_operations = stats->nr_ops[BLOCK_ACCT_FLUSH];
+        s->stats->wr_total_time_ns = stats->total_time_ns[BLOCK_ACCT_WRITE];
+        s->stats->rd_total_time_ns = stats->total_time_ns[BLOCK_ACCT_READ];
+        s->stats->flush_total_time_ns = stats->total_time_ns[BLOCK_ACCT_FLUSH];
+
+        s->stats->has_idle_time_ns = stats->last_access_time_ns > 0;
+        if (s->stats->has_idle_time_ns) {
+            s->stats->idle_time_ns = block_acct_idle_time_ns(stats);
+        }
+
+        s->stats->account_invalid = stats->account_invalid;
+        s->stats->account_failed = stats->account_failed;
+
+        while ((ts = block_acct_interval_next(stats, ts))) {
+            BlockDeviceTimedStatsList *timed_stats =
+                g_malloc0(sizeof(*timed_stats));
+            BlockDeviceTimedStats *dev_stats = g_malloc0(sizeof(*dev_stats));
+            timed_stats->next = s->stats->timed_stats;
+            timed_stats->value = dev_stats;
+            s->stats->timed_stats = timed_stats;
+
+            TimedAverage *rd = &ts->latency[BLOCK_ACCT_READ];
+            TimedAverage *wr = &ts->latency[BLOCK_ACCT_WRITE];
+            TimedAverage *fl = &ts->latency[BLOCK_ACCT_FLUSH];
+
+            dev_stats->interval_length = ts->interval_length;
+
+            dev_stats->min_rd_latency_ns = timed_average_min(rd);
+            dev_stats->max_rd_latency_ns = timed_average_max(rd);
+            dev_stats->avg_rd_latency_ns = timed_average_avg(rd);
+
+            dev_stats->min_wr_latency_ns = timed_average_min(wr);
+            dev_stats->max_wr_latency_ns = timed_average_max(wr);
+            dev_stats->avg_wr_latency_ns = timed_average_avg(wr);
+
+            dev_stats->min_flush_latency_ns = timed_average_min(fl);
+            dev_stats->max_flush_latency_ns = timed_average_max(fl);
+            dev_stats->avg_flush_latency_ns = timed_average_avg(fl);
+
+            dev_stats->avg_rd_queue_depth =
+                block_acct_queue_depth(ts, BLOCK_ACCT_READ);
+            dev_stats->avg_wr_queue_depth =
+                block_acct_queue_depth(ts, BLOCK_ACCT_WRITE);
+        }
+    }
+
+    s->stats->wr_highest_offset = bs->wr_highest_offset;
 
     if (bs->file) {
         s->has_parent = true;
-        s->parent = bdrv_query_stats(bs->file, query_backing);
+        s->parent = bdrv_query_stats(bs->file->bs, query_backing);
     }
 
-    if (query_backing && bs->backing_hd) {
+    if (query_backing && bs->backing) {
         s->has_backing = true;
-        s->backing = bdrv_query_stats(bs->backing_hd, query_backing);
+        s->backing = bdrv_query_stats(bs->backing->bs, query_backing);
     }
 
     return s;
@@ -381,7 +436,9 @@ BlockInfoList *qmp_query_block(Error **errp)
         bdrv_query_info(blk, &info->value, &local_err);
         if (local_err) {
             error_propagate(errp, local_err);
-            goto err;
+            g_free(info);
+            qapi_free_BlockInfoList(head);
+            return NULL;
         }
 
         *p_next = info;
@@ -389,10 +446,6 @@ BlockInfoList *qmp_query_block(Error **errp)
     }
 
     return head;
-
- err:
-    qapi_free_BlockInfoList(head);
-    return NULL;
 }
 
 BlockStatsList *qmp_query_blockstats(bool has_query_nodes,
diff --git a/block/qcow.c b/block/qcow.c
index 01fba54ce..635085e27 100644
--- a/block/qcow.c
+++ b/block/qcow.c
@@ -100,7 +100,7 @@ static int qcow_open(BlockDriverState *bs, QDict *options, int flags,
     int ret;
     QCowHeader header;
 
-    ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
+    ret = bdrv_pread(bs->file->bs, 0, &header, sizeof(header));
     if (ret < 0) {
         goto fail;
     }
@@ -193,7 +193,7 @@ static int qcow_open(BlockDriverState *bs, QDict *options, int flags,
         goto fail;
     }
 
-    ret = bdrv_pread(bs->file, s->l1_table_offset, s->l1_table,
+    ret = bdrv_pread(bs->file->bs, s->l1_table_offset, s->l1_table,
                s->l1_size * sizeof(uint64_t));
     if (ret < 0) {
         goto fail;
@@ -205,7 +205,7 @@ static int qcow_open(BlockDriverState *bs, QDict *options, int flags,
 
     /* alloc L2 cache (max. 64k * 16 * 8 = 8 MB) */
     s->l2_cache =
-        qemu_try_blockalign(bs->file,
+        qemu_try_blockalign(bs->file->bs,
                             s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));
     if (s->l2_cache == NULL) {
         error_setg(errp, "Could not allocate L2 table cache");
@@ -224,7 +224,7 @@ static int qcow_open(BlockDriverState *bs, QDict *options, int flags,
             ret = -EINVAL;
             goto fail;
         }
-        ret = bdrv_pread(bs->file, header.backing_file_offset,
+        ret = bdrv_pread(bs->file->bs, header.backing_file_offset,
                    bs->backing_file, len);
         if (ret < 0) {
             goto fail;
@@ -369,13 +369,13 @@ static uint64_t get_cluster_offset(BlockDriverState *bs,
         if (!allocate)
             return 0;
         /* allocate a new l2 entry */
-        l2_offset = bdrv_getlength(bs->file);
+        l2_offset = bdrv_getlength(bs->file->bs);
         /* 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,
+        if (bdrv_pwrite_sync(bs->file->bs,
                 s->l1_table_offset + l1_index * sizeof(tmp),
                 &tmp, sizeof(tmp)) < 0)
             return 0;
@@ -405,11 +405,12 @@ static uint64_t get_cluster_offset(BlockDriverState *bs,
     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,
+        if (bdrv_pwrite_sync(bs->file->bs, 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)) !=
+        if (bdrv_pread(bs->file->bs, l2_offset, l2_table,
+                       s->l2_size * sizeof(uint64_t)) !=
             s->l2_size * sizeof(uint64_t))
             return 0;
     }
@@ -430,20 +431,21 @@ static uint64_t get_cluster_offset(BlockDriverState *bs,
                overwritten */
             if (decompress_cluster(bs, cluster_offset) < 0)
                 return 0;
-            cluster_offset = bdrv_getlength(bs->file);
+            cluster_offset = bdrv_getlength(bs->file->bs);
             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) !=
+            if (bdrv_pwrite(bs->file->bs, cluster_offset, s->cluster_cache,
+                            s->cluster_size) !=
                 s->cluster_size)
                 return -1;
         } else {
-            cluster_offset = bdrv_getlength(bs->file);
+            cluster_offset = bdrv_getlength(bs->file->bs);
             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);
+                bdrv_truncate(bs->file->bs, cluster_offset + s->cluster_size);
                 /* if encrypted, we must initialize the cluster
                    content which won't be written */
                 if (bs->encrypted &&
@@ -463,7 +465,8 @@ static uint64_t get_cluster_offset(BlockDriverState *bs,
                                 errno = EIO;
                                 return -1;
                             }
-                            if (bdrv_pwrite(bs->file, cluster_offset + i * 512,
+                            if (bdrv_pwrite(bs->file->bs,
+                                            cluster_offset + i * 512,
                                             s->cluster_data, 512) != 512)
                                 return -1;
                         }
@@ -477,7 +480,7 @@ static uint64_t get_cluster_offset(BlockDriverState *bs,
         /* 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),
+        if (bdrv_pwrite_sync(bs->file->bs, l2_offset + l2_index * sizeof(tmp),
                 &tmp, sizeof(tmp)) < 0)
             return 0;
     }
@@ -546,7 +549,7 @@ static int decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
     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);
+        ret = bdrv_pread(bs->file->bs, coffset, s->cluster_data, csize);
         if (ret != csize)
             return -1;
         if (decompress_buffer(s->cluster_cache, s->cluster_size,
@@ -594,13 +597,13 @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num,
         }
 
         if (!cluster_offset) {
-            if (bs->backing_hd) {
+            if (bs->backing) {
                 /* 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,
+                ret = bdrv_co_readv(bs->backing->bs, sector_num,
                                     n, &hd_qiov);
                 qemu_co_mutex_lock(&s->lock);
                 if (ret < 0) {
@@ -625,7 +628,7 @@ static coroutine_fn int qcow_co_readv(BlockDriverState *bs, int64_t sector_num,
             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,
+            ret = bdrv_co_readv(bs->file->bs,
                                 (cluster_offset >> 9) + index_in_cluster,
                                 n, &hd_qiov);
             qemu_co_mutex_lock(&s->lock);
@@ -727,7 +730,7 @@ static coroutine_fn int qcow_co_writev(BlockDriverState *bs, int64_t sector_num,
         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,
+        ret = bdrv_co_writev(bs->file->bs,
                              (cluster_offset >> 9) + index_in_cluster,
                              n, &hd_qiov);
         qemu_co_mutex_lock(&s->lock);
@@ -793,7 +796,7 @@ static int qcow_create(const char *filename, QemuOpts *opts, Error **errp)
 
     qcow_bs = NULL;
     ret = bdrv_open(&qcow_bs, filename, NULL, NULL,
-                    BDRV_O_RDWR | BDRV_O_PROTOCOL, NULL, &local_err);
+                    BDRV_O_RDWR | BDRV_O_PROTOCOL, &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto cleanup;
@@ -879,10 +882,10 @@ static int qcow_make_empty(BlockDriverState *bs)
     int ret;
 
     memset(s->l1_table, 0, l1_length);
-    if (bdrv_pwrite_sync(bs->file, s->l1_table_offset, s->l1_table,
+    if (bdrv_pwrite_sync(bs->file->bs, s->l1_table_offset, s->l1_table,
             l1_length) < 0)
         return -1;
-    ret = bdrv_truncate(bs->file, s->l1_table_offset + l1_length);
+    ret = bdrv_truncate(bs->file->bs, s->l1_table_offset + l1_length);
     if (ret < 0)
         return ret;
 
@@ -962,7 +965,7 @@ static int qcow_write_compressed(BlockDriverState *bs, int64_t sector_num,
         }
 
         cluster_offset &= s->cluster_offset_mask;
-        ret = bdrv_pwrite(bs->file, cluster_offset, out_buf, out_len);
+        ret = bdrv_pwrite(bs->file->bs, cluster_offset, out_buf, out_len);
         if (ret < 0) {
             goto fail;
         }
diff --git a/block/qcow2-cache.c b/block/qcow2-cache.c
index 53b8afc3d..86dd7f2bd 100644
--- a/block/qcow2-cache.c
+++ b/block/qcow2-cache.c
@@ -22,16 +22,24 @@
  * THE SOFTWARE.
  */
 
+/* Needed for CONFIG_MADVISE */
+#include "config-host.h"
+
+#if defined(CONFIG_MADVISE) || defined(CONFIG_POSIX_MADVISE)
+#include <sys/mman.h>
+#endif
+
 #include "block/block_int.h"
 #include "qemu-common.h"
+#include "qemu/osdep.h"
 #include "qcow2.h"
 #include "trace.h"
 
 typedef struct Qcow2CachedTable {
     int64_t  offset;
-    bool     dirty;
     uint64_t lru_counter;
     int      ref;
+    bool     dirty;
 } Qcow2CachedTable;
 
 struct Qcow2Cache {
@@ -41,34 +49,85 @@ struct Qcow2Cache {
     bool                    depends_on_flush;
     void                   *table_array;
     uint64_t                lru_counter;
+    uint64_t                cache_clean_lru_counter;
 };
 
 static inline void *qcow2_cache_get_table_addr(BlockDriverState *bs,
                     Qcow2Cache *c, int table)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     return (uint8_t *) c->table_array + (size_t) table * s->cluster_size;
 }
 
 static inline int qcow2_cache_get_table_idx(BlockDriverState *bs,
                   Qcow2Cache *c, void *table)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     ptrdiff_t table_offset = (uint8_t *) table - (uint8_t *) c->table_array;
     int idx = table_offset / s->cluster_size;
     assert(idx >= 0 && idx < c->size && table_offset % s->cluster_size == 0);
     return idx;
 }
 
+static void qcow2_cache_table_release(BlockDriverState *bs, Qcow2Cache *c,
+                                      int i, int num_tables)
+{
+#if QEMU_MADV_DONTNEED != QEMU_MADV_INVALID
+    BDRVQcow2State *s = bs->opaque;
+    void *t = qcow2_cache_get_table_addr(bs, c, i);
+    int align = getpagesize();
+    size_t mem_size = (size_t) s->cluster_size * num_tables;
+    size_t offset = QEMU_ALIGN_UP((uintptr_t) t, align) - (uintptr_t) t;
+    size_t length = QEMU_ALIGN_DOWN(mem_size - offset, align);
+    if (length > 0) {
+        qemu_madvise((uint8_t *) t + offset, length, QEMU_MADV_DONTNEED);
+    }
+#endif
+}
+
+static inline bool can_clean_entry(Qcow2Cache *c, int i)
+{
+    Qcow2CachedTable *t = &c->entries[i];
+    return t->ref == 0 && !t->dirty && t->offset != 0 &&
+        t->lru_counter <= c->cache_clean_lru_counter;
+}
+
+void qcow2_cache_clean_unused(BlockDriverState *bs, Qcow2Cache *c)
+{
+    int i = 0;
+    while (i < c->size) {
+        int to_clean = 0;
+
+        /* Skip the entries that we don't need to clean */
+        while (i < c->size && !can_clean_entry(c, i)) {
+            i++;
+        }
+
+        /* And count how many we can clean in a row */
+        while (i < c->size && can_clean_entry(c, i)) {
+            c->entries[i].offset = 0;
+            c->entries[i].lru_counter = 0;
+            i++;
+            to_clean++;
+        }
+
+        if (to_clean > 0) {
+            qcow2_cache_table_release(bs, c, i - to_clean, to_clean);
+        }
+    }
+
+    c->cache_clean_lru_counter = c->lru_counter;
+}
+
 Qcow2Cache *qcow2_cache_create(BlockDriverState *bs, int num_tables)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     Qcow2Cache *c;
 
     c = g_new0(Qcow2Cache, 1);
     c->size = num_tables;
     c->entries = g_try_new0(Qcow2CachedTable, num_tables);
-    c->table_array = qemu_try_blockalign(bs->file,
+    c->table_array = qemu_try_blockalign(bs->file->bs,
                                          (size_t) num_tables * s->cluster_size);
 
     if (!c->entries || !c->table_array) {
@@ -113,7 +172,7 @@ static int qcow2_cache_flush_dependency(BlockDriverState *bs, Qcow2Cache *c)
 
 static int qcow2_cache_entry_flush(BlockDriverState *bs, Qcow2Cache *c, int i)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret = 0;
 
     if (!c->entries[i].dirty || !c->entries[i].offset) {
@@ -126,7 +185,7 @@ static int qcow2_cache_entry_flush(BlockDriverState *bs, Qcow2Cache *c, int i)
     if (c->depends) {
         ret = qcow2_cache_flush_dependency(bs, c);
     } else if (c->depends_on_flush) {
-        ret = bdrv_flush(bs->file);
+        ret = bdrv_flush(bs->file->bs);
         if (ret >= 0) {
             c->depends_on_flush = false;
         }
@@ -157,7 +216,7 @@ static int qcow2_cache_entry_flush(BlockDriverState *bs, Qcow2Cache *c, int i)
         BLKDBG_EVENT(bs->file, BLKDBG_L2_UPDATE);
     }
 
-    ret = bdrv_pwrite(bs->file, c->entries[i].offset,
+    ret = bdrv_pwrite(bs->file->bs, c->entries[i].offset,
                       qcow2_cache_get_table_addr(bs, c, i), s->cluster_size);
     if (ret < 0) {
         return ret;
@@ -170,7 +229,7 @@ static int qcow2_cache_entry_flush(BlockDriverState *bs, Qcow2Cache *c, int i)
 
 int qcow2_cache_flush(BlockDriverState *bs, Qcow2Cache *c)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int result = 0;
     int ret;
     int i;
@@ -185,7 +244,7 @@ int qcow2_cache_flush(BlockDriverState *bs, Qcow2Cache *c)
     }
 
     if (result == 0) {
-        ret = bdrv_flush(bs->file);
+        ret = bdrv_flush(bs->file->bs);
         if (ret < 0) {
             result = ret;
         }
@@ -237,6 +296,8 @@ int qcow2_cache_empty(BlockDriverState *bs, Qcow2Cache *c)
         c->entries[i].lru_counter = 0;
     }
 
+    qcow2_cache_table_release(bs, c, 0, c->size);
+
     c->lru_counter = 0;
 
     return 0;
@@ -245,7 +306,7 @@ int qcow2_cache_empty(BlockDriverState *bs, Qcow2Cache *c)
 static int qcow2_cache_do_get(BlockDriverState *bs, Qcow2Cache *c,
     uint64_t offset, void **table, bool read_from_disk)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int i;
     int ret;
     int lookup_index;
@@ -295,7 +356,8 @@ static int qcow2_cache_do_get(BlockDriverState *bs, Qcow2Cache *c,
             BLKDBG_EVENT(bs->file, BLKDBG_L2_LOAD);
         }
 
-        ret = bdrv_pread(bs->file, offset, qcow2_cache_get_table_addr(bs, c, i),
+        ret = bdrv_pread(bs->file->bs, offset,
+                         qcow2_cache_get_table_addr(bs, c, i),
                          s->cluster_size);
         if (ret < 0) {
             return ret;
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index 7e94fe70e..24a60e223 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -32,7 +32,7 @@
 int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
                         bool exact_size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int new_l1_size2, ret, i;
     uint64_t *new_l1_table;
     int64_t old_l1_table_offset, old_l1_size;
@@ -72,7 +72,7 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
 #endif
 
     new_l1_size2 = sizeof(uint64_t) * new_l1_size;
-    new_l1_table = qemu_try_blockalign(bs->file,
+    new_l1_table = qemu_try_blockalign(bs->file->bs,
                                        align_offset(new_l1_size2, 512));
     if (new_l1_table == NULL) {
         return -ENOMEM;
@@ -105,7 +105,8 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
     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);
+    ret = bdrv_pwrite_sync(bs->file->bs, new_l1_table_offset,
+                           new_l1_table, new_l1_size2);
     if (ret < 0)
         goto fail;
     for(i = 0; i < s->l1_size; i++)
@@ -115,7 +116,8 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
     BLKDBG_EVENT(bs->file, BLKDBG_L1_GROW_ACTIVATE_TABLE);
     cpu_to_be32w((uint32_t*)data, new_l1_size);
     stq_be_p(data + 4, new_l1_table_offset);
-    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, l1_size), data,sizeof(data));
+    ret = bdrv_pwrite_sync(bs->file->bs, offsetof(QCowHeader, l1_size),
+                           data, sizeof(data));
     if (ret < 0) {
         goto fail;
     }
@@ -148,7 +150,7 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
 static int l2_load(BlockDriverState *bs, uint64_t l2_offset,
     uint64_t **l2_table)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret;
 
     ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset, (void**) l2_table);
@@ -163,7 +165,7 @@ static int l2_load(BlockDriverState *bs, uint64_t l2_offset,
 #define L1_ENTRIES_PER_SECTOR (512 / 8)
 int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t buf[L1_ENTRIES_PER_SECTOR] = { 0 };
     int l1_start_index;
     int i, ret;
@@ -182,8 +184,9 @@ int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index)
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_L1_UPDATE);
-    ret = bdrv_pwrite_sync(bs->file, s->l1_table_offset + 8 * l1_start_index,
-        buf, sizeof(buf));
+    ret = bdrv_pwrite_sync(bs->file->bs,
+                           s->l1_table_offset + 8 * l1_start_index,
+                           buf, sizeof(buf));
     if (ret < 0) {
         return ret;
     }
@@ -203,7 +206,7 @@ int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index)
 
 static int l2_allocate(BlockDriverState *bs, int l1_index, uint64_t **table)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t old_l2_offset;
     uint64_t *l2_table = NULL;
     int64_t l2_offset;
@@ -309,7 +312,7 @@ static int count_contiguous_clusters(int nb_clusters, int cluster_size,
     if (!offset)
         return 0;
 
-    assert(qcow2_get_cluster_type(first_entry) != QCOW2_CLUSTER_COMPRESSED);
+    assert(qcow2_get_cluster_type(first_entry) == QCOW2_CLUSTER_NORMAL);
 
     for (i = 0; i < nb_clusters; i++) {
         uint64_t l2_entry = be64_to_cpu(l2_table[i]) & mask;
@@ -321,14 +324,16 @@ static int count_contiguous_clusters(int nb_clusters, int cluster_size,
 	return i;
 }
 
-static int count_contiguous_free_clusters(int nb_clusters, uint64_t *l2_table)
+static int count_contiguous_clusters_by_type(int nb_clusters,
+                                             uint64_t *l2_table,
+                                             int wanted_type)
 {
     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) {
+        if (type != wanted_type) {
             break;
         }
     }
@@ -339,7 +344,7 @@ static int count_contiguous_free_clusters(int nb_clusters, uint64_t *l2_table)
 /* The crypt function is compatible with the linux cryptoloop
    algorithm for < 4 GB images. NOTE: out_buf == in_buf is
    supported */
-int qcow2_encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
+int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num,
                           uint8_t *out_buf, const uint8_t *in_buf,
                           int nb_sectors, bool enc,
                           Error **errp)
@@ -387,7 +392,7 @@ static int coroutine_fn copy_sectors(BlockDriverState *bs,
                                      uint64_t cluster_offset,
                                      int n_start, int n_end)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QEMUIOVector qiov;
     struct iovec iov;
     int n, ret;
@@ -440,7 +445,8 @@ static int coroutine_fn copy_sectors(BlockDriverState *bs,
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_COW_WRITE);
-    ret = bdrv_co_writev(bs->file, (cluster_offset >> 9) + n_start, n, &qiov);
+    ret = bdrv_co_writev(bs->file->bs, (cluster_offset >> 9) + n_start, n,
+                         &qiov);
     if (ret < 0) {
         goto out;
     }
@@ -469,7 +475,7 @@ out:
 int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
     int *num, uint64_t *cluster_offset)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     unsigned int l2_index;
     uint64_t l1_index, l2_offset, *l2_table;
     int l1_bits, c;
@@ -499,7 +505,7 @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
 
     *cluster_offset = 0;
 
-    /* seek the the l2 offset in the l1 table */
+    /* seek to the l2 offset in the l1 table */
 
     l1_index = offset >> l1_bits;
     if (l1_index >= s->l1_size) {
@@ -550,13 +556,14 @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
             ret = -EIO;
             goto fail;
         }
-        c = count_contiguous_clusters(nb_clusters, s->cluster_size,
-                &l2_table[l2_index], QCOW_OFLAG_ZERO);
+        c = count_contiguous_clusters_by_type(nb_clusters, &l2_table[l2_index],
+                                              QCOW2_CLUSTER_ZERO);
         *cluster_offset = 0;
         break;
     case QCOW2_CLUSTER_UNALLOCATED:
         /* how many empty clusters ? */
-        c = count_contiguous_free_clusters(nb_clusters, &l2_table[l2_index]);
+        c = count_contiguous_clusters_by_type(nb_clusters, &l2_table[l2_index],
+                                              QCOW2_CLUSTER_UNALLOCATED);
         *cluster_offset = 0;
         break;
     case QCOW2_CLUSTER_NORMAL:
@@ -609,13 +616,13 @@ static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
                              uint64_t **new_l2_table,
                              int *new_l2_index)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *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 */
+    /* seek to the l2 offset in the l1 table */
 
     l1_index = offset >> (s->l2_bits + s->cluster_bits);
     if (l1_index >= s->l1_size) {
@@ -683,7 +690,7 @@ uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
                                                uint64_t offset,
                                                int compressed_size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int l2_index, ret;
     uint64_t *l2_table;
     int64_t cluster_offset;
@@ -728,7 +735,7 @@ uint64_t qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
 
 static int perform_cow(BlockDriverState *bs, QCowL2Meta *m, Qcow2COWRegion *r)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret;
 
     if (r->nb_sectors == 0) {
@@ -757,7 +764,7 @@ static int perform_cow(BlockDriverState *bs, QCowL2Meta *m, Qcow2COWRegion *r)
 
 int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int i, j = 0, l2_index, ret;
     uint64_t *old_cluster, *l2_table;
     uint64_t cluster_offset = m->alloc_offset;
@@ -817,7 +824,6 @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
 
     /*
      * 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.
@@ -840,7 +846,7 @@ err:
  * 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,
+static int count_cow_clusters(BDRVQcow2State *s, int nb_clusters,
     uint64_t *l2_table, int l2_index)
 {
     int i;
@@ -886,7 +892,7 @@ out:
 static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
     uint64_t *cur_bytes, QCowL2Meta **m)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowL2Meta *old_alloc;
     uint64_t bytes = *cur_bytes;
 
@@ -959,7 +965,7 @@ static int handle_dependencies(BlockDriverState *bs, uint64_t guest_offset,
 static int handle_copied(BlockDriverState *bs, uint64_t guest_offset,
     uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int l2_index;
     uint64_t cluster_offset;
     uint64_t *l2_table;
@@ -1067,7 +1073,7 @@ out:
 static int do_alloc_cluster_offset(BlockDriverState *bs, uint64_t guest_offset,
                                    uint64_t *host_offset, uint64_t *nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     trace_qcow2_do_alloc_clusters_offset(qemu_coroutine_self(), guest_offset,
                                          *host_offset, *nb_clusters);
@@ -1115,7 +1121,7 @@ static int do_alloc_cluster_offset(BlockDriverState *bs, uint64_t guest_offset,
 static int handle_alloc(BlockDriverState *bs, uint64_t guest_offset,
     uint64_t *host_offset, uint64_t *bytes, QCowL2Meta **m)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int l2_index;
     uint64_t *l2_table;
     uint64_t entry;
@@ -1268,7 +1274,7 @@ fail:
 int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
     int *num, uint64_t *host_offset, QCowL2Meta **m)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t start, remaining;
     uint64_t cluster_offset;
     uint64_t cur_bytes;
@@ -1402,7 +1408,7 @@ static int decompress_buffer(uint8_t *out_buf, int out_buf_size,
 
 int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret, csize, nb_csectors, sector_offset;
     uint64_t coffset;
 
@@ -1412,7 +1418,8 @@ int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset)
         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);
+        ret = bdrv_read(bs->file->bs, coffset >> 9, s->cluster_data,
+                        nb_csectors);
         if (ret < 0) {
             return ret;
         }
@@ -1434,7 +1441,7 @@ static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
                              uint64_t nb_clusters, enum qcow2_discard_type type,
                              bool full_discard)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l2_table;
     int l2_index;
     int ret;
@@ -1469,7 +1476,7 @@ static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
          */
         switch (qcow2_get_cluster_type(old_l2_entry)) {
             case QCOW2_CLUSTER_UNALLOCATED:
-                if (full_discard || !bs->backing_hd) {
+                if (full_discard || !bs->backing) {
                     continue;
                 }
                 break;
@@ -1508,7 +1515,7 @@ static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
 int qcow2_discard_clusters(BlockDriverState *bs, uint64_t offset,
     int nb_sectors, enum qcow2_discard_type type, bool full_discard)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t end_offset;
     uint64_t nb_clusters;
     int ret;
@@ -1554,7 +1561,7 @@ fail:
 static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
                           uint64_t nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l2_table;
     int l2_index;
     int ret;
@@ -1591,7 +1598,7 @@ static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
 
 int qcow2_zero_clusters(BlockDriverState *bs, uint64_t offset, int nb_sectors)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t nb_clusters;
     int ret;
 
@@ -1636,7 +1643,7 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                                       int64_t l1_entries,
                                       BlockDriverAmendStatusCB *status_cb)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     bool is_active_l1 = (l1_table == s->l1_table);
     uint64_t *l2_table = NULL;
     int ret;
@@ -1645,7 +1652,7 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
     if (!is_active_l1) {
         /* inactive L2 tables require a buffer to be stored in when loading
          * them from disk */
-        l2_table = qemu_try_blockalign(bs->file, s->cluster_size);
+        l2_table = qemu_try_blockalign(bs->file->bs, s->cluster_size);
         if (l2_table == NULL) {
             return -ENOMEM;
         }
@@ -1679,8 +1686,8 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                     (void **)&l2_table);
         } else {
             /* load inactive L2 tables from disk */
-            ret = bdrv_read(bs->file, l2_offset / BDRV_SECTOR_SIZE,
-                    (void *)l2_table, s->cluster_sectors);
+            ret = bdrv_read(bs->file->bs, l2_offset / BDRV_SECTOR_SIZE,
+                            (void *)l2_table, s->cluster_sectors);
         }
         if (ret < 0) {
             goto fail;
@@ -1703,7 +1710,7 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
             }
 
             if (!preallocated) {
-                if (!bs->backing_hd) {
+                if (!bs->backing) {
                     /* not backed; therefore we can simply deallocate the
                      * cluster */
                     l2_table[j] = 0;
@@ -1754,7 +1761,7 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                 goto fail;
             }
 
-            ret = bdrv_write_zeroes(bs->file, offset / BDRV_SECTOR_SIZE,
+            ret = bdrv_write_zeroes(bs->file->bs, offset / BDRV_SECTOR_SIZE,
                                     s->cluster_sectors, 0);
             if (ret < 0) {
                 if (!preallocated) {
@@ -1787,8 +1794,8 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                     goto fail;
                 }
 
-                ret = bdrv_write(bs->file, l2_offset / BDRV_SECTOR_SIZE,
-                        (void *)l2_table, s->cluster_sectors);
+                ret = bdrv_write(bs->file->bs, l2_offset / BDRV_SECTOR_SIZE,
+                                 (void *)l2_table, s->cluster_sectors);
                 if (ret < 0) {
                     goto fail;
                 }
@@ -1823,7 +1830,7 @@ fail:
 int qcow2_expand_zero_clusters(BlockDriverState *bs,
                                BlockDriverAmendStatusCB *status_cb)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l1_table = NULL;
     int64_t l1_entries = 0, visited_l1_entries = 0;
     int ret;
@@ -1861,8 +1868,9 @@ int qcow2_expand_zero_clusters(BlockDriverState *bs,
 
         l1_table = g_realloc(l1_table, l1_sectors * BDRV_SECTOR_SIZE);
 
-        ret = bdrv_read(bs->file, s->snapshots[i].l1_table_offset /
-                BDRV_SECTOR_SIZE, (void *)l1_table, l1_sectors);
+        ret = bdrv_read(bs->file->bs,
+                        s->snapshots[i].l1_table_offset / BDRV_SECTOR_SIZE,
+                        (void *)l1_table, l1_sectors);
         if (ret < 0) {
             goto fail;
         }
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 0b6c302ee..820f412ab 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -82,7 +82,7 @@ static Qcow2SetRefcountFunc *const set_refcount_funcs[] = {
 
 int qcow2_refcount_init(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     unsigned int refcount_table_size2, i;
     int ret;
 
@@ -101,7 +101,7 @@ int qcow2_refcount_init(BlockDriverState *bs)
             goto fail;
         }
         BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD);
-        ret = bdrv_pread(bs->file, s->refcount_table_offset,
+        ret = bdrv_pread(bs->file->bs, s->refcount_table_offset,
                          s->refcount_table, refcount_table_size2);
         if (ret < 0) {
             goto fail;
@@ -116,7 +116,7 @@ int qcow2_refcount_init(BlockDriverState *bs)
 
 void qcow2_refcount_close(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     g_free(s->refcount_table);
 }
 
@@ -214,7 +214,7 @@ static int load_refcount_block(BlockDriverState *bs,
                                int64_t refcount_block_offset,
                                void **refcount_block)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret;
 
     BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_LOAD);
@@ -231,7 +231,7 @@ static int load_refcount_block(BlockDriverState *bs,
 int qcow2_get_refcount(BlockDriverState *bs, int64_t cluster_index,
                        uint64_t *refcount)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t refcount_table_index, block_index;
     int64_t refcount_block_offset;
     int ret;
@@ -274,7 +274,7 @@ int qcow2_get_refcount(BlockDriverState *bs, int64_t cluster_index,
  * 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,
+static unsigned int next_refcount_table_size(BDRVQcow2State *s,
     unsigned int min_size)
 {
     unsigned int min_clusters = (min_size >> (s->cluster_bits - 3)) + 1;
@@ -290,7 +290,7 @@ static unsigned int next_refcount_table_size(BDRVQcowState *s,
 
 
 /* Checks if two offsets are described by the same refcount block */
-static int in_same_refcount_block(BDRVQcowState *s, uint64_t offset_a,
+static int in_same_refcount_block(BDRVQcow2State *s, uint64_t offset_a,
     uint64_t offset_b)
 {
     uint64_t block_a = offset_a >> (s->cluster_bits + s->refcount_block_bits);
@@ -308,7 +308,7 @@ static int in_same_refcount_block(BDRVQcowState *s, uint64_t offset_a,
 static int alloc_refcount_block(BlockDriverState *bs,
                                 int64_t cluster_index, void **refcount_block)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     unsigned int refcount_table_index;
     int ret;
 
@@ -431,7 +431,7 @@ static int alloc_refcount_block(BlockDriverState *bs,
     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,
+        ret = bdrv_pwrite_sync(bs->file->bs,
             s->refcount_table_offset + refcount_table_index * sizeof(uint64_t),
             &data64, sizeof(data64));
         if (ret < 0) {
@@ -535,7 +535,7 @@ static int alloc_refcount_block(BlockDriverState *bs,
 
     /* Write refcount blocks to disk */
     BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_BLOCKS);
-    ret = bdrv_pwrite_sync(bs->file, meta_offset, new_blocks,
+    ret = bdrv_pwrite_sync(bs->file->bs, meta_offset, new_blocks,
         blocks_clusters * s->cluster_size);
     g_free(new_blocks);
     new_blocks = NULL;
@@ -549,7 +549,7 @@ static int alloc_refcount_block(BlockDriverState *bs,
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_TABLE);
-    ret = bdrv_pwrite_sync(bs->file, table_offset, new_table,
+    ret = bdrv_pwrite_sync(bs->file->bs, table_offset, new_table,
         table_size * sizeof(uint64_t));
     if (ret < 0) {
         goto fail_table;
@@ -560,12 +560,16 @@ static int alloc_refcount_block(BlockDriverState *bs,
     }
 
     /* 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);
+    struct QEMU_PACKED {
+        uint64_t d64;
+        uint32_t d32;
+    } data;
+    cpu_to_be64w(&data.d64, table_offset);
+    cpu_to_be32w(&data.d32, 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));
+    ret = bdrv_pwrite_sync(bs->file->bs,
+                           offsetof(QCowHeader, refcount_table_offset),
+                           &data, sizeof(data));
     if (ret < 0) {
         goto fail_table;
     }
@@ -605,7 +609,7 @@ fail_block:
 
 void qcow2_process_discards(BlockDriverState *bs, int ret)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     Qcow2DiscardRegion *d, *next;
 
     QTAILQ_FOREACH_SAFE(d, &s->discards, next, next) {
@@ -613,7 +617,7 @@ void qcow2_process_discards(BlockDriverState *bs, int ret)
 
         /* Discard is optional, ignore the return value */
         if (ret >= 0) {
-            bdrv_discard(bs->file,
+            bdrv_discard(bs->file->bs,
                          d->offset >> BDRV_SECTOR_BITS,
                          d->bytes >> BDRV_SECTOR_BITS);
         }
@@ -625,7 +629,7 @@ void qcow2_process_discards(BlockDriverState *bs, int ret)
 static void update_refcount_discard(BlockDriverState *bs,
                                     uint64_t offset, uint64_t length)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     Qcow2DiscardRegion *d, *p, *next;
 
     QTAILQ_FOREACH(d, &s->discards, next) {
@@ -682,7 +686,7 @@ static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
                                                    bool decrease,
                                                    enum qcow2_discard_type type)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t start, last, cluster_offset;
     void *refcount_block = NULL;
     int64_t old_table_index = -1;
@@ -793,7 +797,7 @@ int qcow2_update_cluster_refcount(BlockDriverState *bs,
                                   uint64_t addend, bool decrease,
                                   enum qcow2_discard_type type)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret;
 
     ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend,
@@ -815,7 +819,7 @@ int qcow2_update_cluster_refcount(BlockDriverState *bs,
 /* return < 0 if error */
 static int64_t alloc_clusters_noref(BlockDriverState *bs, uint64_t size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t i, nb_clusters, refcount;
     int ret;
 
@@ -878,7 +882,7 @@ int64_t qcow2_alloc_clusters(BlockDriverState *bs, uint64_t size)
 int64_t qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
                                 int64_t nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t cluster_index, refcount;
     uint64_t i;
     int ret;
@@ -916,7 +920,7 @@ int64_t qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
    contiguous sectors. size must be <= cluster_size */
 int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t offset;
     size_t free_in_cluster;
     int ret;
@@ -949,11 +953,17 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
 
             if (!offset || ROUND_UP(offset, s->cluster_size) != new_cluster) {
                 offset = new_cluster;
+                free_in_cluster = s->cluster_size;
+            } else {
+                free_in_cluster += s->cluster_size;
             }
         }
 
         assert(offset);
         ret = update_refcount(bs, offset, size, 1, false, QCOW2_DISCARD_NEVER);
+        if (ret < 0) {
+            offset = 0;
+        }
     } while (ret == -EAGAIN);
     if (ret < 0) {
         return ret;
@@ -992,7 +1002,7 @@ void qcow2_free_clusters(BlockDriverState *bs,
 void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
                              int nb_clusters, enum qcow2_discard_type type)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     switch (qcow2_get_cluster_type(l2_entry)) {
     case QCOW2_CLUSTER_COMPRESSED:
@@ -1036,7 +1046,7 @@ void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
 int qcow2_update_snapshot_refcount(BlockDriverState *bs,
     int64_t l1_table_offset, int l1_size, int addend)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l1_table, *l2_table, l2_offset, offset, l1_size2, refcount;
     bool l1_allocated = false;
     int64_t old_offset, old_l2_offset;
@@ -1062,7 +1072,7 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
         }
         l1_allocated = true;
 
-        ret = bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2);
+        ret = bdrv_pread(bs->file->bs, l1_table_offset, l1_table, l1_size2);
         if (ret < 0) {
             goto fail;
         }
@@ -1215,7 +1225,8 @@ fail:
             cpu_to_be64s(&l1_table[i]);
         }
 
-        ret = bdrv_pwrite_sync(bs->file, l1_table_offset, l1_table, l1_size2);
+        ret = bdrv_pwrite_sync(bs->file->bs, l1_table_offset,
+                               l1_table, l1_size2);
 
         for (i = 0; i < l1_size; i++) {
             be64_to_cpus(&l1_table[i]);
@@ -1233,7 +1244,7 @@ fail:
 /* refcount checking functions */
 
 
-static size_t refcount_array_byte_size(BDRVQcowState *s, uint64_t entries)
+static uint64_t refcount_array_byte_size(BDRVQcow2State *s, uint64_t entries)
 {
     /* This assertion holds because there is no way we can address more than
      * 2^(64 - 9) clusters at once (with cluster size 512 = 2^9, and because
@@ -1256,7 +1267,7 @@ static size_t refcount_array_byte_size(BDRVQcowState *s, uint64_t entries)
  * refcount array buffer will be aligned to a cluster boundary, and the newly
  * allocated area will be zeroed.
  */
-static int realloc_refcount_array(BDRVQcowState *s, void **array,
+static int realloc_refcount_array(BDRVQcow2State *s, void **array,
                                   int64_t *size, int64_t new_size)
 {
     int64_t old_byte_size, new_byte_size;
@@ -1298,7 +1309,7 @@ static int realloc_refcount_array(BDRVQcowState *s, void **array,
 /*
  * 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.
+ * which can be compared to the refcount table saved in the image.
  *
  * Modifies the number of errors in res.
  */
@@ -1308,7 +1319,7 @@ static int inc_refcounts(BlockDriverState *bs,
                          int64_t *refcount_table_size,
                          int64_t offset, int64_t size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t start, last, cluster_offset, k, refcount;
     int ret;
 
@@ -1361,7 +1372,7 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
                               int64_t *refcount_table_size, int64_t l2_offset,
                               int flags)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l2_table, l2_entry;
     uint64_t next_contiguous_offset = 0;
     int i, l2_size, nb_csectors, ret;
@@ -1370,7 +1381,7 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
     l2_size = s->l2_size * sizeof(uint64_t);
     l2_table = g_malloc(l2_size);
 
-    ret = bdrv_pread(bs->file, l2_offset, l2_table, l2_size);
+    ret = bdrv_pread(bs->file->bs, l2_offset, l2_table, l2_size);
     if (ret < 0) {
         fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n");
         res->check_errors++;
@@ -1481,7 +1492,7 @@ static int check_refcounts_l1(BlockDriverState *bs,
                               int64_t l1_table_offset, int l1_size,
                               int flags)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l1_table = NULL, l2_offset, l1_size2;
     int i, ret;
 
@@ -1502,7 +1513,7 @@ static int check_refcounts_l1(BlockDriverState *bs,
             res->check_errors++;
             goto fail;
         }
-        ret = bdrv_pread(bs->file, l1_table_offset, l1_table, l1_size2);
+        ret = bdrv_pread(bs->file->bs, l1_table_offset, l1_table, l1_size2);
         if (ret < 0) {
             fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
             res->check_errors++;
@@ -1558,7 +1569,7 @@ fail:
 static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
                               BdrvCheckMode fix)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t *l2_table = qemu_blockalign(bs, s->cluster_size);
     int ret;
     uint64_t refcount;
@@ -1600,7 +1611,7 @@ static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
             }
         }
 
-        ret = bdrv_pread(bs->file, l2_offset, l2_table,
+        ret = bdrv_pread(bs->file->bs, l2_offset, l2_table,
                          s->l2_size * sizeof(uint64_t));
         if (ret < 0) {
             fprintf(stderr, "ERROR: Could not read L2 table: %s\n",
@@ -1652,7 +1663,8 @@ static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res,
                 goto fail;
             }
 
-            ret = bdrv_pwrite(bs->file, l2_offset, l2_table, s->cluster_size);
+            ret = bdrv_pwrite(bs->file->bs, l2_offset, l2_table,
+                              s->cluster_size);
             if (ret < 0) {
                 fprintf(stderr, "ERROR: Could not write L2 table: %s\n",
                         strerror(-ret));
@@ -1677,7 +1689,7 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
                            BdrvCheckMode fix, bool *rebuild,
                            void **refcount_table, int64_t *nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t i, size;
     int ret;
 
@@ -1707,11 +1719,11 @@ static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res,
                     goto resize_fail;
                 }
 
-                ret = bdrv_truncate(bs->file, offset + s->cluster_size);
+                ret = bdrv_truncate(bs->file->bs, offset + s->cluster_size);
                 if (ret < 0) {
                     goto resize_fail;
                 }
-                size = bdrv_getlength(bs->file);
+                size = bdrv_getlength(bs->file->bs);
                 if (size < 0) {
                     ret = size;
                     goto resize_fail;
@@ -1780,7 +1792,7 @@ static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                                BdrvCheckMode fix, bool *rebuild,
                                void **refcount_table, int64_t *nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t i;
     QCowSnapshot *sn;
     int ret;
@@ -1844,7 +1856,7 @@ static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                               int64_t *highest_cluster,
                               void *refcount_table, int64_t nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t i;
     uint64_t refcount1, refcount2;
     int ret;
@@ -1921,7 +1933,7 @@ static int64_t alloc_clusters_imrt(BlockDriverState *bs,
                                    int64_t *imrt_nb_clusters,
                                    int64_t *first_free_cluster)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t cluster = *first_free_cluster, i;
     bool first_gap = true;
     int contiguous_free_clusters;
@@ -1991,7 +2003,7 @@ static int rebuild_refcount_structure(BlockDriverState *bs,
                                       void **refcount_table,
                                       int64_t *nb_clusters)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t first_free_cluster = 0, reftable_offset = -1, cluster = 0;
     int64_t refblock_offset, refblock_start, refblock_index;
     uint32_t reftable_size = 0;
@@ -2085,7 +2097,7 @@ write_refblocks:
         on_disk_refblock = (void *)((char *) *refcount_table +
                                     refblock_index * s->cluster_size);
 
-        ret = bdrv_write(bs->file, refblock_offset / BDRV_SECTOR_SIZE,
+        ret = bdrv_write(bs->file->bs, refblock_offset / BDRV_SECTOR_SIZE,
                          on_disk_refblock, s->cluster_sectors);
         if (ret < 0) {
             fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret));
@@ -2134,7 +2146,7 @@ write_refblocks:
     }
 
     assert(reftable_size < INT_MAX / sizeof(uint64_t));
-    ret = bdrv_pwrite(bs->file, reftable_offset, on_disk_reftable,
+    ret = bdrv_pwrite(bs->file->bs, reftable_offset, on_disk_reftable,
                       reftable_size * sizeof(uint64_t));
     if (ret < 0) {
         fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret));
@@ -2146,8 +2158,8 @@ write_refblocks:
                  reftable_offset);
     cpu_to_be32w(&reftable_offset_and_clusters.reftable_clusters,
                  size_to_clusters(s, reftable_size * sizeof(uint64_t)));
-    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader,
-                                              refcount_table_offset),
+    ret = bdrv_pwrite_sync(bs->file->bs, offsetof(QCowHeader,
+                                                  refcount_table_offset),
                            &reftable_offset_and_clusters,
                            sizeof(reftable_offset_and_clusters));
     if (ret < 0) {
@@ -2178,14 +2190,14 @@ fail:
 int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                           BdrvCheckMode fix)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     BdrvCheckResult pre_compare_res;
     int64_t size, highest_cluster, nb_clusters;
     void *refcount_table = NULL;
     bool rebuild = false;
     int ret;
 
-    size = bdrv_getlength(bs->file);
+    size = bdrv_getlength(bs->file->bs);
     if (size < 0) {
         res->check_errors++;
         return size;
@@ -2315,7 +2327,7 @@ fail:
 int qcow2_check_metadata_overlap(BlockDriverState *bs, int ign, int64_t offset,
                                  int64_t size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int chk = s->overlap_check & ~ign;
     int i, j;
 
@@ -2394,7 +2406,7 @@ int qcow2_check_metadata_overlap(BlockDriverState *bs, int ign, int64_t offset,
                 return -ENOMEM;
             }
 
-            ret = bdrv_pread(bs->file, l1_ofs, l1, l1_sz2);
+            ret = bdrv_pread(bs->file->bs, l1_ofs, l1, l1_sz2);
             if (ret < 0) {
                 g_free(l1);
                 return ret;
diff --git a/block/qcow2-snapshot.c b/block/qcow2-snapshot.c
index b6f58c13e..def720164 100644
--- a/block/qcow2-snapshot.c
+++ b/block/qcow2-snapshot.c
@@ -29,7 +29,7 @@
 
 void qcow2_free_snapshots(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int i;
 
     for(i = 0; i < s->nb_snapshots; i++) {
@@ -43,7 +43,7 @@ void qcow2_free_snapshots(BlockDriverState *bs)
 
 int qcow2_read_snapshots(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshotHeader h;
     QCowSnapshotExtraData extra;
     QCowSnapshot *sn;
@@ -64,7 +64,7 @@ int qcow2_read_snapshots(BlockDriverState *bs)
     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));
+        ret = bdrv_pread(bs->file->bs, offset, &h, sizeof(h));
         if (ret < 0) {
             goto fail;
         }
@@ -83,7 +83,7 @@ int qcow2_read_snapshots(BlockDriverState *bs)
         name_size = be16_to_cpu(h.name_size);
 
         /* Read extra data */
-        ret = bdrv_pread(bs->file, offset, &extra,
+        ret = bdrv_pread(bs->file->bs, offset, &extra,
                          MIN(sizeof(extra), extra_data_size));
         if (ret < 0) {
             goto fail;
@@ -102,7 +102,7 @@ int qcow2_read_snapshots(BlockDriverState *bs)
 
         /* Read snapshot ID */
         sn->id_str = g_malloc(id_str_size + 1);
-        ret = bdrv_pread(bs->file, offset, sn->id_str, id_str_size);
+        ret = bdrv_pread(bs->file->bs, offset, sn->id_str, id_str_size);
         if (ret < 0) {
             goto fail;
         }
@@ -111,7 +111,7 @@ int qcow2_read_snapshots(BlockDriverState *bs)
 
         /* Read snapshot name */
         sn->name = g_malloc(name_size + 1);
-        ret = bdrv_pread(bs->file, offset, sn->name, name_size);
+        ret = bdrv_pread(bs->file->bs, offset, sn->name, name_size);
         if (ret < 0) {
             goto fail;
         }
@@ -136,7 +136,7 @@ fail:
 /* add at the end of the file a new list of snapshots */
 static int qcow2_write_snapshots(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshot *sn;
     QCowSnapshotHeader h;
     QCowSnapshotExtraData extra;
@@ -214,25 +214,25 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
         h.name_size = cpu_to_be16(name_size);
         offset = align_offset(offset, 8);
 
-        ret = bdrv_pwrite(bs->file, offset, &h, sizeof(h));
+        ret = bdrv_pwrite(bs->file->bs, offset, &h, sizeof(h));
         if (ret < 0) {
             goto fail;
         }
         offset += sizeof(h);
 
-        ret = bdrv_pwrite(bs->file, offset, &extra, sizeof(extra));
+        ret = bdrv_pwrite(bs->file->bs, offset, &extra, sizeof(extra));
         if (ret < 0) {
             goto fail;
         }
         offset += sizeof(extra);
 
-        ret = bdrv_pwrite(bs->file, offset, sn->id_str, id_str_size);
+        ret = bdrv_pwrite(bs->file->bs, 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);
+        ret = bdrv_pwrite(bs->file->bs, offset, sn->name, name_size);
         if (ret < 0) {
             goto fail;
         }
@@ -254,7 +254,7 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
     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),
+    ret = bdrv_pwrite_sync(bs->file->bs, offsetof(QCowHeader, nb_snapshots),
                            &header_data, sizeof(header_data));
     if (ret < 0) {
         goto fail;
@@ -278,7 +278,7 @@ fail:
 static void find_new_snapshot_id(BlockDriverState *bs,
                                  char *id_str, int id_str_size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshot *sn;
     int i;
     unsigned long id, id_max = 0;
@@ -296,7 +296,7 @@ static int find_snapshot_by_id_and_name(BlockDriverState *bs,
                                         const char *id,
                                         const char *name)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int i;
 
     if (id && name) {
@@ -338,7 +338,7 @@ static int find_snapshot_by_id_or_name(BlockDriverState *bs,
 /* if no id is provided, a new one is constructed */
 int qcow2_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshot *new_snapshot_list = NULL;
     QCowSnapshot *old_snapshot_list = NULL;
     QCowSnapshot sn1, *sn = &sn1;
@@ -396,7 +396,7 @@ int qcow2_snapshot_create(BlockDriverState *bs, QEMUSnapshotInfo *sn_info)
         goto fail;
     }
 
-    ret = bdrv_pwrite(bs->file, sn->l1_table_offset, l1_table,
+    ret = bdrv_pwrite(bs->file->bs, sn->l1_table_offset, l1_table,
                       s->l1_size * sizeof(uint64_t));
     if (ret < 0) {
         goto fail;
@@ -461,7 +461,7 @@ fail:
 /* copy the snapshot 'snapshot_name' into the current disk image */
 int qcow2_snapshot_goto(BlockDriverState *bs, const char *snapshot_id)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshot *sn;
     int i, snapshot_index;
     int cur_l1_bytes, sn_l1_bytes;
@@ -509,7 +509,8 @@ int qcow2_snapshot_goto(BlockDriverState *bs, const char *snapshot_id)
         goto fail;
     }
 
-    ret = bdrv_pread(bs->file, sn->l1_table_offset, sn_l1_table, sn_l1_bytes);
+    ret = bdrv_pread(bs->file->bs, sn->l1_table_offset,
+                     sn_l1_table, sn_l1_bytes);
     if (ret < 0) {
         goto fail;
     }
@@ -526,7 +527,7 @@ int qcow2_snapshot_goto(BlockDriverState *bs, const char *snapshot_id)
         goto fail;
     }
 
-    ret = bdrv_pwrite_sync(bs->file, s->l1_table_offset, sn_l1_table,
+    ret = bdrv_pwrite_sync(bs->file->bs, s->l1_table_offset, sn_l1_table,
                            cur_l1_bytes);
     if (ret < 0) {
         goto fail;
@@ -587,7 +588,7 @@ int qcow2_snapshot_delete(BlockDriverState *bs,
                           const char *name,
                           Error **errp)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshot sn;
     int snapshot_index, ret;
 
@@ -650,7 +651,7 @@ int qcow2_snapshot_delete(BlockDriverState *bs,
 
 int qcow2_snapshot_list(BlockDriverState *bs, QEMUSnapshotInfo **psn_tab)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QEMUSnapshotInfo *sn_tab, *sn_info;
     QCowSnapshot *sn;
     int i;
@@ -683,7 +684,7 @@ int qcow2_snapshot_load_tmp(BlockDriverState *bs,
                             Error **errp)
 {
     int i, snapshot_index;
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowSnapshot *sn;
     uint64_t *new_l1_table;
     int new_l1_bytes;
@@ -706,13 +707,14 @@ int qcow2_snapshot_load_tmp(BlockDriverState *bs,
         return -EFBIG;
     }
     new_l1_bytes = sn->l1_size * sizeof(uint64_t);
-    new_l1_table = qemu_try_blockalign(bs->file,
+    new_l1_table = qemu_try_blockalign(bs->file->bs,
                                        align_offset(new_l1_bytes, 512));
     if (new_l1_table == NULL) {
         return -ENOMEM;
     }
 
-    ret = bdrv_pread(bs->file, sn->l1_table_offset, new_l1_table, new_l1_bytes);
+    ret = bdrv_pread(bs->file->bs, sn->l1_table_offset,
+                     new_l1_table, new_l1_bytes);
     if (ret < 0) {
         error_setg(errp, "Failed to read l1 table for snapshot");
         qemu_vfree(new_l1_table);
diff --git a/block/qcow2.c b/block/qcow2.c
index 76c331b38..88f56c886 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -85,7 +85,7 @@ static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
                                  uint64_t end_offset, void **p_feature_table,
                                  Error **errp)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowExtension ext;
     uint64_t offset;
     int ret;
@@ -104,7 +104,7 @@ static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
         printf("attempting to read extended header in offset %lu\n", offset);
 #endif
 
-        ret = bdrv_pread(bs->file, offset, &ext, sizeof(ext));
+        ret = bdrv_pread(bs->file->bs, offset, &ext, sizeof(ext));
         if (ret < 0) {
             error_setg_errno(errp, -ret, "qcow2_read_extension: ERROR: "
                              "pread fail from offset %" PRIu64, offset);
@@ -132,7 +132,7 @@ static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
                            sizeof(bs->backing_format));
                 return 2;
             }
-            ret = bdrv_pread(bs->file, offset, bs->backing_format, ext.len);
+            ret = bdrv_pread(bs->file->bs, offset, bs->backing_format, ext.len);
             if (ret < 0) {
                 error_setg_errno(errp, -ret, "ERROR: ext_backing_format: "
                                  "Could not read format name");
@@ -148,7 +148,7 @@ static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
         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);
+                ret = bdrv_pread(bs->file->bs, offset , feature_table, ext.len);
                 if (ret < 0) {
                     error_setg_errno(errp, -ret, "ERROR: ext_feature_table: "
                                      "Could not read table");
@@ -169,7 +169,7 @@ static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
                 uext->len = ext.len;
                 QLIST_INSERT_HEAD(&s->unknown_header_ext, uext, next);
 
-                ret = bdrv_pread(bs->file, offset , uext->data, uext->len);
+                ret = bdrv_pread(bs->file->bs, offset , uext->data, uext->len);
                 if (ret < 0) {
                     error_setg_errno(errp, -ret, "ERROR: unknown extension: "
                                      "Could not read data");
@@ -187,7 +187,7 @@ static int qcow2_read_extensions(BlockDriverState *bs, uint64_t start_offset,
 
 static void cleanup_unknown_header_ext(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     Qcow2UnknownHeaderExtension *uext, *next;
 
     QLIST_FOREACH_SAFE(uext, &s->unknown_header_ext, next, next) {
@@ -249,7 +249,7 @@ static void report_unsupported_feature(BlockDriverState *bs,
  */
 int qcow2_mark_dirty(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t val;
     int ret;
 
@@ -260,12 +260,12 @@ int qcow2_mark_dirty(BlockDriverState *bs)
     }
 
     val = cpu_to_be64(s->incompatible_features | QCOW2_INCOMPAT_DIRTY);
-    ret = bdrv_pwrite(bs->file, offsetof(QCowHeader, incompatible_features),
+    ret = bdrv_pwrite(bs->file->bs, offsetof(QCowHeader, incompatible_features),
                       &val, sizeof(val));
     if (ret < 0) {
         return ret;
     }
-    ret = bdrv_flush(bs->file);
+    ret = bdrv_flush(bs->file->bs);
     if (ret < 0) {
         return ret;
     }
@@ -282,7 +282,7 @@ int qcow2_mark_dirty(BlockDriverState *bs)
  */
 static int qcow2_mark_clean(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     if (s->incompatible_features & QCOW2_INCOMPAT_DIRTY) {
         int ret;
@@ -304,7 +304,7 @@ static int qcow2_mark_clean(BlockDriverState *bs)
  */
 int qcow2_mark_corrupt(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     s->incompatible_features |= QCOW2_INCOMPAT_CORRUPT;
     return qcow2_update_header(bs);
@@ -316,7 +316,7 @@ int qcow2_mark_corrupt(BlockDriverState *bs)
  */
 int qcow2_mark_consistent(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     if (s->incompatible_features & QCOW2_INCOMPAT_CORRUPT) {
         int ret = bdrv_flush(bs);
@@ -351,7 +351,7 @@ static int qcow2_check(BlockDriverState *bs, BdrvCheckResult *result,
 static int validate_table_offset(BlockDriverState *bs, uint64_t offset,
                                  uint64_t entries, size_t entry_len)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t size;
 
     /* Use signed INT64_MAX as the maximum even for uint64_t header fields,
@@ -467,6 +467,11 @@ static QemuOptsList qcow2_runtime_opts = {
             .type = QEMU_OPT_SIZE,
             .help = "Maximum refcount block cache size",
         },
+        {
+            .name = QCOW2_OPT_CACHE_CLEAN_INTERVAL,
+            .type = QEMU_OPT_NUMBER,
+            .help = "Clean unused cache entries after this time (in seconds)",
+        },
         { /* end of list */ }
     },
 };
@@ -482,11 +487,54 @@ static const char *overlap_bool_option_names[QCOW2_OL_MAX_BITNR] = {
     [QCOW2_OL_INACTIVE_L2_BITNR]    = QCOW2_OPT_OVERLAP_INACTIVE_L2,
 };
 
+static void cache_clean_timer_cb(void *opaque)
+{
+    BlockDriverState *bs = opaque;
+    BDRVQcow2State *s = bs->opaque;
+    qcow2_cache_clean_unused(bs, s->l2_table_cache);
+    qcow2_cache_clean_unused(bs, s->refcount_block_cache);
+    timer_mod(s->cache_clean_timer, qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) +
+              (int64_t) s->cache_clean_interval * 1000);
+}
+
+static void cache_clean_timer_init(BlockDriverState *bs, AioContext *context)
+{
+    BDRVQcow2State *s = bs->opaque;
+    if (s->cache_clean_interval > 0) {
+        s->cache_clean_timer = aio_timer_new(context, QEMU_CLOCK_VIRTUAL,
+                                             SCALE_MS, cache_clean_timer_cb,
+                                             bs);
+        timer_mod(s->cache_clean_timer, qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) +
+                  (int64_t) s->cache_clean_interval * 1000);
+    }
+}
+
+static void cache_clean_timer_del(BlockDriverState *bs)
+{
+    BDRVQcow2State *s = bs->opaque;
+    if (s->cache_clean_timer) {
+        timer_del(s->cache_clean_timer);
+        timer_free(s->cache_clean_timer);
+        s->cache_clean_timer = NULL;
+    }
+}
+
+static void qcow2_detach_aio_context(BlockDriverState *bs)
+{
+    cache_clean_timer_del(bs);
+}
+
+static void qcow2_attach_aio_context(BlockDriverState *bs,
+                                     AioContext *new_context)
+{
+    cache_clean_timer_init(bs, new_context);
+}
+
 static void read_cache_sizes(BlockDriverState *bs, QemuOpts *opts,
                              uint64_t *l2_cache_size,
                              uint64_t *refcount_cache_size, Error **errp)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t combined_cache_size;
     bool l2_cache_size_set, refcount_cache_size_set, combined_cache_size_set;
 
@@ -541,22 +589,246 @@ static void read_cache_sizes(BlockDriverState *bs, QemuOpts *opts,
     }
 }
 
+typedef struct Qcow2ReopenState {
+    Qcow2Cache *l2_table_cache;
+    Qcow2Cache *refcount_block_cache;
+    bool use_lazy_refcounts;
+    int overlap_check;
+    bool discard_passthrough[QCOW2_DISCARD_MAX];
+    uint64_t cache_clean_interval;
+} Qcow2ReopenState;
+
+static int qcow2_update_options_prepare(BlockDriverState *bs,
+                                        Qcow2ReopenState *r,
+                                        QDict *options, int flags,
+                                        Error **errp)
+{
+    BDRVQcow2State *s = bs->opaque;
+    QemuOpts *opts = NULL;
+    const char *opt_overlap_check, *opt_overlap_check_template;
+    int overlap_check_template = 0;
+    uint64_t l2_cache_size, refcount_cache_size;
+    int i;
+    Error *local_err = NULL;
+    int ret;
+
+    opts = qemu_opts_create(&qcow2_runtime_opts, NULL, 0, &error_abort);
+    qemu_opts_absorb_qdict(opts, options, &local_err);
+    if (local_err) {
+        error_propagate(errp, local_err);
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    /* get L2 table/refcount block cache size from command line options */
+    read_cache_sizes(bs, opts, &l2_cache_size, &refcount_cache_size,
+                     &local_err);
+    if (local_err) {
+        error_propagate(errp, local_err);
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    l2_cache_size /= s->cluster_size;
+    if (l2_cache_size < MIN_L2_CACHE_SIZE) {
+        l2_cache_size = MIN_L2_CACHE_SIZE;
+    }
+    if (l2_cache_size > INT_MAX) {
+        error_setg(errp, "L2 cache size too big");
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    refcount_cache_size /= s->cluster_size;
+    if (refcount_cache_size < MIN_REFCOUNT_CACHE_SIZE) {
+        refcount_cache_size = MIN_REFCOUNT_CACHE_SIZE;
+    }
+    if (refcount_cache_size > INT_MAX) {
+        error_setg(errp, "Refcount cache size too big");
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    /* alloc new L2 table/refcount block cache, flush old one */
+    if (s->l2_table_cache) {
+        ret = qcow2_cache_flush(bs, s->l2_table_cache);
+        if (ret) {
+            error_setg_errno(errp, -ret, "Failed to flush the L2 table cache");
+            goto fail;
+        }
+    }
+
+    if (s->refcount_block_cache) {
+        ret = qcow2_cache_flush(bs, s->refcount_block_cache);
+        if (ret) {
+            error_setg_errno(errp, -ret,
+                             "Failed to flush the refcount block cache");
+            goto fail;
+        }
+    }
+
+    r->l2_table_cache = qcow2_cache_create(bs, l2_cache_size);
+    r->refcount_block_cache = qcow2_cache_create(bs, refcount_cache_size);
+    if (r->l2_table_cache == NULL || r->refcount_block_cache == NULL) {
+        error_setg(errp, "Could not allocate metadata caches");
+        ret = -ENOMEM;
+        goto fail;
+    }
+
+    /* New interval for cache cleanup timer */
+    r->cache_clean_interval =
+        qemu_opt_get_number(opts, QCOW2_OPT_CACHE_CLEAN_INTERVAL,
+                            s->cache_clean_interval);
+    if (r->cache_clean_interval > UINT_MAX) {
+        error_setg(errp, "Cache clean interval too big");
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    /* lazy-refcounts; flush if going from enabled to disabled */
+    r->use_lazy_refcounts = qemu_opt_get_bool(opts, QCOW2_OPT_LAZY_REFCOUNTS,
+        (s->compatible_features & QCOW2_COMPAT_LAZY_REFCOUNTS));
+    if (r->use_lazy_refcounts && s->qcow_version < 3) {
+        error_setg(errp, "Lazy refcounts require a qcow2 image with at least "
+                   "qemu 1.1 compatibility level");
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    if (s->use_lazy_refcounts && !r->use_lazy_refcounts) {
+        ret = qcow2_mark_clean(bs);
+        if (ret < 0) {
+            error_setg_errno(errp, -ret, "Failed to disable lazy refcounts");
+            goto fail;
+        }
+    }
+
+    /* Overlap check options */
+    opt_overlap_check = qemu_opt_get(opts, QCOW2_OPT_OVERLAP);
+    opt_overlap_check_template = qemu_opt_get(opts, QCOW2_OPT_OVERLAP_TEMPLATE);
+    if (opt_overlap_check_template && opt_overlap_check &&
+        strcmp(opt_overlap_check_template, opt_overlap_check))
+    {
+        error_setg(errp, "Conflicting values for qcow2 options '"
+                   QCOW2_OPT_OVERLAP "' ('%s') and '" QCOW2_OPT_OVERLAP_TEMPLATE
+                   "' ('%s')", opt_overlap_check, opt_overlap_check_template);
+        ret = -EINVAL;
+        goto fail;
+    }
+    if (!opt_overlap_check) {
+        opt_overlap_check = opt_overlap_check_template ?: "cached";
+    }
+
+    if (!strcmp(opt_overlap_check, "none")) {
+        overlap_check_template = 0;
+    } else if (!strcmp(opt_overlap_check, "constant")) {
+        overlap_check_template = QCOW2_OL_CONSTANT;
+    } else if (!strcmp(opt_overlap_check, "cached")) {
+        overlap_check_template = QCOW2_OL_CACHED;
+    } else if (!strcmp(opt_overlap_check, "all")) {
+        overlap_check_template = QCOW2_OL_ALL;
+    } else {
+        error_setg(errp, "Unsupported value '%s' for qcow2 option "
+                   "'overlap-check'. Allowed are any of the following: "
+                   "none, constant, cached, all", opt_overlap_check);
+        ret = -EINVAL;
+        goto fail;
+    }
+
+    r->overlap_check = 0;
+    for (i = 0; i < QCOW2_OL_MAX_BITNR; i++) {
+        /* overlap-check defines a template bitmask, but every flag may be
+         * overwritten through the associated boolean option */
+        r->overlap_check |=
+            qemu_opt_get_bool(opts, overlap_bool_option_names[i],
+                              overlap_check_template & (1 << i)) << i;
+    }
+
+    r->discard_passthrough[QCOW2_DISCARD_NEVER] = false;
+    r->discard_passthrough[QCOW2_DISCARD_ALWAYS] = true;
+    r->discard_passthrough[QCOW2_DISCARD_REQUEST] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_REQUEST,
+                          flags & BDRV_O_UNMAP);
+    r->discard_passthrough[QCOW2_DISCARD_SNAPSHOT] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_SNAPSHOT, true);
+    r->discard_passthrough[QCOW2_DISCARD_OTHER] =
+        qemu_opt_get_bool(opts, QCOW2_OPT_DISCARD_OTHER, false);
+
+    ret = 0;
+fail:
+    qemu_opts_del(opts);
+    opts = NULL;
+    return ret;
+}
+
+static void qcow2_update_options_commit(BlockDriverState *bs,
+                                        Qcow2ReopenState *r)
+{
+    BDRVQcow2State *s = bs->opaque;
+    int i;
+
+    if (s->l2_table_cache) {
+        qcow2_cache_destroy(bs, s->l2_table_cache);
+    }
+    if (s->refcount_block_cache) {
+        qcow2_cache_destroy(bs, s->refcount_block_cache);
+    }
+    s->l2_table_cache = r->l2_table_cache;
+    s->refcount_block_cache = r->refcount_block_cache;
+
+    s->overlap_check = r->overlap_check;
+    s->use_lazy_refcounts = r->use_lazy_refcounts;
+
+    for (i = 0; i < QCOW2_DISCARD_MAX; i++) {
+        s->discard_passthrough[i] = r->discard_passthrough[i];
+    }
+
+    if (s->cache_clean_interval != r->cache_clean_interval) {
+        cache_clean_timer_del(bs);
+        s->cache_clean_interval = r->cache_clean_interval;
+        cache_clean_timer_init(bs, bdrv_get_aio_context(bs));
+    }
+}
+
+static void qcow2_update_options_abort(BlockDriverState *bs,
+                                       Qcow2ReopenState *r)
+{
+    if (r->l2_table_cache) {
+        qcow2_cache_destroy(bs, r->l2_table_cache);
+    }
+    if (r->refcount_block_cache) {
+        qcow2_cache_destroy(bs, r->refcount_block_cache);
+    }
+}
+
+static int qcow2_update_options(BlockDriverState *bs, QDict *options,
+                                int flags, Error **errp)
+{
+    Qcow2ReopenState r = {};
+    int ret;
+
+    ret = qcow2_update_options_prepare(bs, &r, options, flags, errp);
+    if (ret >= 0) {
+        qcow2_update_options_commit(bs, &r);
+    } else {
+        qcow2_update_options_abort(bs, &r);
+    }
+
+    return ret;
+}
+
 static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
                       Error **errp)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     unsigned int len, i;
     int ret = 0;
     QCowHeader header;
-    QemuOpts *opts = NULL;
     Error *local_err = NULL;
     uint64_t ext_end;
     uint64_t l1_vm_state_index;
-    const char *opt_overlap_check, *opt_overlap_check_template;
-    int overlap_check_template = 0;
-    uint64_t l2_cache_size, refcount_cache_size;
 
-    ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
+    ret = bdrv_pread(bs->file->bs, 0, &header, sizeof(header));
     if (ret < 0) {
         error_setg_errno(errp, -ret, "Could not read qcow2 header");
         goto fail;
@@ -631,7 +903,7 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
     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,
+        ret = bdrv_pread(bs->file->bs, sizeof(header), s->unknown_header_fields,
                          s->unknown_header_fields_size);
         if (ret < 0) {
             error_setg_errno(errp, -ret, "Could not read unknown qcow2 header "
@@ -784,14 +1056,14 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
 
 
     if (s->l1_size > 0) {
-        s->l1_table = qemu_try_blockalign(bs->file,
+        s->l1_table = qemu_try_blockalign(bs->file->bs,
             align_offset(s->l1_size * sizeof(uint64_t), 512));
         if (s->l1_table == NULL) {
             error_setg(errp, "Could not allocate L1 table");
             ret = -ENOMEM;
             goto fail;
         }
-        ret = bdrv_pread(bs->file, s->l1_table_offset, s->l1_table,
+        ret = bdrv_pread(bs->file->bs, s->l1_table_offset, s->l1_table,
                          s->l1_size * sizeof(uint64_t));
         if (ret < 0) {
             error_setg_errno(errp, -ret, "Could not read L1 table");
@@ -802,55 +1074,15 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
         }
     }
 
-    /* get L2 table/refcount block cache size from command line options */
-    opts = qemu_opts_create(&qcow2_runtime_opts, NULL, 0, &error_abort);
-    qemu_opts_absorb_qdict(opts, options, &local_err);
-    if (local_err) {
-        error_propagate(errp, local_err);
-        ret = -EINVAL;
-        goto fail;
-    }
-
-    read_cache_sizes(bs, opts, &l2_cache_size, &refcount_cache_size,
-                     &local_err);
-    if (local_err) {
-        error_propagate(errp, local_err);
-        ret = -EINVAL;
-        goto fail;
-    }
-
-    l2_cache_size /= s->cluster_size;
-    if (l2_cache_size < MIN_L2_CACHE_SIZE) {
-        l2_cache_size = MIN_L2_CACHE_SIZE;
-    }
-    if (l2_cache_size > INT_MAX) {
-        error_setg(errp, "L2 cache size too big");
-        ret = -EINVAL;
-        goto fail;
-    }
-
-    refcount_cache_size /= s->cluster_size;
-    if (refcount_cache_size < MIN_REFCOUNT_CACHE_SIZE) {
-        refcount_cache_size = MIN_REFCOUNT_CACHE_SIZE;
-    }
-    if (refcount_cache_size > INT_MAX) {
-        error_setg(errp, "Refcount cache size too big");
-        ret = -EINVAL;
-        goto fail;
-    }
-
-    /* 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);
-    if (s->l2_table_cache == NULL || s->refcount_block_cache == NULL) {
-        error_setg(errp, "Could not allocate metadata caches");
-        ret = -ENOMEM;
+    /* Parse driver-specific options */
+    ret = qcow2_update_options(bs, options, flags, errp);
+    if (ret < 0) {
         goto fail;
     }
 
     s->cluster_cache = g_malloc(s->cluster_size);
     /* one more sector for decompressed data alignment */
-    s->cluster_data = qemu_try_blockalign(bs->file, QCOW_MAX_CRYPT_CLUSTERS
+    s->cluster_data = qemu_try_blockalign(bs->file->bs, QCOW_MAX_CRYPT_CLUSTERS
                                                     * s->cluster_size + 512);
     if (s->cluster_data == NULL) {
         error_setg(errp, "Could not allocate temporary cluster buffer");
@@ -887,7 +1119,7 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
             ret = -EINVAL;
             goto fail;
         }
-        ret = bdrv_pread(bs->file, header.backing_file_offset,
+        ret = bdrv_pread(bs->file->bs, header.backing_file_offset,
                          bs->backing_file, len);
         if (ret < 0) {
             error_setg_errno(errp, -ret, "Could not read backing file name");
@@ -932,70 +1164,6 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
         }
     }
 
-    /* Enable lazy_refcounts according to image and command line options */
-    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);
-
-    opt_overlap_check = qemu_opt_get(opts, QCOW2_OPT_OVERLAP);
-    opt_overlap_check_template = qemu_opt_get(opts, QCOW2_OPT_OVERLAP_TEMPLATE);
-    if (opt_overlap_check_template && opt_overlap_check &&
-        strcmp(opt_overlap_check_template, opt_overlap_check))
-    {
-        error_setg(errp, "Conflicting values for qcow2 options '"
-                   QCOW2_OPT_OVERLAP "' ('%s') and '" QCOW2_OPT_OVERLAP_TEMPLATE
-                   "' ('%s')", opt_overlap_check, opt_overlap_check_template);
-        ret = -EINVAL;
-        goto fail;
-    }
-    if (!opt_overlap_check) {
-        opt_overlap_check = opt_overlap_check_template ?: "cached";
-    }
-
-    if (!strcmp(opt_overlap_check, "none")) {
-        overlap_check_template = 0;
-    } else if (!strcmp(opt_overlap_check, "constant")) {
-        overlap_check_template = QCOW2_OL_CONSTANT;
-    } else if (!strcmp(opt_overlap_check, "cached")) {
-        overlap_check_template = QCOW2_OL_CACHED;
-    } else if (!strcmp(opt_overlap_check, "all")) {
-        overlap_check_template = QCOW2_OL_ALL;
-    } else {
-        error_setg(errp, "Unsupported value '%s' for qcow2 option "
-                   "'overlap-check'. Allowed are either of the following: "
-                   "none, constant, cached, all", opt_overlap_check);
-        ret = -EINVAL;
-        goto fail;
-    }
-
-    s->overlap_check = 0;
-    for (i = 0; i < QCOW2_OL_MAX_BITNR; i++) {
-        /* overlap-check defines a template bitmask, but every flag may be
-         * overwritten through the associated boolean option */
-        s->overlap_check |=
-            qemu_opt_get_bool(opts, overlap_bool_option_names[i],
-                              overlap_check_template & (1 << i)) << i;
-    }
-
-    qemu_opts_del(opts);
-    opts = NULL;
-
-    if (s->use_lazy_refcounts && s->qcow_version < 3) {
-        error_setg(errp, "Lazy refcounts require a qcow2 image with at least "
-                   "qemu 1.1 compatibility level");
-        ret = -EINVAL;
-        goto fail;
-    }
-
 #ifdef DEBUG_ALLOC
     {
         BdrvCheckResult result = {0};
@@ -1005,7 +1173,6 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
     return ret;
 
  fail:
-    qemu_opts_del(opts);
     g_free(s->unknown_header_fields);
     cleanup_unknown_header_ext(bs);
     qcow2_free_snapshots(bs);
@@ -1013,6 +1180,7 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
     qemu_vfree(s->l1_table);
     /* else pre-write overlap checks in cache_destroy may crash */
     s->l1_table = NULL;
+    cache_clean_timer_del(bs);
     if (s->l2_table_cache) {
         qcow2_cache_destroy(bs, s->l2_table_cache);
     }
@@ -1026,14 +1194,14 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags,
 
 static void qcow2_refresh_limits(BlockDriverState *bs, Error **errp)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     bs->bl.write_zeroes_alignment = s->cluster_sectors;
 }
 
 static int qcow2_set_key(BlockDriverState *bs, const char *key)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint8_t keybuf[16];
     int len, i;
     Error *err = NULL;
@@ -1066,32 +1234,58 @@ static int qcow2_set_key(BlockDriverState *bs, const char *key)
     return 0;
 }
 
-/* We have no actual commit/abort logic for qcow2, but we need to write out any
- * unwritten data if we reopen read-only. */
 static int qcow2_reopen_prepare(BDRVReopenState *state,
                                 BlockReopenQueue *queue, Error **errp)
 {
+    Qcow2ReopenState *r;
     int ret;
 
+    r = g_new0(Qcow2ReopenState, 1);
+    state->opaque = r;
+
+    ret = qcow2_update_options_prepare(state->bs, r, state->options,
+                                       state->flags, errp);
+    if (ret < 0) {
+        goto fail;
+    }
+
+    /* We need to write out any unwritten data if we reopen read-only. */
     if ((state->flags & BDRV_O_RDWR) == 0) {
         ret = bdrv_flush(state->bs);
         if (ret < 0) {
-            return ret;
+            goto fail;
         }
 
         ret = qcow2_mark_clean(state->bs);
         if (ret < 0) {
-            return ret;
+            goto fail;
         }
     }
 
     return 0;
+
+fail:
+    qcow2_update_options_abort(state->bs, r);
+    g_free(r);
+    return ret;
+}
+
+static void qcow2_reopen_commit(BDRVReopenState *state)
+{
+    qcow2_update_options_commit(state->bs, state->opaque);
+    g_free(state->opaque);
+}
+
+static void qcow2_reopen_abort(BDRVReopenState *state)
+{
+    qcow2_update_options_abort(state->bs, state->opaque);
+    g_free(state->opaque);
 }
 
 static int64_t coroutine_fn qcow2_co_get_block_status(BlockDriverState *bs,
         int64_t sector_num, int nb_sectors, int *pnum)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t cluster_offset;
     int index_in_cluster, ret;
     int64_t status = 0;
@@ -1138,7 +1332,7 @@ int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov,
 static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
                           int remaining_sectors, QEMUIOVector *qiov)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int index_in_cluster, n1;
     int ret;
     int cur_nr_sectors; /* number of sectors in current iteration */
@@ -1175,9 +1369,9 @@ static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
         switch (ret) {
         case QCOW2_CLUSTER_UNALLOCATED:
 
-            if (bs->backing_hd) {
+            if (bs->backing) {
                 /* read from the base image */
-                n1 = qcow2_backing_read1(bs->backing_hd, &hd_qiov,
+                n1 = qcow2_backing_read1(bs->backing->bs, &hd_qiov,
                     sector_num, cur_nr_sectors);
                 if (n1 > 0) {
                     QEMUIOVector local_qiov;
@@ -1188,7 +1382,7 @@ static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
 
                     BLKDBG_EVENT(bs->file, BLKDBG_READ_BACKING_AIO);
                     qemu_co_mutex_unlock(&s->lock);
-                    ret = bdrv_co_readv(bs->backing_hd, sector_num,
+                    ret = bdrv_co_readv(bs->backing->bs, sector_num,
                                         n1, &local_qiov);
                     qemu_co_mutex_lock(&s->lock);
 
@@ -1235,8 +1429,9 @@ static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
                  */
                 if (!cluster_data) {
                     cluster_data =
-                        qemu_try_blockalign(bs->file, QCOW_MAX_CRYPT_CLUSTERS
-                                                      * s->cluster_size);
+                        qemu_try_blockalign(bs->file->bs,
+                                            QCOW_MAX_CRYPT_CLUSTERS
+                                            * s->cluster_size);
                     if (cluster_data == NULL) {
                         ret = -ENOMEM;
                         goto fail;
@@ -1252,7 +1447,7 @@ static coroutine_fn int qcow2_co_readv(BlockDriverState *bs, int64_t sector_num,
 
             BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
             qemu_co_mutex_unlock(&s->lock);
-            ret = bdrv_co_readv(bs->file,
+            ret = bdrv_co_readv(bs->file->bs,
                                 (cluster_offset >> 9) + index_in_cluster,
                                 cur_nr_sectors, &hd_qiov);
             qemu_co_mutex_lock(&s->lock);
@@ -1300,7 +1495,7 @@ static coroutine_fn int qcow2_co_writev(BlockDriverState *bs,
                            int remaining_sectors,
                            QEMUIOVector *qiov)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int index_in_cluster;
     int ret;
     int cur_nr_sectors; /* number of sectors in current iteration */
@@ -1349,7 +1544,7 @@ static coroutine_fn int qcow2_co_writev(BlockDriverState *bs,
             Error *err = NULL;
             assert(s->cipher);
             if (!cluster_data) {
-                cluster_data = qemu_try_blockalign(bs->file,
+                cluster_data = qemu_try_blockalign(bs->file->bs,
                                                    QCOW_MAX_CRYPT_CLUSTERS
                                                    * s->cluster_size);
                 if (cluster_data == NULL) {
@@ -1386,7 +1581,7 @@ static coroutine_fn int qcow2_co_writev(BlockDriverState *bs,
         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,
+        ret = bdrv_co_writev(bs->file->bs,
                              (cluster_offset >> 9) + index_in_cluster,
                              cur_nr_sectors, &hd_qiov);
         qemu_co_mutex_lock(&s->lock);
@@ -1446,7 +1641,7 @@ fail:
 
 static void qcow2_close(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     qemu_vfree(s->l1_table);
     /* else pre-write overlap checks in cache_destroy may crash */
     s->l1_table = NULL;
@@ -1471,6 +1666,7 @@ static void qcow2_close(BlockDriverState *bs)
         }
     }
 
+    cache_clean_timer_del(bs);
     qcow2_cache_destroy(bs, s->l2_table_cache);
     qcow2_cache_destroy(bs, s->refcount_block_cache);
 
@@ -1491,7 +1687,7 @@ static void qcow2_close(BlockDriverState *bs)
 
 static void qcow2_invalidate_cache(BlockDriverState *bs, Error **errp)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int flags = s->flags;
     QCryptoCipher *cipher = NULL;
     QDict *options;
@@ -1508,13 +1704,13 @@ static void qcow2_invalidate_cache(BlockDriverState *bs, Error **errp)
 
     qcow2_close(bs);
 
-    bdrv_invalidate_cache(bs->file, &local_err);
+    bdrv_invalidate_cache(bs->file->bs, &local_err);
     if (local_err) {
         error_propagate(errp, local_err);
         return;
     }
 
-    memset(s, 0, sizeof(BDRVQcowState));
+    memset(s, 0, sizeof(BDRVQcow2State));
     options = qdict_clone_shallow(bs->options);
 
     ret = qcow2_open(bs, options, flags, &local_err);
@@ -1561,7 +1757,7 @@ static size_t header_ext_add(char *buf, uint32_t magic, const void *s,
  */
 int qcow2_update_header(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     QCowHeader *header;
     char *buf;
     size_t buflen = s->cluster_size;
@@ -1716,7 +1912,7 @@ int qcow2_update_header(BlockDriverState *bs)
     }
 
     /* Write the new header */
-    ret = bdrv_pwrite(bs->file, 0, header, s->cluster_size);
+    ret = bdrv_pwrite(bs->file->bs, 0, header, s->cluster_size);
     if (ret < 0) {
         goto fail;
     }
@@ -1730,7 +1926,7 @@ fail:
 static int qcow2_change_backing_file(BlockDriverState *bs,
     const char *backing_file, const char *backing_fmt)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     pstrcpy(bs->backing_file, sizeof(bs->backing_file), backing_file ?: "");
     pstrcpy(bs->backing_format, sizeof(bs->backing_format), backing_fmt ?: "");
@@ -1796,7 +1992,8 @@ static int preallocate(BlockDriverState *bs)
     if (host_offset != 0) {
         uint8_t buf[BDRV_SECTOR_SIZE];
         memset(buf, 0, BDRV_SECTOR_SIZE);
-        ret = bdrv_write(bs->file, (host_offset >> BDRV_SECTOR_BITS) + num - 1,
+        ret = bdrv_write(bs->file->bs,
+                         (host_offset >> BDRV_SECTOR_BITS) + num - 1,
                          buf, 1);
         if (ret < 0) {
             return ret;
@@ -1812,8 +2009,10 @@ static int qcow2_create2(const char *filename, int64_t total_size,
                          QemuOpts *opts, int version, int refcount_order,
                          Error **errp)
 {
-    /* Calculate cluster_bits */
     int cluster_bits;
+    QDict *options;
+
+    /* Calculate cluster_bits */
     cluster_bits = ctz32(cluster_size);
     if (cluster_bits < MIN_CLUSTER_BITS || cluster_bits > MAX_CLUSTER_BITS ||
         (1 << cluster_bits) != cluster_size)
@@ -1912,7 +2111,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
 
     bs = NULL;
     ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
-                    NULL, &local_err);
+                    &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         return ret;
@@ -1971,9 +2170,11 @@ static int qcow2_create2(const char *filename, int64_t total_size,
      * refcount of the cluster that is occupied by the header and the refcount
      * table)
      */
-    ret = bdrv_open(&bs, filename, NULL, NULL,
+    options = qdict_new();
+    qdict_put(options, "driver", qstring_from_str("qcow2"));
+    ret = bdrv_open(&bs, filename, NULL, options,
                     BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_NO_FLUSH,
-                    &bdrv_qcow2, &local_err);
+                    &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto out;
@@ -2009,7 +2210,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
 
     /* And if we're supposed to preallocate metadata, do that now */
     if (prealloc != PREALLOC_MODE_OFF) {
-        BDRVQcowState *s = bs->opaque;
+        BDRVQcow2State *s = bs->opaque;
         qemu_co_mutex_lock(&s->lock);
         ret = preallocate(bs);
         qemu_co_mutex_unlock(&s->lock);
@@ -2023,9 +2224,11 @@ static int qcow2_create2(const char *filename, int64_t total_size,
     bs = NULL;
 
     /* Reopen the image without BDRV_O_NO_FLUSH to flush it before returning */
-    ret = bdrv_open(&bs, filename, NULL, NULL,
+    options = qdict_new();
+    qdict_put(options, "driver", qstring_from_str("qcow2"));
+    ret = bdrv_open(&bs, filename, NULL, options,
                     BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_NO_BACKING,
-                    &bdrv_qcow2, &local_err);
+                    &local_err);
     if (local_err) {
         error_propagate(errp, local_err);
         goto out;
@@ -2142,7 +2345,7 @@ static coroutine_fn int qcow2_co_write_zeroes(BlockDriverState *bs,
     int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
 {
     int ret;
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     /* Emulate misaligned zero writes */
     if (sector_num % s->cluster_sectors || nb_sectors % s->cluster_sectors) {
@@ -2162,7 +2365,7 @@ static coroutine_fn int qcow2_co_discard(BlockDriverState *bs,
     int64_t sector_num, int nb_sectors)
 {
     int ret;
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
 
     qemu_co_mutex_lock(&s->lock);
     ret = qcow2_discard_clusters(bs, sector_num << BDRV_SECTOR_BITS,
@@ -2173,7 +2376,7 @@ static coroutine_fn int qcow2_co_discard(BlockDriverState *bs,
 
 static int qcow2_truncate(BlockDriverState *bs, int64_t offset)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t new_l1_size;
     int ret;
 
@@ -2202,7 +2405,7 @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset)
 
     /* write updated header.size */
     offset = cpu_to_be64(offset);
-    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, size),
+    ret = bdrv_pwrite_sync(bs->file->bs, offsetof(QCowHeader, size),
                            &offset, sizeof(uint64_t));
     if (ret < 0) {
         return ret;
@@ -2217,7 +2420,7 @@ static int qcow2_truncate(BlockDriverState *bs, int64_t offset)
 static int qcow2_write_compressed(BlockDriverState *bs, int64_t sector_num,
                                   const uint8_t *buf, int nb_sectors)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     z_stream strm;
     int ret, out_len;
     uint8_t *out_buf;
@@ -2226,8 +2429,8 @@ static int qcow2_write_compressed(BlockDriverState *bs, int64_t sector_num,
     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);
-        return bdrv_truncate(bs->file, cluster_offset);
+        cluster_offset = bdrv_getlength(bs->file->bs);
+        return bdrv_truncate(bs->file->bs, cluster_offset);
     }
 
     if (nb_sectors != s->cluster_sectors) {
@@ -2294,7 +2497,7 @@ static int qcow2_write_compressed(BlockDriverState *bs, int64_t sector_num,
         }
 
         BLKDBG_EVENT(bs->file, BLKDBG_WRITE_COMPRESSED);
-        ret = bdrv_pwrite(bs->file, cluster_offset, out_buf, out_len);
+        ret = bdrv_pwrite(bs->file->bs, cluster_offset, out_buf, out_len);
         if (ret < 0) {
             goto fail;
         }
@@ -2308,7 +2511,7 @@ fail:
 
 static int make_completely_empty(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret, l1_clusters;
     int64_t offset;
     uint64_t *new_reftable = NULL;
@@ -2343,7 +2546,7 @@ static int make_completely_empty(BlockDriverState *bs)
     /* After this call, neither the in-memory nor the on-disk refcount
      * information accurately describe the actual references */
 
-    ret = bdrv_write_zeroes(bs->file, s->l1_table_offset / BDRV_SECTOR_SIZE,
+    ret = bdrv_write_zeroes(bs->file->bs, s->l1_table_offset / BDRV_SECTOR_SIZE,
                             l1_clusters * s->cluster_sectors, 0);
     if (ret < 0) {
         goto fail_broken_refcounts;
@@ -2357,7 +2560,7 @@ static int make_completely_empty(BlockDriverState *bs)
      * overwrite parts of the existing refcount and L1 table, which is not
      * an issue because the dirty flag is set, complete data loss is in fact
      * desired and partial data loss is consequently fine as well */
-    ret = bdrv_write_zeroes(bs->file, s->cluster_size / BDRV_SECTOR_SIZE,
+    ret = bdrv_write_zeroes(bs->file->bs, s->cluster_size / BDRV_SECTOR_SIZE,
                             (2 + l1_clusters) * s->cluster_size /
                             BDRV_SECTOR_SIZE, 0);
     /* This call (even if it failed overall) may have overwritten on-disk
@@ -2377,7 +2580,7 @@ static int make_completely_empty(BlockDriverState *bs)
     cpu_to_be64w(&l1_ofs_rt_ofs_cls.l1_offset, 3 * s->cluster_size);
     cpu_to_be64w(&l1_ofs_rt_ofs_cls.reftable_offset, s->cluster_size);
     cpu_to_be32w(&l1_ofs_rt_ofs_cls.reftable_clusters, 1);
-    ret = bdrv_pwrite_sync(bs->file, offsetof(QCowHeader, l1_table_offset),
+    ret = bdrv_pwrite_sync(bs->file->bs, offsetof(QCowHeader, l1_table_offset),
                            &l1_ofs_rt_ofs_cls, sizeof(l1_ofs_rt_ofs_cls));
     if (ret < 0) {
         goto fail_broken_refcounts;
@@ -2408,7 +2611,7 @@ static int make_completely_empty(BlockDriverState *bs)
 
     /* Enter the first refblock into the reftable */
     rt_entry = cpu_to_be64(2 * s->cluster_size);
-    ret = bdrv_pwrite_sync(bs->file, s->cluster_size,
+    ret = bdrv_pwrite_sync(bs->file->bs, s->cluster_size,
                            &rt_entry, sizeof(rt_entry));
     if (ret < 0) {
         goto fail_broken_refcounts;
@@ -2433,7 +2636,7 @@ static int make_completely_empty(BlockDriverState *bs)
         goto fail;
     }
 
-    ret = bdrv_truncate(bs->file, (3 + l1_clusters) * s->cluster_size);
+    ret = bdrv_truncate(bs->file->bs, (3 + l1_clusters) * s->cluster_size);
     if (ret < 0) {
         goto fail;
     }
@@ -2456,7 +2659,7 @@ fail:
 
 static int qcow2_make_empty(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     uint64_t start_sector;
     int sector_step = INT_MAX / BDRV_SECTOR_SIZE;
     int l1_clusters, ret = 0;
@@ -2497,7 +2700,7 @@ static int qcow2_make_empty(BlockDriverState *bs)
 
 static coroutine_fn int qcow2_co_flush_to_os(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int ret;
 
     qemu_co_mutex_lock(&s->lock);
@@ -2521,7 +2724,7 @@ static coroutine_fn int qcow2_co_flush_to_os(BlockDriverState *bs)
 
 static int qcow2_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     bdi->unallocated_blocks_are_zero = true;
     bdi->can_write_zeroes_with_unmap = (s->qcow_version >= 3);
     bdi->cluster_size = s->cluster_size;
@@ -2531,22 +2734,20 @@ static int qcow2_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
 
 static ImageInfoSpecific *qcow2_get_specific_info(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     ImageInfoSpecific *spec_info = g_new(ImageInfoSpecific, 1);
 
     *spec_info = (ImageInfoSpecific){
-        .kind  = IMAGE_INFO_SPECIFIC_KIND_QCOW2,
-        {
-            .qcow2 = g_new(ImageInfoSpecificQCow2, 1),
-        },
+        .type  = IMAGE_INFO_SPECIFIC_KIND_QCOW2,
+        .u.qcow2 = g_new(ImageInfoSpecificQCow2, 1),
     };
     if (s->qcow_version == 2) {
-        *spec_info->qcow2 = (ImageInfoSpecificQCow2){
+        *spec_info->u.qcow2 = (ImageInfoSpecificQCow2){
             .compat             = g_strdup("0.10"),
             .refcount_bits      = s->refcount_bits,
         };
     } else if (s->qcow_version == 3) {
-        *spec_info->qcow2 = (ImageInfoSpecificQCow2){
+        *spec_info->u.qcow2 = (ImageInfoSpecificQCow2){
             .compat             = g_strdup("1.1"),
             .lazy_refcounts     = s->compatible_features &
                                   QCOW2_COMPAT_LAZY_REFCOUNTS,
@@ -2564,11 +2765,11 @@ static ImageInfoSpecific *qcow2_get_specific_info(BlockDriverState *bs)
 #if 0
 static void dump_refcounts(BlockDriverState *bs)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t nb_clusters, k, k1, size;
     int refcount;
 
-    size = bdrv_getlength(bs->file);
+    size = bdrv_getlength(bs->file->bs);
     nb_clusters = size_to_clusters(s, size);
     for(k = 0; k < nb_clusters;) {
         k1 = k;
@@ -2585,7 +2786,7 @@ static void dump_refcounts(BlockDriverState *bs)
 static int qcow2_save_vmstate(BlockDriverState *bs, QEMUIOVector *qiov,
                               int64_t pos)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int64_t total_sectors = bs->total_sectors;
     bool zero_beyond_eof = bs->zero_beyond_eof;
     int ret;
@@ -2606,7 +2807,7 @@ static int qcow2_save_vmstate(BlockDriverState *bs, QEMUIOVector *qiov,
 static int qcow2_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                               int64_t pos, int size)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     bool zero_beyond_eof = bs->zero_beyond_eof;
     int ret;
 
@@ -2625,7 +2826,7 @@ static int qcow2_load_vmstate(BlockDriverState *bs, uint8_t *buf,
 static int qcow2_downgrade(BlockDriverState *bs, int target_version,
                            BlockDriverAmendStatusCB *status_cb)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int current_version = s->qcow_version;
     int ret;
 
@@ -2689,7 +2890,7 @@ static int qcow2_downgrade(BlockDriverState *bs, int target_version,
 static int qcow2_amend_options(BlockDriverState *bs, QemuOpts *opts,
                                BlockDriverAmendStatusCB *status_cb)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     int old_version = s->qcow_version, new_version = old_version;
     uint64_t new_size = 0;
     const char *backing_file = NULL, *backing_format = NULL;
@@ -2836,7 +3037,7 @@ static int qcow2_amend_options(BlockDriverState *bs, QemuOpts *opts,
 void qcow2_signal_corruption(BlockDriverState *bs, bool fatal, int64_t offset,
                              int64_t size, const char *message_format, ...)
 {
-    BDRVQcowState *s = bs->opaque;
+    BDRVQcow2State *s = bs->opaque;
     const char *node_name;
     char *message;
     va_list ap;
@@ -2937,11 +3138,13 @@ static QemuOptsList qcow2_create_opts = {
 
 BlockDriver bdrv_qcow2 = {
     .format_name        = "qcow2",
-    .instance_size      = sizeof(BDRVQcowState),
+    .instance_size      = sizeof(BDRVQcow2State),
     .bdrv_probe         = qcow2_probe,
     .bdrv_open          = qcow2_open,
     .bdrv_close         = qcow2_close,
     .bdrv_reopen_prepare  = qcow2_reopen_prepare,
+    .bdrv_reopen_commit   = qcow2_reopen_commit,
+    .bdrv_reopen_abort    = qcow2_reopen_abort,
     .bdrv_create        = qcow2_create,
     .bdrv_has_zero_init = bdrv_has_zero_init_1,
     .bdrv_co_get_block_status = qcow2_co_get_block_status,
@@ -2977,6 +3180,9 @@ BlockDriver bdrv_qcow2 = {
     .create_opts         = &qcow2_create_opts,
     .bdrv_check          = qcow2_check,
     .bdrv_amend_options  = qcow2_amend_options,
+
+    .bdrv_detach_aio_context  = qcow2_detach_aio_context,
+    .bdrv_attach_aio_context  = qcow2_attach_aio_context,
 };
 
 static void bdrv_qcow2_init(void)
diff --git a/block/qcow2.h b/block/qcow2.h
index 4b5a6afc8..b8c500b9d 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -26,7 +26,7 @@
 #define BLOCK_QCOW2_H
 
 #include "crypto/cipher.h"
-#include "block/coroutine.h"
+#include "qemu/coroutine.h"
 
 //#define DEBUG_ALLOC
 //#define DEBUG_ALLOC2
@@ -96,6 +96,7 @@
 #define QCOW2_OPT_CACHE_SIZE "cache-size"
 #define QCOW2_OPT_L2_CACHE_SIZE "l2-cache-size"
 #define QCOW2_OPT_REFCOUNT_CACHE_SIZE "refcount-cache-size"
+#define QCOW2_OPT_CACHE_CLEAN_INTERVAL "cache-clean-interval"
 
 typedef struct QCowHeader {
     uint32_t magic;
@@ -221,7 +222,7 @@ typedef uint64_t Qcow2GetRefcountFunc(const void *refcount_array,
 typedef void Qcow2SetRefcountFunc(void *refcount_array,
                                   uint64_t index, uint64_t value);
 
-typedef struct BDRVQcowState {
+typedef struct BDRVQcow2State {
     int cluster_bits;
     int cluster_size;
     int cluster_sectors;
@@ -239,6 +240,8 @@ typedef struct BDRVQcowState {
 
     Qcow2Cache* l2_table_cache;
     Qcow2Cache* refcount_block_cache;
+    QEMUTimer *cache_clean_timer;
+    unsigned cache_clean_interval;
 
     uint8_t *cluster_cache;
     uint8_t *cluster_data;
@@ -290,9 +293,7 @@ typedef struct BDRVQcowState {
      * override) */
     char *image_backing_file;
     char *image_backing_format;
-} BDRVQcowState;
-
-struct QCowAIOCB;
+} BDRVQcow2State;
 
 typedef struct Qcow2COWRegion {
     /**
@@ -402,28 +403,28 @@ typedef enum QCow2MetadataOverlap {
 
 #define REFT_OFFSET_MASK 0xfffffffffffffe00ULL
 
-static inline int64_t start_of_cluster(BDRVQcowState *s, int64_t offset)
+static inline int64_t start_of_cluster(BDRVQcow2State *s, int64_t offset)
 {
     return offset & ~(s->cluster_size - 1);
 }
 
-static inline int64_t offset_into_cluster(BDRVQcowState *s, int64_t offset)
+static inline int64_t offset_into_cluster(BDRVQcow2State *s, int64_t offset)
 {
     return offset & (s->cluster_size - 1);
 }
 
-static inline uint64_t size_to_clusters(BDRVQcowState *s, uint64_t size)
+static inline uint64_t size_to_clusters(BDRVQcow2State *s, uint64_t size)
 {
     return (size + (s->cluster_size - 1)) >> s->cluster_bits;
 }
 
-static inline int64_t size_to_l1(BDRVQcowState *s, int64_t size)
+static inline int64_t size_to_l1(BDRVQcow2State *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)
+static inline int offset_to_l2_index(BDRVQcow2State *s, int64_t offset)
 {
     return (offset >> s->cluster_bits) & (s->l2_size - 1);
 }
@@ -434,12 +435,12 @@ static inline int64_t align_offset(int64_t offset, int n)
     return offset;
 }
 
-static inline int64_t qcow2_vm_state_offset(BDRVQcowState *s)
+static inline int64_t qcow2_vm_state_offset(BDRVQcow2State *s)
 {
     return (int64_t)s->l1_vm_state_index << (s->cluster_bits + s->l2_bits);
 }
 
-static inline uint64_t qcow2_max_refcount_clusters(BDRVQcowState *s)
+static inline uint64_t qcow2_max_refcount_clusters(BDRVQcow2State *s)
 {
     return QCOW_MAX_REFTABLE_SIZE >> s->cluster_bits;
 }
@@ -458,7 +459,7 @@ static inline int qcow2_get_cluster_type(uint64_t l2_entry)
 }
 
 /* Check whether refcounts are eager or lazy */
-static inline bool qcow2_need_accurate_refcounts(BDRVQcowState *s)
+static inline bool qcow2_need_accurate_refcounts(BDRVQcow2State *s)
 {
     return !(s->incompatible_features & QCOW2_INCOMPAT_DIRTY);
 }
@@ -534,7 +535,7 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
 int qcow2_write_l1_entry(BlockDriverState *bs, int l1_index);
 void qcow2_l2_cache_reset(BlockDriverState *bs);
 int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset);
-int qcow2_encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
+int qcow2_encrypt_sectors(BDRVQcow2State *s, int64_t sector_num,
                           uint8_t *out_buf, const uint8_t *in_buf,
                           int nb_sectors, bool enc, Error **errp);
 
@@ -581,6 +582,7 @@ int qcow2_cache_set_dependency(BlockDriverState *bs, Qcow2Cache *c,
     Qcow2Cache *dependency);
 void qcow2_cache_depends_on_flush(Qcow2Cache *c);
 
+void qcow2_cache_clean_unused(BlockDriverState *bs, Qcow2Cache *c);
 int qcow2_cache_empty(BlockDriverState *bs, Qcow2Cache *c);
 
 int qcow2_cache_get(BlockDriverState *bs, Qcow2Cache *c, uint64_t offset,
diff --git a/block/qed-table.c b/block/qed-table.c
index 513aa872c..f4219b8ac 100644
--- a/block/qed-table.c
+++ b/block/qed-table.c
@@ -63,7 +63,7 @@ static void qed_read_table(BDRVQEDState *s, uint64_t offset, QEDTable *table,
     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,
+    bdrv_aio_readv(s->bs->file->bs, offset / BDRV_SECTOR_SIZE, qiov,
                    qiov->size / BDRV_SECTOR_SIZE,
                    qed_read_table_cb, read_table_cb);
 }
@@ -152,7 +152,7 @@ static void qed_write_table(BDRVQEDState *s, uint64_t offset, QEDTable *table,
     /* Adjust for offset into table */
     offset += start * sizeof(uint64_t);
 
-    bdrv_aio_writev(s->bs->file, offset / BDRV_SECTOR_SIZE,
+    bdrv_aio_writev(s->bs->file->bs, offset / BDRV_SECTOR_SIZE,
                     &write_table_cb->qiov,
                     write_table_cb->qiov.size / BDRV_SECTOR_SIZE,
                     qed_write_table_cb, write_table_cb);
diff --git a/block/qed.c b/block/qed.c
index 954ed007c..9b8889503 100644
--- a/block/qed.c
+++ b/block/qed.c
@@ -82,7 +82,7 @@ int qed_write_header_sync(BDRVQEDState *s)
     int ret;
 
     qed_header_cpu_to_le(&s->header, &le);
-    ret = bdrv_pwrite(s->bs->file, 0, &le, sizeof(le));
+    ret = bdrv_pwrite(s->bs->file->bs, 0, &le, sizeof(le));
     if (ret != sizeof(le)) {
         return ret;
     }
@@ -119,7 +119,7 @@ static void qed_write_header_read_cb(void *opaque, int ret)
     /* 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,
+    bdrv_aio_writev(s->bs->file->bs, 0, &write_header_cb->qiov,
                     write_header_cb->nsectors, qed_write_header_cb,
                     write_header_cb);
 }
@@ -152,7 +152,7 @@ static void qed_write_header(BDRVQEDState *s, BlockCompletionFunc cb,
     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,
+    bdrv_aio_readv(s->bs->file->bs, 0, &write_header_cb->qiov, nsectors,
                    qed_write_header_read_cb, write_header_cb);
 }
 
@@ -354,12 +354,6 @@ static void qed_cancel_need_check_timer(BDRVQEDState *s)
     timer_del(s->need_check_timer);
 }
 
-static void bdrv_qed_rebind(BlockDriverState *bs)
-{
-    BDRVQEDState *s = bs->opaque;
-    s->bs = bs;
-}
-
 static void bdrv_qed_detach_aio_context(BlockDriverState *bs)
 {
     BDRVQEDState *s = bs->opaque;
@@ -381,6 +375,18 @@ static void bdrv_qed_attach_aio_context(BlockDriverState *bs,
     }
 }
 
+static void bdrv_qed_drain(BlockDriverState *bs)
+{
+    BDRVQEDState *s = bs->opaque;
+
+    /* Cancel timer and start doing I/O that were meant to happen as if it
+     * fired, that way we get bdrv_drain() taking care of the ongoing requests
+     * correctly. */
+    qed_cancel_need_check_timer(s);
+    qed_plug_allocating_write_reqs(s);
+    bdrv_aio_flush(s->bs, qed_clear_need_check, s);
+}
+
 static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
                          Error **errp)
 {
@@ -392,7 +398,7 @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
     s->bs = bs;
     QSIMPLEQ_INIT(&s->allocating_write_reqs);
 
-    ret = bdrv_pread(bs->file, 0, &le_header, sizeof(le_header));
+    ret = bdrv_pread(bs->file->bs, 0, &le_header, sizeof(le_header));
     if (ret < 0) {
         return ret;
     }
@@ -416,7 +422,7 @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     /* Round down file size to the last cluster */
-    file_size = bdrv_getlength(bs->file);
+    file_size = bdrv_getlength(bs->file->bs);
     if (file_size < 0) {
         return file_size;
     }
@@ -452,7 +458,7 @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
             return -EINVAL;
         }
 
-        ret = qed_read_string(bs->file, s->header.backing_filename_offset,
+        ret = qed_read_string(bs->file->bs, s->header.backing_filename_offset,
                               s->header.backing_filename_size, bs->backing_file,
                               sizeof(bs->backing_file));
         if (ret < 0) {
@@ -471,7 +477,7 @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
      * 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)) {
+        !bdrv_is_read_only(bs->file->bs) && !(flags & BDRV_O_INCOMING)) {
         s->header.autoclear_features &= QED_AUTOCLEAR_FEATURE_MASK;
 
         ret = qed_write_header_sync(s);
@@ -480,7 +486,7 @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
         }
 
         /* From here on only known autoclear feature bits are valid */
-        bdrv_flush(bs->file);
+        bdrv_flush(bs->file->bs);
     }
 
     s->l1_table = qed_alloc_table(s);
@@ -498,7 +504,7 @@ static int bdrv_qed_open(BlockDriverState *bs, QDict *options, int flags,
          * 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) &&
+        if (!bdrv_is_read_only(bs->file->bs) &&
             !(flags & BDRV_O_INCOMING)) {
             BdrvCheckResult result = {0};
 
@@ -541,7 +547,7 @@ static void bdrv_qed_close(BlockDriverState *bs)
     bdrv_qed_detach_aio_context(bs);
 
     /* Ensure writes reach stable storage */
-    bdrv_flush(bs->file);
+    bdrv_flush(bs->file->bs);
 
     /* Clean shutdown, no check required on next open */
     if (s->header.features & QED_F_NEED_CHECK) {
@@ -583,7 +589,7 @@ static int qed_create(const char *filename, uint32_t cluster_size,
 
     bs = NULL;
     ret = bdrv_open(&bs, filename, NULL, NULL,
-                    BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_PROTOCOL, NULL,
+                    BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_PROTOCOL,
                     &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
@@ -772,8 +778,8 @@ static void qed_read_backing_file(BDRVQEDState *s, uint64_t pos,
     /* 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 (s->bs->backing) {
+        int64_t l = bdrv_getlength(s->bs->backing->bs);
         if (l < 0) {
             cb(opaque, l);
             return;
@@ -802,7 +808,7 @@ static void qed_read_backing_file(BDRVQEDState *s, uint64_t pos,
     qemu_iovec_concat(*backing_qiov, qiov, 0, size);
 
     BLKDBG_EVENT(s->bs->file, BLKDBG_READ_BACKING_AIO);
-    bdrv_aio_readv(s->bs->backing_hd, pos / BDRV_SECTOR_SIZE,
+    bdrv_aio_readv(s->bs->backing->bs, pos / BDRV_SECTOR_SIZE,
                    *backing_qiov, size / BDRV_SECTOR_SIZE, cb, opaque);
 }
 
@@ -839,7 +845,7 @@ static void qed_copy_from_backing_file_write(void *opaque, int ret)
     }
 
     BLKDBG_EVENT(s->bs->file, BLKDBG_COW_WRITE);
-    bdrv_aio_writev(s->bs->file, copy_cb->offset / BDRV_SECTOR_SIZE,
+    bdrv_aio_writev(s->bs->file->bs, copy_cb->offset / BDRV_SECTOR_SIZE,
                     &copy_cb->qiov, copy_cb->qiov.size / BDRV_SECTOR_SIZE,
                     qed_copy_from_backing_file_cb, copy_cb);
 }
@@ -1055,7 +1061,7 @@ 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)) {
+    if (!bdrv_aio_flush(s->bs->file->bs, qed_aio_write_l2_update_cb, opaque)) {
         qed_aio_complete(acb, -EIO);
     }
 }
@@ -1081,7 +1087,7 @@ static void qed_aio_write_main(void *opaque, int ret)
     if (acb->find_cluster_ret == QED_CLUSTER_FOUND) {
         next_fn = qed_aio_next_io;
     } else {
-        if (s->bs->backing_hd) {
+        if (s->bs->backing) {
             next_fn = qed_aio_write_flush_before_l2_update;
         } else {
             next_fn = qed_aio_write_l2_update_cb;
@@ -1089,7 +1095,7 @@ static void qed_aio_write_main(void *opaque, int ret)
     }
 
     BLKDBG_EVENT(s->bs->file, BLKDBG_WRITE_AIO);
-    bdrv_aio_writev(s->bs->file, offset / BDRV_SECTOR_SIZE,
+    bdrv_aio_writev(s->bs->file->bs, offset / BDRV_SECTOR_SIZE,
                     &acb->cur_qiov, acb->cur_qiov.size / BDRV_SECTOR_SIZE,
                     next_fn, acb);
 }
@@ -1139,7 +1145,7 @@ static void qed_aio_write_prefill(void *opaque, int ret)
 static bool qed_should_set_need_check(BDRVQEDState *s)
 {
     /* The flush before L2 update path ensures consistency */
-    if (s->bs->backing_hd) {
+    if (s->bs->backing) {
         return false;
     }
 
@@ -1321,7 +1327,7 @@ static void qed_aio_read_data(void *opaque, int ret,
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
-    bdrv_aio_readv(bs->file, offset / BDRV_SECTOR_SIZE,
+    bdrv_aio_readv(bs->file->bs, offset / BDRV_SECTOR_SIZE,
                    &acb->cur_qiov, acb->cur_qiov.size / BDRV_SECTOR_SIZE,
                    qed_aio_next_io, acb);
     return;
@@ -1443,7 +1449,7 @@ static int coroutine_fn bdrv_qed_co_write_zeroes(BlockDriverState *bs,
     struct iovec iov;
 
     /* Refuse if there are untouched backing file sectors */
-    if (bs->backing_hd) {
+    if (bs->backing) {
         if (qed_offset_into_cluster(s, sector_num * BDRV_SECTOR_SIZE) != 0) {
             return -ENOTSUP;
         }
@@ -1580,7 +1586,7 @@ static int bdrv_qed_change_backing_file(BlockDriverState *bs,
     }
 
     /* Write new header */
-    ret = bdrv_pwrite_sync(bs->file, 0, buffer, buffer_len);
+    ret = bdrv_pwrite_sync(bs->file->bs, 0, buffer, buffer_len);
     g_free(buffer);
     if (ret == 0) {
         memcpy(&s->header, &new_header, sizeof(new_header));
@@ -1596,7 +1602,7 @@ static void bdrv_qed_invalidate_cache(BlockDriverState *bs, Error **errp)
 
     bdrv_qed_close(bs);
 
-    bdrv_invalidate_cache(bs->file, &local_err);
+    bdrv_invalidate_cache(bs->file->bs, &local_err);
     if (local_err) {
         error_propagate(errp, local_err);
         return;
@@ -1664,7 +1670,6 @@ static BlockDriver bdrv_qed = {
     .supports_backing         = true,
 
     .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,
@@ -1683,6 +1688,7 @@ static BlockDriver bdrv_qed = {
     .bdrv_check               = bdrv_qed_check,
     .bdrv_detach_aio_context  = bdrv_qed_detach_aio_context,
     .bdrv_attach_aio_context  = bdrv_qed_attach_aio_context,
+    .bdrv_drain               = bdrv_qed_drain,
 };
 
 static void bdrv_qed_init(void)
diff --git a/block/quorum.c b/block/quorum.c
index 2f6c45f76..e640688eb 100644
--- a/block/quorum.c
+++ b/block/quorum.c
@@ -64,7 +64,7 @@ typedef struct QuorumVotes {
 
 /* the following structure holds the state of one quorum instance */
 typedef struct BDRVQuorumState {
-    BlockDriverState **bs; /* children BlockDriverStates */
+    BdrvChild **children;  /* children BlockDriverStates */
     int num_children;      /* children count */
     int threshold;         /* if less than threshold children reads gave the
                             * same result a quorum error occurs.
@@ -336,7 +336,7 @@ static void quorum_report_bad_versions(BDRVQuorumState *s,
             continue;
         }
         QLIST_FOREACH(item, &version->items, next) {
-            quorum_report_bad(acb, s->bs[item->index]->node_name, 0);
+            quorum_report_bad(acb, s->children[item->index]->bs->node_name, 0);
         }
     }
 }
@@ -369,8 +369,9 @@ static bool quorum_rewrite_bad_versions(BDRVQuorumState *s, QuorumAIOCB *acb,
             continue;
         }
         QLIST_FOREACH(item, &version->items, next) {
-            bdrv_aio_writev(s->bs[item->index], acb->sector_num, acb->qiov,
-                            acb->nb_sectors, quorum_rewrite_aio_cb, acb);
+            bdrv_aio_writev(s->children[item->index]->bs, acb->sector_num,
+                            acb->qiov, acb->nb_sectors, quorum_rewrite_aio_cb,
+                            acb);
         }
     }
 
@@ -639,14 +640,15 @@ static BlockAIOCB *read_quorum_children(QuorumAIOCB *acb)
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        acb->qcrs[i].buf = qemu_blockalign(s->bs[i], acb->qiov->size);
+        acb->qcrs[i].buf = qemu_blockalign(s->children[i]->bs, acb->qiov->size);
         qemu_iovec_init(&acb->qcrs[i].qiov, acb->qiov->niov);
         qemu_iovec_clone(&acb->qcrs[i].qiov, acb->qiov, acb->qcrs[i].buf);
     }
 
     for (i = 0; i < s->num_children; i++) {
-        bdrv_aio_readv(s->bs[i], acb->sector_num, &acb->qcrs[i].qiov,
-                       acb->nb_sectors, quorum_aio_cb, &acb->qcrs[i]);
+        acb->qcrs[i].aiocb = bdrv_aio_readv(s->children[i]->bs, acb->sector_num,
+                                            &acb->qcrs[i].qiov, acb->nb_sectors,
+                                            quorum_aio_cb, &acb->qcrs[i]);
     }
 
     return &acb->common;
@@ -656,14 +658,15 @@ static BlockAIOCB *read_fifo_child(QuorumAIOCB *acb)
 {
     BDRVQuorumState *s = acb->common.bs->opaque;
 
-    acb->qcrs[acb->child_iter].buf = qemu_blockalign(s->bs[acb->child_iter],
-                                                     acb->qiov->size);
+    acb->qcrs[acb->child_iter].buf =
+        qemu_blockalign(s->children[acb->child_iter]->bs, acb->qiov->size);
     qemu_iovec_init(&acb->qcrs[acb->child_iter].qiov, acb->qiov->niov);
     qemu_iovec_clone(&acb->qcrs[acb->child_iter].qiov, acb->qiov,
                      acb->qcrs[acb->child_iter].buf);
-    bdrv_aio_readv(s->bs[acb->child_iter], acb->sector_num,
-                   &acb->qcrs[acb->child_iter].qiov, acb->nb_sectors,
-                   quorum_aio_cb, &acb->qcrs[acb->child_iter]);
+    acb->qcrs[acb->child_iter].aiocb =
+        bdrv_aio_readv(s->children[acb->child_iter]->bs, acb->sector_num,
+                       &acb->qcrs[acb->child_iter].qiov, acb->nb_sectors,
+                       quorum_aio_cb, &acb->qcrs[acb->child_iter]);
 
     return &acb->common;
 }
@@ -702,8 +705,8 @@ static BlockAIOCB *quorum_aio_writev(BlockDriverState *bs,
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        acb->qcrs[i].aiocb = bdrv_aio_writev(s->bs[i], sector_num, qiov,
-                                             nb_sectors, &quorum_aio_cb,
+        acb->qcrs[i].aiocb = bdrv_aio_writev(s->children[i]->bs, sector_num,
+                                             qiov, nb_sectors, &quorum_aio_cb,
                                              &acb->qcrs[i]);
     }
 
@@ -717,12 +720,12 @@ static int64_t quorum_getlength(BlockDriverState *bs)
     int i;
 
     /* check that all file have the same length */
-    result = bdrv_getlength(s->bs[0]);
+    result = bdrv_getlength(s->children[0]->bs);
     if (result < 0) {
         return result;
     }
     for (i = 1; i < s->num_children; i++) {
-        int64_t value = bdrv_getlength(s->bs[i]);
+        int64_t value = bdrv_getlength(s->children[i]->bs);
         if (value < 0) {
             return value;
         }
@@ -741,7 +744,7 @@ static void quorum_invalidate_cache(BlockDriverState *bs, Error **errp)
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        bdrv_invalidate_cache(s->bs[i], &local_err);
+        bdrv_invalidate_cache(s->children[i]->bs, &local_err);
         if (local_err) {
             error_propagate(errp, local_err);
             return;
@@ -762,7 +765,7 @@ static coroutine_fn int quorum_co_flush(BlockDriverState *bs)
     error_votes.compare = quorum_64bits_compare;
 
     for (i = 0; i < s->num_children; i++) {
-        result = bdrv_co_flush(s->bs[i]);
+        result = bdrv_co_flush(s->children[i]->bs);
         result_value.l = result;
         quorum_count_vote(&error_votes, &result_value, i);
     }
@@ -782,7 +785,7 @@ static bool quorum_recurse_is_first_non_filter(BlockDriverState *bs,
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        bool perm = bdrv_recurse_is_first_non_filter(s->bs[i],
+        bool perm = bdrv_recurse_is_first_non_filter(s->children[i]->bs,
                                                      candidate);
         if (perm) {
             return true;
@@ -889,6 +892,12 @@ static int quorum_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     s->threshold = qemu_opt_get_number(opts, QUORUM_OPT_VOTE_THRESHOLD, 0);
+    /* and validate it against s->num_children */
+    ret = quorum_valid_threshold(s->threshold, s->num_children, &local_err);
+    if (ret < 0) {
+        goto exit;
+    }
+
     ret = parse_read_pattern(qemu_opt_get(opts, QUORUM_OPT_READ_PATTERN));
     if (ret < 0) {
         error_setg(&local_err, "Please set read-pattern as fifo or quorum");
@@ -897,12 +906,6 @@ static int quorum_open(BlockDriverState *bs, QDict *options, int flags,
     s->read_pattern = ret;
 
     if (s->read_pattern == QUORUM_READ_PATTERN_QUORUM) {
-        /* and validate it against s->num_children */
-        ret = quorum_valid_threshold(s->threshold, s->num_children, &local_err);
-        if (ret < 0) {
-            goto exit;
-        }
-
         /* is the driver in blkverify mode */
         if (qemu_opt_get_bool(opts, QUORUM_OPT_BLKVERIFY, false) &&
             s->num_children == 2 && s->threshold == 2) {
@@ -922,8 +925,8 @@ static int quorum_open(BlockDriverState *bs, QDict *options, int flags,
         }
     }
 
-    /* allocate the children BlockDriverState array */
-    s->bs = g_new0(BlockDriverState *, s->num_children);
+    /* allocate the children array */
+    s->children = g_new0(BdrvChild *, s->num_children);
     opened = g_new0(bool, s->num_children);
 
     for (i = 0; i < s->num_children; i++) {
@@ -931,9 +934,10 @@ static int quorum_open(BlockDriverState *bs, QDict *options, int flags,
         ret = snprintf(indexstr, 32, "children.%d", i);
         assert(ret < 32);
 
-        ret = bdrv_open_image(&s->bs[i], NULL, options, indexstr, bs,
-                              &child_format, false, &local_err);
-        if (ret < 0) {
+        s->children[i] = bdrv_open_child(NULL, options, indexstr, bs,
+                                         &child_format, false, &local_err);
+        if (local_err) {
+            ret = -EINVAL;
             goto close_exit;
         }
 
@@ -949,9 +953,9 @@ close_exit:
         if (!opened[i]) {
             continue;
         }
-        bdrv_unref(s->bs[i]);
+        bdrv_unref_child(bs, s->children[i]);
     }
-    g_free(s->bs);
+    g_free(s->children);
     g_free(opened);
 exit:
     qemu_opts_del(opts);
@@ -968,10 +972,10 @@ static void quorum_close(BlockDriverState *bs)
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        bdrv_unref(s->bs[i]);
+        bdrv_unref_child(bs, s->children[i]);
     }
 
-    g_free(s->bs);
+    g_free(s->children);
 }
 
 static void quorum_detach_aio_context(BlockDriverState *bs)
@@ -980,7 +984,7 @@ static void quorum_detach_aio_context(BlockDriverState *bs)
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        bdrv_detach_aio_context(s->bs[i]);
+        bdrv_detach_aio_context(s->children[i]->bs);
     }
 }
 
@@ -991,7 +995,7 @@ static void quorum_attach_aio_context(BlockDriverState *bs,
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        bdrv_attach_aio_context(s->bs[i], new_context);
+        bdrv_attach_aio_context(s->children[i]->bs, new_context);
     }
 }
 
@@ -1003,16 +1007,17 @@ static void quorum_refresh_filename(BlockDriverState *bs)
     int i;
 
     for (i = 0; i < s->num_children; i++) {
-        bdrv_refresh_filename(s->bs[i]);
-        if (!s->bs[i]->full_open_options) {
+        bdrv_refresh_filename(s->children[i]->bs);
+        if (!s->children[i]->bs->full_open_options) {
             return;
         }
     }
 
     children = qlist_new();
     for (i = 0; i < s->num_children; i++) {
-        QINCREF(s->bs[i]->full_open_options);
-        qlist_append_obj(children, QOBJECT(s->bs[i]->full_open_options));
+        QINCREF(s->children[i]->bs->full_open_options);
+        qlist_append_obj(children,
+                         QOBJECT(s->children[i]->bs->full_open_options));
     }
 
     opts = qdict_new();
diff --git a/block/raw-posix.c b/block/raw-posix.c
index 855febed5..2fff1843c 100644
--- a/block/raw-posix.c
+++ b/block/raw-posix.c
@@ -127,11 +127,6 @@ do { \
 
 #define FTYPE_FILE   0
 #define FTYPE_CD     1
-#define FTYPE_FD     2
-
-/* if the FD is not accessed during that time (in ns), we try to
-   reopen it to see if the disk has been changed */
-#define FD_OPEN_TIMEOUT (1000000000)
 
 #define MAX_BLOCKSIZE	4096
 
@@ -141,13 +136,6 @@ typedef struct BDRVRawState {
     int open_flags;
     size_t buf_align;
 
-#if defined(__linux__)
-    /* linux floppy specific */
-    int64_t fd_open_time;
-    int64_t fd_error_time;
-    int fd_got_error;
-    int fd_media_changed;
-#endif
 #ifdef CONFIG_LINUX_AIO
     int use_aio;
     void *aio_ctx;
@@ -519,7 +507,16 @@ static int raw_open_common(BlockDriverState *bs, QDict *options,
                      "future QEMU versions.\n",
                      bs->filename);
     }
-#endif
+#else
+    if (bdrv_flags & BDRV_O_NATIVE_AIO) {
+        error_printf("WARNING: aio=native was specified for '%s', but "
+                     "is not supported in this build. Falling back to "
+                     "aio=threads.\n"
+                     "         This will become an error condition in "
+                     "future QEMU versions.\n",
+                     bs->filename);
+    }
+#endif /* !defined(CONFIG_LINUX_AIO) */
 
     s->has_discard = true;
     s->has_write_zeroes = true;
@@ -626,7 +623,7 @@ static int raw_reopen_prepare(BDRVReopenState *state,
     }
 #endif
 
-    if (s->type == FTYPE_FD || s->type == FTYPE_CD) {
+    if (s->type == FTYPE_CD) {
         raw_s->open_flags |= O_NONBLOCK;
     }
 
@@ -670,11 +667,17 @@ static int raw_reopen_prepare(BDRVReopenState *state,
 
     /* If we cannot use fcntl, or fcntl failed, fall back to qemu_open() */
     if (raw_s->fd == -1) {
-        assert(!(raw_s->open_flags & O_CREAT));
-        raw_s->fd = qemu_open(state->bs->filename, raw_s->open_flags);
-        if (raw_s->fd == -1) {
-            error_setg_errno(errp, errno, "Could not reopen file");
-            ret = -1;
+        const char *normalized_filename = state->bs->filename;
+        ret = raw_normalize_devicepath(&normalized_filename);
+        if (ret < 0) {
+            error_setg_errno(errp, -ret, "Could not normalize device path");
+        } else {
+            assert(!(raw_s->open_flags & O_CREAT));
+            raw_s->fd = qemu_open(normalized_filename, raw_s->open_flags);
+            if (raw_s->fd == -1) {
+                error_setg_errno(errp, errno, "Could not reopen file");
+                ret = -1;
+            }
         }
     }
 
@@ -780,7 +783,6 @@ static int hdev_probe_geometry(BlockDriverState *bs, HDGeometry *geo)
 {
     BDRVRawState *s = bs->opaque;
     struct hd_geometry ioctl_geo = {0};
-    uint32_t blksize;
 
     /* If DASD, get its geometry */
     if (check_for_dasd(s->fd) < 0) {
@@ -800,12 +802,6 @@ static int hdev_probe_geometry(BlockDriverState *bs, HDGeometry *geo)
     }
     geo->heads = ioctl_geo.heads;
     geo->sectors = ioctl_geo.sectors;
-    if (!probe_physical_blocksize(s->fd, &blksize)) {
-        /* overwrite cyls: HDIO_GETGEO result is incorrect for big drives */
-        geo->cylinders = bdrv_nb_sectors(bs) / (blksize / BDRV_SECTOR_SIZE)
-                                             / (geo->heads * geo->sectors);
-        return 0;
-    }
     geo->cylinders = ioctl_geo.cylinders;
 
     return 0;
@@ -1253,7 +1249,7 @@ static int aio_worker(void *arg)
         break;
     }
 
-    g_slice_free(RawPosixAIOData, aiocb);
+    g_free(aiocb);
     return ret;
 }
 
@@ -1261,7 +1257,7 @@ static int paio_submit_co(BlockDriverState *bs, int fd,
         int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
         int type)
 {
-    RawPosixAIOData *acb = g_slice_new(RawPosixAIOData);
+    RawPosixAIOData *acb = g_new(RawPosixAIOData, 1);
     ThreadPool *pool;
 
     acb->bs = bs;
@@ -1286,7 +1282,7 @@ static BlockAIOCB *paio_submit(BlockDriverState *bs, int fd,
         int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
         BlockCompletionFunc *cb, void *opaque, int type)
 {
-    RawPosixAIOData *acb = g_slice_new(RawPosixAIOData);
+    RawPosixAIOData *acb = g_new(RawPosixAIOData, 1);
     ThreadPool *pool;
 
     acb->bs = bs;
@@ -1642,7 +1638,7 @@ static int raw_create(const char *filename, QemuOpts *opts, Error **errp)
         goto out;
     }
 
-    fd = qemu_open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY,
+    fd = qemu_open(filename, O_RDWR | O_CREAT | O_TRUNC | O_BINARY,
                    0644);
     if (fd < 0) {
         result = -errno;
@@ -1973,8 +1969,8 @@ BlockDriver bdrv_file = {
 
 #if defined(__APPLE__) && defined(__MACH__)
 static kern_return_t FindEjectableCDMedia( io_iterator_t *mediaIterator );
-static kern_return_t GetBSDPath( io_iterator_t mediaIterator, char *bsdPath, CFIndex maxPathSize );
-
+static kern_return_t GetBSDPath(io_iterator_t mediaIterator, char *bsdPath,
+                                CFIndex maxPathSize, int flags);
 kern_return_t FindEjectableCDMedia( io_iterator_t *mediaIterator )
 {
     kern_return_t       kernResult;
@@ -2001,7 +1997,8 @@ kern_return_t FindEjectableCDMedia( io_iterator_t *mediaIterator )
     return kernResult;
 }
 
-kern_return_t GetBSDPath( io_iterator_t mediaIterator, char *bsdPath, CFIndex maxPathSize )
+kern_return_t GetBSDPath(io_iterator_t mediaIterator, char *bsdPath,
+                         CFIndex maxPathSize, int flags)
 {
     io_object_t     nextMedia;
     kern_return_t   kernResult = KERN_FAILURE;
@@ -2014,7 +2011,9 @@ kern_return_t GetBSDPath( io_iterator_t mediaIterator, char *bsdPath, CFIndex ma
         if ( bsdPathAsCFString ) {
             size_t devPathLength;
             strcpy( bsdPath, _PATH_DEV );
-            strcat( bsdPath, "r" );
+            if (flags & BDRV_O_NOCACHE) {
+                strcat(bsdPath, "r");
+            }
             devPathLength = strlen( bsdPath );
             if ( CFStringGetCString( bsdPathAsCFString, bsdPath + devPathLength, maxPathSize - devPathLength, kCFStringEncodingASCII ) ) {
                 kernResult = KERN_SUCCESS;
@@ -2126,8 +2125,8 @@ static int hdev_open(BlockDriverState *bs, QDict *options, int flags,
         int fd;
 
         kernResult = FindEjectableCDMedia( &mediaIterator );
-        kernResult = GetBSDPath( mediaIterator, bsdPath, sizeof( bsdPath ) );
-
+        kernResult = GetBSDPath(mediaIterator, bsdPath, sizeof(bsdPath),
+                                flags);
         if ( bsdPath[ 0 ] != '\0' ) {
             strcat(bsdPath,"s0");
             /* some CDs don't have a partition 0 */
@@ -2172,53 +2171,6 @@ static int hdev_open(BlockDriverState *bs, QDict *options, int flags,
 }
 
 #if defined(__linux__)
-/* Note: we do not have a reliable method to detect if the floppy is
-   present. The current method is to try to open the floppy at every
-   I/O and to keep it opened during a few hundreds of ms. */
-static int fd_open(BlockDriverState *bs)
-{
-    BDRVRawState *s = bs->opaque;
-    int last_media_present;
-
-    if (s->type != FTYPE_FD)
-        return 0;
-    last_media_present = (s->fd >= 0);
-    if (s->fd >= 0 &&
-        (qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - s->fd_open_time) >= FD_OPEN_TIMEOUT) {
-        qemu_close(s->fd);
-        s->fd = -1;
-        DPRINTF("Floppy closed\n");
-    }
-    if (s->fd < 0) {
-        if (s->fd_got_error &&
-            (qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - s->fd_error_time) < FD_OPEN_TIMEOUT) {
-            DPRINTF("No floppy (open delayed)\n");
-            return -EIO;
-        }
-        s->fd = qemu_open(bs->filename, s->open_flags & ~O_NONBLOCK);
-        if (s->fd < 0) {
-            s->fd_error_time = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
-            s->fd_got_error = 1;
-            if (last_media_present)
-                s->fd_media_changed = 1;
-            DPRINTF("No floppy\n");
-            return -EIO;
-        }
-        DPRINTF("Floppy opened\n");
-    }
-    if (!last_media_present)
-        s->fd_media_changed = 1;
-    s->fd_open_time = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
-    s->fd_got_error = 0;
-    return 0;
-}
-
-static int hdev_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
-{
-    BDRVRawState *s = bs->opaque;
-
-    return ioctl(s->fd, req, buf);
-}
 
 static BlockAIOCB *hdev_aio_ioctl(BlockDriverState *bs,
         unsigned long int req, void *buf,
@@ -2231,7 +2183,7 @@ static BlockAIOCB *hdev_aio_ioctl(BlockDriverState *bs,
     if (fd_open(bs) < 0)
         return NULL;
 
-    acb = g_slice_new(RawPosixAIOData);
+    acb = g_new(RawPosixAIOData, 1);
     acb->bs = bs;
     acb->aio_type = QEMU_AIO_IOCTL;
     acb->aio_fildes = s->fd;
@@ -2241,8 +2193,8 @@ static BlockAIOCB *hdev_aio_ioctl(BlockDriverState *bs,
     pool = aio_get_thread_pool(bdrv_get_aio_context(bs));
     return thread_pool_submit_aio(pool, aio_worker, acb, cb, opaque);
 }
+#endif /* linux */
 
-#elif defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
 static int fd_open(BlockDriverState *bs)
 {
     BDRVRawState *s = bs->opaque;
@@ -2252,14 +2204,6 @@ static int fd_open(BlockDriverState *bs)
         return 0;
     return -EIO;
 }
-#else /* !linux && !FreeBSD */
-
-static int fd_open(BlockDriverState *bs)
-{
-    return 0;
-}
-
-#endif /* !linux && !FreeBSD */
 
 static coroutine_fn BlockAIOCB *hdev_aio_discard(BlockDriverState *bs,
     int64_t sector_num, int nb_sectors,
@@ -2303,17 +2247,22 @@ static int hdev_create(const char *filename, QemuOpts *opts,
     int64_t total_size = 0;
     bool has_prefix;
 
-    /* This function is used by all three protocol block drivers and therefore
-     * any of these three prefixes may be given.
+    /* This function is used by both protocol block drivers and therefore either
+     * of these prefixes may be given.
      * The return value has to be stored somewhere, otherwise this is an error
      * due to -Werror=unused-value. */
     has_prefix =
         strstart(filename, "host_device:", &filename) ||
-        strstart(filename, "host_cdrom:" , &filename) ||
-        strstart(filename, "host_floppy:", &filename);
+        strstart(filename, "host_cdrom:" , &filename);
 
     (void)has_prefix;
 
+    ret = raw_normalize_devicepath(&filename);
+    if (ret < 0) {
+        error_setg_errno(errp, -ret, "Could not normalize device path");
+        return ret;
+    }
+
     /* Read out options */
     total_size = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
                           BDRV_SECTOR_SIZE);
@@ -2379,160 +2328,10 @@ static BlockDriver bdrv_host_device = {
 
     /* generic scsi device */
 #ifdef __linux__
-    .bdrv_ioctl         = hdev_ioctl,
     .bdrv_aio_ioctl     = hdev_aio_ioctl,
 #endif
 };
 
-#ifdef __linux__
-static void floppy_parse_filename(const char *filename, QDict *options,
-                                  Error **errp)
-{
-    /* The prefix is optional, just as for "file". */
-    strstart(filename, "host_floppy:", &filename);
-
-    qdict_put_obj(options, "filename", QOBJECT(qstring_from_str(filename)));
-}
-
-static int floppy_open(BlockDriverState *bs, QDict *options, int flags,
-                       Error **errp)
-{
-    BDRVRawState *s = bs->opaque;
-    Error *local_err = NULL;
-    int ret;
-
-    s->type = FTYPE_FD;
-
-    /* open will not fail even if no floppy is inserted, so add O_NONBLOCK */
-    ret = raw_open_common(bs, options, flags, O_NONBLOCK, &local_err);
-    if (ret) {
-        if (local_err) {
-            error_propagate(errp, local_err);
-        }
-        return ret;
-    }
-
-    /* close fd so that we can reopen it as needed */
-    qemu_close(s->fd);
-    s->fd = -1;
-    s->fd_media_changed = 1;
-
-    error_report("Host floppy pass-through is deprecated");
-    error_printf("Support for it will be removed in a future release.\n");
-    return 0;
-}
-
-static int floppy_probe_device(const char *filename)
-{
-    int fd, ret;
-    int prio = 0;
-    struct floppy_struct fdparam;
-    struct stat st;
-
-    if (strstart(filename, "/dev/fd", NULL) &&
-        !strstart(filename, "/dev/fdset/", NULL) &&
-        !strstart(filename, "/dev/fd/", NULL)) {
-        prio = 50;
-    }
-
-    fd = qemu_open(filename, O_RDONLY | O_NONBLOCK);
-    if (fd < 0) {
-        goto out;
-    }
-    ret = fstat(fd, &st);
-    if (ret == -1 || !S_ISBLK(st.st_mode)) {
-        goto outc;
-    }
-
-    /* Attempt to detect via a floppy specific ioctl */
-    ret = ioctl(fd, FDGETPRM, &fdparam);
-    if (ret >= 0)
-        prio = 100;
-
-outc:
-    qemu_close(fd);
-out:
-    return prio;
-}
-
-
-static int floppy_is_inserted(BlockDriverState *bs)
-{
-    return fd_open(bs) >= 0;
-}
-
-static int floppy_media_changed(BlockDriverState *bs)
-{
-    BDRVRawState *s = bs->opaque;
-    int ret;
-
-    /*
-     * XXX: we do not have a true media changed indication.
-     * It does not work if the floppy is changed without trying to read it.
-     */
-    fd_open(bs);
-    ret = s->fd_media_changed;
-    s->fd_media_changed = 0;
-    DPRINTF("Floppy changed=%d\n", ret);
-    return ret;
-}
-
-static void floppy_eject(BlockDriverState *bs, bool eject_flag)
-{
-    BDRVRawState *s = bs->opaque;
-    int fd;
-
-    if (s->fd >= 0) {
-        qemu_close(s->fd);
-        s->fd = -1;
-    }
-    fd = qemu_open(bs->filename, s->open_flags | O_NONBLOCK);
-    if (fd >= 0) {
-        if (ioctl(fd, FDEJECT, 0) < 0)
-            perror("FDEJECT");
-        qemu_close(fd);
-    }
-}
-
-static BlockDriver bdrv_host_floppy = {
-    .format_name        = "host_floppy",
-    .protocol_name      = "host_floppy",
-    .instance_size      = sizeof(BDRVRawState),
-    .bdrv_needs_filename = true,
-    .bdrv_probe_device	= floppy_probe_device,
-    .bdrv_parse_filename = floppy_parse_filename,
-    .bdrv_file_open     = floppy_open,
-    .bdrv_close         = raw_close,
-    .bdrv_reopen_prepare = raw_reopen_prepare,
-    .bdrv_reopen_commit  = raw_reopen_commit,
-    .bdrv_reopen_abort   = raw_reopen_abort,
-    .bdrv_create         = hdev_create,
-    .create_opts         = &raw_create_opts,
-
-    .bdrv_aio_readv     = raw_aio_readv,
-    .bdrv_aio_writev    = raw_aio_writev,
-    .bdrv_aio_flush	= raw_aio_flush,
-    .bdrv_refresh_limits = raw_refresh_limits,
-    .bdrv_io_plug = raw_aio_plug,
-    .bdrv_io_unplug = raw_aio_unplug,
-    .bdrv_flush_io_queue = raw_aio_flush_io_queue,
-
-    .bdrv_truncate      = raw_truncate,
-    .bdrv_getlength      = raw_getlength,
-    .has_variable_length = true,
-    .bdrv_get_allocated_file_size
-                        = raw_get_allocated_file_size,
-
-    .bdrv_detach_aio_context = raw_detach_aio_context,
-    .bdrv_attach_aio_context = raw_attach_aio_context,
-
-    /* removable device support */
-    .bdrv_is_inserted   = floppy_is_inserted,
-    .bdrv_media_changed = floppy_media_changed,
-    .bdrv_eject         = floppy_eject,
-};
-#endif
-
 #if defined(__linux__) || defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
 static void cdrom_parse_filename(const char *filename, QDict *options,
                                  Error **errp)
@@ -2588,15 +2387,13 @@ out:
     return prio;
 }
 
-static int cdrom_is_inserted(BlockDriverState *bs)
+static bool cdrom_is_inserted(BlockDriverState *bs)
 {
     BDRVRawState *s = bs->opaque;
     int ret;
 
     ret = ioctl(s->fd, CDROM_DRIVE_STATUS, CDSL_CURRENT);
-    if (ret == CDS_DISC_OK)
-        return 1;
-    return 0;
+    return ret == CDS_DISC_OK;
 }
 
 static void cdrom_eject(BlockDriverState *bs, bool eject_flag)
@@ -2663,7 +2460,6 @@ static BlockDriver bdrv_host_cdrom = {
     .bdrv_lock_medium   = cdrom_lock_medium,
 
     /* generic scsi device */
-    .bdrv_ioctl         = hdev_ioctl,
     .bdrv_aio_ioctl     = hdev_aio_ioctl,
 };
 #endif /* __linux__ */
@@ -2722,7 +2518,7 @@ static int cdrom_reopen(BlockDriverState *bs)
     return 0;
 }
 
-static int cdrom_is_inserted(BlockDriverState *bs)
+static bool cdrom_is_inserted(BlockDriverState *bs)
 {
     return raw_getlength(bs) > 0;
 }
@@ -2810,7 +2606,6 @@ static void bdrv_file_init(void)
     bdrv_register(&bdrv_file);
     bdrv_register(&bdrv_host_device);
 #ifdef __linux__
-    bdrv_register(&bdrv_host_floppy);
     bdrv_register(&bdrv_host_cdrom);
 #endif
 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
diff --git a/block/raw-win32.c b/block/raw-win32.c
index 68f2338ac..2d0907a82 100644
--- a/block/raw-win32.c
+++ b/block/raw-win32.c
@@ -119,9 +119,9 @@ static int aio_worker(void *arg)
     case QEMU_AIO_WRITE:
         count = handle_aiocb_rw(aiocb);
         if (count == aiocb->aio_nbytes) {
-            count = 0;
+            ret = 0;
         } else {
-            count = -EINVAL;
+            ret = -EINVAL;
         }
         break;
     case QEMU_AIO_FLUSH:
@@ -135,7 +135,7 @@ static int aio_worker(void *arg)
         break;
     }
 
-    g_slice_free(RawWin32AIOData, aiocb);
+    g_free(aiocb);
     return ret;
 }
 
@@ -143,7 +143,7 @@ static BlockAIOCB *paio_submit(BlockDriverState *bs, HANDLE hfile,
         int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
         BlockCompletionFunc *cb, void *opaque, int type)
 {
-    RawWin32AIOData *acb = g_slice_new(RawWin32AIOData);
+    RawWin32AIOData *acb = g_new(RawWin32AIOData, 1);
     ThreadPool *pool;
 
     acb->bs = bs;
diff --git a/block/raw_bsd.c b/block/raw_bsd.c
index e3d2d0468..915d6fd0e 100644
--- a/block/raw_bsd.c
+++ b/block/raw_bsd.c
@@ -52,7 +52,7 @@ static int coroutine_fn raw_co_readv(BlockDriverState *bs, int64_t sector_num,
                                      int nb_sectors, QEMUIOVector *qiov)
 {
     BLKDBG_EVENT(bs->file, BLKDBG_READ_AIO);
-    return bdrv_co_readv(bs->file, sector_num, nb_sectors, qiov);
+    return bdrv_co_readv(bs->file->bs, sector_num, nb_sectors, qiov);
 }
 
 static int coroutine_fn raw_co_writev(BlockDriverState *bs, int64_t sector_num,
@@ -75,7 +75,7 @@ static int coroutine_fn raw_co_writev(BlockDriverState *bs, int64_t sector_num,
             return 0;
         }
 
-        buf = qemu_try_blockalign(bs->file, 512);
+        buf = qemu_try_blockalign(bs->file->bs, 512);
         if (!buf) {
             ret = -ENOMEM;
             goto fail;
@@ -102,7 +102,7 @@ static int coroutine_fn raw_co_writev(BlockDriverState *bs, int64_t sector_num,
     }
 
     BLKDBG_EVENT(bs->file, BLKDBG_WRITE_AIO);
-    ret = bdrv_co_writev(bs->file, sector_num, nb_sectors, qiov);
+    ret = bdrv_co_writev(bs->file->bs, sector_num, nb_sectors, qiov);
 
 fail:
     if (qiov == &local_qiov) {
@@ -125,58 +125,48 @@ static int coroutine_fn raw_co_write_zeroes(BlockDriverState *bs,
                                             int64_t sector_num, int nb_sectors,
                                             BdrvRequestFlags flags)
 {
-    return bdrv_co_write_zeroes(bs->file, sector_num, nb_sectors, flags);
+    return bdrv_co_write_zeroes(bs->file->bs, sector_num, nb_sectors, flags);
 }
 
 static int coroutine_fn raw_co_discard(BlockDriverState *bs,
                                        int64_t sector_num, int nb_sectors)
 {
-    return bdrv_co_discard(bs->file, sector_num, nb_sectors);
+    return bdrv_co_discard(bs->file->bs, sector_num, nb_sectors);
 }
 
 static int64_t raw_getlength(BlockDriverState *bs)
 {
-    return bdrv_getlength(bs->file);
+    return bdrv_getlength(bs->file->bs);
 }
 
 static int raw_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
 {
-    return bdrv_get_info(bs->file, bdi);
+    return bdrv_get_info(bs->file->bs, bdi);
 }
 
 static void raw_refresh_limits(BlockDriverState *bs, Error **errp)
 {
-    bs->bl = bs->file->bl;
+    bs->bl = bs->file->bs->bl;
 }
 
 static int raw_truncate(BlockDriverState *bs, int64_t offset)
 {
-    return bdrv_truncate(bs->file, offset);
-}
-
-static int raw_is_inserted(BlockDriverState *bs)
-{
-    return bdrv_is_inserted(bs->file);
+    return bdrv_truncate(bs->file->bs, offset);
 }
 
 static int raw_media_changed(BlockDriverState *bs)
 {
-    return bdrv_media_changed(bs->file);
+    return bdrv_media_changed(bs->file->bs);
 }
 
 static void raw_eject(BlockDriverState *bs, bool eject_flag)
 {
-    bdrv_eject(bs->file, eject_flag);
+    bdrv_eject(bs->file->bs, eject_flag);
 }
 
 static void raw_lock_medium(BlockDriverState *bs, bool locked)
 {
-    bdrv_lock_medium(bs->file, locked);
-}
-
-static int raw_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
-{
-    return bdrv_ioctl(bs->file, req, buf);
+    bdrv_lock_medium(bs->file->bs, locked);
 }
 
 static BlockAIOCB *raw_aio_ioctl(BlockDriverState *bs,
@@ -184,12 +174,12 @@ static BlockAIOCB *raw_aio_ioctl(BlockDriverState *bs,
                                  BlockCompletionFunc *cb,
                                  void *opaque)
 {
-    return bdrv_aio_ioctl(bs->file, req, buf, cb, opaque);
+    return bdrv_aio_ioctl(bs->file->bs, req, buf, cb, opaque);
 }
 
 static int raw_has_zero_init(BlockDriverState *bs)
 {
-    return bdrv_has_zero_init(bs->file);
+    return bdrv_has_zero_init(bs->file->bs);
 }
 
 static int raw_create(const char *filename, QemuOpts *opts, Error **errp)
@@ -207,7 +197,7 @@ static int raw_create(const char *filename, QemuOpts *opts, Error **errp)
 static int raw_open(BlockDriverState *bs, QDict *options, int flags,
                     Error **errp)
 {
-    bs->sg = bs->file->sg;
+    bs->sg = bs->file->bs->sg;
 
     if (bs->probed && !bdrv_is_read_only(bs)) {
         fprintf(stderr,
@@ -217,7 +207,7 @@ static int raw_open(BlockDriverState *bs, QDict *options, int flags,
                 "raw images, write operations on block 0 will be restricted.\n"
                 "         Specify the 'raw' format explicitly to remove the "
                 "restrictions.\n",
-                bs->file->filename);
+                bs->file->bs->filename);
     }
 
     return 0;
@@ -237,12 +227,12 @@ static int raw_probe(const uint8_t *buf, int buf_size, const char *filename)
 
 static int raw_probe_blocksizes(BlockDriverState *bs, BlockSizes *bsz)
 {
-    return bdrv_probe_blocksizes(bs->file, bsz);
+    return bdrv_probe_blocksizes(bs->file->bs, bsz);
 }
 
 static int raw_probe_geometry(BlockDriverState *bs, HDGeometry *geo)
 {
-    return bdrv_probe_geometry(bs->file, geo);
+    return bdrv_probe_geometry(bs->file->bs, geo);
 }
 
 BlockDriver bdrv_raw = {
@@ -264,11 +254,9 @@ BlockDriver bdrv_raw = {
     .bdrv_refresh_limits  = &raw_refresh_limits,
     .bdrv_probe_blocksizes = &raw_probe_blocksizes,
     .bdrv_probe_geometry  = &raw_probe_geometry,
-    .bdrv_is_inserted     = &raw_is_inserted,
     .bdrv_media_changed   = &raw_media_changed,
     .bdrv_eject           = &raw_eject,
     .bdrv_lock_medium     = &raw_lock_medium,
-    .bdrv_ioctl           = &raw_ioctl,
     .bdrv_aio_ioctl       = &raw_aio_ioctl,
     .create_opts          = &raw_create_opts,
     .bdrv_has_zero_init   = &raw_has_zero_init
diff --git a/block/sheepdog.c b/block/sheepdog.c
index 9585beb73..d80e4ed18 100644
--- a/block/sheepdog.c
+++ b/block/sheepdog.c
@@ -28,7 +28,6 @@
 #define SD_OP_READ_OBJ       0x02
 #define SD_OP_WRITE_OBJ      0x03
 /* 0x04 is used internally by Sheepdog */
-#define SD_OP_DISCARD_OBJ    0x05
 
 #define SD_OP_NEW_VDI        0x11
 #define SD_OP_LOCK_VDI       0x12
@@ -318,7 +317,7 @@ enum AIOCBState {
     AIOCB_DISCARD_OBJ,
 };
 
-#define AIOCBOverwrapping(x, y)                                 \
+#define AIOCBOverlapping(x, y)                                 \
     (!(x->max_affect_data_idx < y->min_affect_data_idx          \
        || y->max_affect_data_idx < x->min_affect_data_idx))
 
@@ -342,6 +341,15 @@ struct SheepdogAIOCB {
     uint32_t min_affect_data_idx;
     uint32_t max_affect_data_idx;
 
+    /*
+     * The difference between affect_data_idx and dirty_data_idx:
+     * affect_data_idx represents range of index of all request types.
+     * dirty_data_idx represents range of index updated by COW requests.
+     * dirty_data_idx is used for updating an inode object.
+     */
+    uint32_t min_dirty_data_idx;
+    uint32_t max_dirty_data_idx;
+
     QLIST_ENTRY(SheepdogAIOCB) aiocb_siblings;
 };
 
@@ -351,9 +359,6 @@ typedef struct BDRVSheepdogState {
 
     SheepdogInode inode;
 
-    uint32_t min_dirty_data_idx;
-    uint32_t max_dirty_data_idx;
-
     char name[SD_MAX_VDI_LEN];
     bool is_snapshot;
     uint32_t cache_flags;
@@ -373,10 +378,15 @@ typedef struct BDRVSheepdogState {
     QLIST_HEAD(inflight_aio_head, AIOReq) inflight_aio_head;
     QLIST_HEAD(failed_aio_head, AIOReq) failed_aio_head;
 
-    CoQueue overwrapping_queue;
+    CoQueue overlapping_queue;
     QLIST_HEAD(inflight_aiocb_head, SheepdogAIOCB) inflight_aiocb_head;
 } BDRVSheepdogState;
 
+typedef struct BDRVSheepdogReopenState {
+    int fd;
+    int cache_flags;
+} BDRVSheepdogReopenState;
+
 static const char * sd_strerror(int err)
 {
     int i;
@@ -556,6 +566,9 @@ static SheepdogAIOCB *sd_aio_setup(BlockDriverState *bs, QEMUIOVector *qiov,
     acb->max_affect_data_idx = (acb->sector_num * BDRV_SECTOR_SIZE +
                               acb->nb_sectors * BDRV_SECTOR_SIZE) / object_size;
 
+    acb->min_dirty_data_idx = UINT32_MAX;
+    acb->max_dirty_data_idx = 0;
+
     return acb;
 }
 
@@ -638,14 +651,16 @@ static coroutine_fn void do_co_req(void *opaque)
     unsigned int *rlen = srco->rlen;
 
     co = qemu_coroutine_self();
-    aio_set_fd_handler(srco->aio_context, sockfd, NULL, restart_co_req, co);
+    aio_set_fd_handler(srco->aio_context, sockfd, false,
+                       NULL, restart_co_req, co);
 
     ret = send_co_req(sockfd, hdr, data, wlen);
     if (ret < 0) {
         goto out;
     }
 
-    aio_set_fd_handler(srco->aio_context, sockfd, restart_co_req, NULL, co);
+    aio_set_fd_handler(srco->aio_context, sockfd, false,
+                       restart_co_req, NULL, co);
 
     ret = qemu_co_recv(sockfd, hdr, sizeof(*hdr));
     if (ret != sizeof(*hdr)) {
@@ -670,7 +685,8 @@ static coroutine_fn void do_co_req(void *opaque)
 out:
     /* there is at most one request for this sockfd, so it is safe to
      * set each handler to NULL. */
-    aio_set_fd_handler(srco->aio_context, sockfd, NULL, NULL, NULL);
+    aio_set_fd_handler(srco->aio_context, sockfd, false,
+                       NULL, NULL, NULL);
 
     srco->ret = ret;
     srco->finished = true;
@@ -722,7 +738,8 @@ static coroutine_fn void reconnect_to_sdog(void *opaque)
     BDRVSheepdogState *s = opaque;
     AIOReq *aio_req, *next;
 
-    aio_set_fd_handler(s->aio_context, s->fd, NULL, NULL, NULL);
+    aio_set_fd_handler(s->aio_context, s->fd, false, NULL,
+                       NULL, NULL);
     close(s->fd);
     s->fd = -1;
 
@@ -819,8 +836,8 @@ static void coroutine_fn aio_read_response(void *opaque)
              */
             if (rsp.result == SD_RES_SUCCESS) {
                 s->inode.data_vdi_id[idx] = s->inode.vdi_id;
-                s->max_dirty_data_idx = MAX(idx, s->max_dirty_data_idx);
-                s->min_dirty_data_idx = MIN(idx, s->min_dirty_data_idx);
+                acb->max_dirty_data_idx = MAX(idx, acb->max_dirty_data_idx);
+                acb->min_dirty_data_idx = MIN(idx, acb->min_dirty_data_idx);
             }
         }
         break;
@@ -847,10 +864,6 @@ static void coroutine_fn aio_read_response(void *opaque)
             rsp.result = SD_RES_SUCCESS;
             s->discard_supported = false;
             break;
-        case SD_RES_SUCCESS:
-            idx = data_oid_to_idx(aio_req->oid);
-            s->inode.data_vdi_id[idx] = 0;
-            break;
         default:
             break;
         }
@@ -929,7 +942,8 @@ static int get_sheep_fd(BDRVSheepdogState *s, Error **errp)
         return fd;
     }
 
-    aio_set_fd_handler(s->aio_context, fd, co_read_response, NULL, s);
+    aio_set_fd_handler(s->aio_context, fd, false,
+                       co_read_response, NULL, s);
     return fd;
 }
 
@@ -1165,7 +1179,13 @@ static void coroutine_fn add_aio_request(BDRVSheepdogState *s, AIOReq *aio_req,
         hdr.flags = SD_FLAG_CMD_WRITE | flags;
         break;
     case AIOCB_DISCARD_OBJ:
-        hdr.opcode = SD_OP_DISCARD_OBJ;
+        hdr.opcode = SD_OP_WRITE_OBJ;
+        hdr.flags = SD_FLAG_CMD_WRITE | flags;
+        s->inode.data_vdi_id[data_oid_to_idx(oid)] = 0;
+        offset = offsetof(SheepdogInode,
+                          data_vdi_id[data_oid_to_idx(oid)]);
+        oid = vid_to_vdi_oid(s->inode.vdi_id);
+        wlen = datalen = sizeof(uint32_t);
         break;
     }
 
@@ -1184,7 +1204,7 @@ static void coroutine_fn add_aio_request(BDRVSheepdogState *s, AIOReq *aio_req,
 
     qemu_co_mutex_lock(&s->lock);
     s->co_send = qemu_coroutine_self();
-    aio_set_fd_handler(s->aio_context, s->fd,
+    aio_set_fd_handler(s->aio_context, s->fd, false,
                        co_read_response, co_write_request, s);
     socket_set_cork(s->fd, 1);
 
@@ -1203,7 +1223,8 @@ static void coroutine_fn add_aio_request(BDRVSheepdogState *s, AIOReq *aio_req,
     }
 out:
     socket_set_cork(s->fd, 0);
-    aio_set_fd_handler(s->aio_context, s->fd, co_read_response, NULL, s);
+    aio_set_fd_handler(s->aio_context, s->fd, false,
+                       co_read_response, NULL, s);
     s->co_send = NULL;
     qemu_co_mutex_unlock(&s->lock);
 }
@@ -1353,7 +1374,8 @@ static void sd_detach_aio_context(BlockDriverState *bs)
 {
     BDRVSheepdogState *s = bs->opaque;
 
-    aio_set_fd_handler(s->aio_context, s->fd, NULL, NULL, NULL);
+    aio_set_fd_handler(s->aio_context, s->fd, false, NULL,
+                       NULL, NULL);
 }
 
 static void sd_attach_aio_context(BlockDriverState *bs,
@@ -1362,7 +1384,8 @@ static void sd_attach_aio_context(BlockDriverState *bs,
     BDRVSheepdogState *s = bs->opaque;
 
     s->aio_context = new_context;
-    aio_set_fd_handler(new_context, s->fd, co_read_response, NULL, s);
+    aio_set_fd_handler(new_context, s->fd, false,
+                       co_read_response, NULL, s);
 }
 
 /* TODO Convert to fine grained options */
@@ -1466,18 +1489,17 @@ static int sd_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     memcpy(&s->inode, buf, sizeof(s->inode));
-    s->min_dirty_data_idx = UINT32_MAX;
-    s->max_dirty_data_idx = 0;
 
     bs->total_sectors = s->inode.vdi_size / BDRV_SECTOR_SIZE;
     pstrcpy(s->name, sizeof(s->name), vdi);
     qemu_co_mutex_init(&s->lock);
-    qemu_co_queue_init(&s->overwrapping_queue);
+    qemu_co_queue_init(&s->overlapping_queue);
     qemu_opts_del(opts);
     g_free(buf);
     return 0;
 out:
-    aio_set_fd_handler(bdrv_get_aio_context(bs), s->fd, NULL, NULL, NULL);
+    aio_set_fd_handler(bdrv_get_aio_context(bs), s->fd,
+                       false, NULL, NULL, NULL);
     if (s->fd >= 0) {
         closesocket(s->fd);
     }
@@ -1486,6 +1508,70 @@ out:
     return ret;
 }
 
+static int sd_reopen_prepare(BDRVReopenState *state, BlockReopenQueue *queue,
+                             Error **errp)
+{
+    BDRVSheepdogState *s = state->bs->opaque;
+    BDRVSheepdogReopenState *re_s;
+    int ret = 0;
+
+    re_s = state->opaque = g_new0(BDRVSheepdogReopenState, 1);
+
+    re_s->cache_flags = SD_FLAG_CMD_CACHE;
+    if (state->flags & BDRV_O_NOCACHE) {
+        re_s->cache_flags = SD_FLAG_CMD_DIRECT;
+    }
+
+    re_s->fd = get_sheep_fd(s, errp);
+    if (re_s->fd < 0) {
+        ret = re_s->fd;
+        return ret;
+    }
+
+    return ret;
+}
+
+static void sd_reopen_commit(BDRVReopenState *state)
+{
+    BDRVSheepdogReopenState *re_s = state->opaque;
+    BDRVSheepdogState *s = state->bs->opaque;
+
+    if (s->fd) {
+        aio_set_fd_handler(s->aio_context, s->fd, false,
+                           NULL, NULL, NULL);
+        closesocket(s->fd);
+    }
+
+    s->fd = re_s->fd;
+    s->cache_flags = re_s->cache_flags;
+
+    g_free(state->opaque);
+    state->opaque = NULL;
+
+    return;
+}
+
+static void sd_reopen_abort(BDRVReopenState *state)
+{
+    BDRVSheepdogReopenState *re_s = state->opaque;
+    BDRVSheepdogState *s = state->bs->opaque;
+
+    if (re_s == NULL) {
+        return;
+    }
+
+    if (re_s->fd) {
+        aio_set_fd_handler(s->aio_context, re_s->fd, false,
+                           NULL, NULL, NULL);
+        closesocket(re_s->fd);
+    }
+
+    g_free(state->opaque);
+    state->opaque = NULL;
+
+    return;
+}
+
 static int do_sd_create(BDRVSheepdogState *s, uint32_t *vdi_id, int snapshot,
                         Error **errp)
 {
@@ -1554,7 +1640,7 @@ static int sd_prealloc(const char *filename, Error **errp)
     int ret;
 
     ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
-                    NULL, errp);
+                    errp);
     if (ret < 0) {
         goto out_with_err_set;
     }
@@ -1746,8 +1832,7 @@ static int sd_create(const char *filename, QemuOpts *opts,
         }
 
         bs = NULL;
-        ret = bdrv_open(&bs, backing_file, NULL, NULL, BDRV_O_PROTOCOL, NULL,
-                        errp);
+        ret = bdrv_open(&bs, backing_file, NULL, NULL, BDRV_O_PROTOCOL, errp);
         if (ret < 0) {
             goto out;
         }
@@ -1861,7 +1946,8 @@ static void sd_close(BlockDriverState *bs)
         error_report("%s, %s", sd_strerror(rsp->result), s->name);
     }
 
-    aio_set_fd_handler(bdrv_get_aio_context(bs), s->fd, NULL, NULL, NULL);
+    aio_set_fd_handler(bdrv_get_aio_context(bs), s->fd,
+                       false, NULL, NULL, NULL);
     closesocket(s->fd);
     g_free(s->host_spec);
 }
@@ -1923,16 +2009,16 @@ static void coroutine_fn sd_write_done(SheepdogAIOCB *acb)
     AIOReq *aio_req;
     uint32_t offset, data_len, mn, mx;
 
-    mn = s->min_dirty_data_idx;
-    mx = s->max_dirty_data_idx;
+    mn = acb->min_dirty_data_idx;
+    mx = acb->max_dirty_data_idx;
     if (mn <= mx) {
         /* we need to update the vdi object. */
         offset = sizeof(s->inode) - sizeof(s->inode.data_vdi_id) +
             mn * sizeof(s->inode.data_vdi_id[0]);
         data_len = (mx - mn + 1) * sizeof(s->inode.data_vdi_id[0]);
 
-        s->min_dirty_data_idx = UINT32_MAX;
-        s->max_dirty_data_idx = 0;
+        acb->min_dirty_data_idx = UINT32_MAX;
+        acb->max_dirty_data_idx = 0;
 
         iov.iov_base = &s->inode;
         iov.iov_len = sizeof(s->inode);
@@ -2141,7 +2227,9 @@ static int coroutine_fn sd_co_rw_vector(void *p)
         }
 
         aio_req = alloc_aio_req(s, acb, oid, len, offset, flags, create,
-                                old_oid, done);
+                                old_oid,
+                                acb->aiocb_type == AIOCB_DISCARD_OBJ ?
+                                0 : done);
         QLIST_INSERT_HEAD(&s->inflight_aio_head, aio_req, aio_siblings);
 
         add_aio_request(s, aio_req, acb->qiov->iov, acb->qiov->niov,
@@ -2158,12 +2246,12 @@ out:
     return 1;
 }
 
-static bool check_overwrapping_aiocb(BDRVSheepdogState *s, SheepdogAIOCB *aiocb)
+static bool check_overlapping_aiocb(BDRVSheepdogState *s, SheepdogAIOCB *aiocb)
 {
     SheepdogAIOCB *cb;
 
     QLIST_FOREACH(cb, &s->inflight_aiocb_head, aiocb_siblings) {
-        if (AIOCBOverwrapping(aiocb, cb)) {
+        if (AIOCBOverlapping(aiocb, cb)) {
             return true;
         }
     }
@@ -2192,15 +2280,15 @@ static coroutine_fn int sd_co_writev(BlockDriverState *bs, int64_t sector_num,
     acb->aiocb_type = AIOCB_WRITE_UDATA;
 
 retry:
-    if (check_overwrapping_aiocb(s, acb)) {
-        qemu_co_queue_wait(&s->overwrapping_queue);
+    if (check_overlapping_aiocb(s, acb)) {
+        qemu_co_queue_wait(&s->overlapping_queue);
         goto retry;
     }
 
     ret = sd_co_rw_vector(acb);
     if (ret <= 0) {
         QLIST_REMOVE(acb, aiocb_siblings);
-        qemu_co_queue_restart_all(&s->overwrapping_queue);
+        qemu_co_queue_restart_all(&s->overlapping_queue);
         qemu_aio_unref(acb);
         return ret;
     }
@@ -2208,7 +2296,7 @@ retry:
     qemu_coroutine_yield();
 
     QLIST_REMOVE(acb, aiocb_siblings);
-    qemu_co_queue_restart_all(&s->overwrapping_queue);
+    qemu_co_queue_restart_all(&s->overlapping_queue);
 
     return acb->ret;
 }
@@ -2225,15 +2313,15 @@ static coroutine_fn int sd_co_readv(BlockDriverState *bs, int64_t sector_num,
     acb->aio_done_func = sd_finish_aiocb;
 
 retry:
-    if (check_overwrapping_aiocb(s, acb)) {
-        qemu_co_queue_wait(&s->overwrapping_queue);
+    if (check_overlapping_aiocb(s, acb)) {
+        qemu_co_queue_wait(&s->overlapping_queue);
         goto retry;
     }
 
     ret = sd_co_rw_vector(acb);
     if (ret <= 0) {
         QLIST_REMOVE(acb, aiocb_siblings);
-        qemu_co_queue_restart_all(&s->overwrapping_queue);
+        qemu_co_queue_restart_all(&s->overlapping_queue);
         qemu_aio_unref(acb);
         return ret;
     }
@@ -2241,7 +2329,7 @@ retry:
     qemu_coroutine_yield();
 
     QLIST_REMOVE(acb, aiocb_siblings);
-    qemu_co_queue_restart_all(&s->overwrapping_queue);
+    qemu_co_queue_restart_all(&s->overlapping_queue);
     return acb->ret;
 }
 
@@ -2577,28 +2665,36 @@ static coroutine_fn int sd_co_discard(BlockDriverState *bs, int64_t sector_num,
                                       int nb_sectors)
 {
     SheepdogAIOCB *acb;
-    QEMUIOVector dummy;
     BDRVSheepdogState *s = bs->opaque;
     int ret;
+    QEMUIOVector discard_iov;
+    struct iovec iov;
+    uint32_t zero = 0;
 
     if (!s->discard_supported) {
             return 0;
     }
 
-    acb = sd_aio_setup(bs, &dummy, sector_num, nb_sectors);
+    memset(&discard_iov, 0, sizeof(discard_iov));
+    memset(&iov, 0, sizeof(iov));
+    iov.iov_base = &zero;
+    iov.iov_len = sizeof(zero);
+    discard_iov.iov = &iov;
+    discard_iov.niov = 1;
+    acb = sd_aio_setup(bs, &discard_iov, sector_num, nb_sectors);
     acb->aiocb_type = AIOCB_DISCARD_OBJ;
     acb->aio_done_func = sd_finish_aiocb;
 
 retry:
-    if (check_overwrapping_aiocb(s, acb)) {
-        qemu_co_queue_wait(&s->overwrapping_queue);
+    if (check_overlapping_aiocb(s, acb)) {
+        qemu_co_queue_wait(&s->overlapping_queue);
         goto retry;
     }
 
     ret = sd_co_rw_vector(acb);
     if (ret <= 0) {
         QLIST_REMOVE(acb, aiocb_siblings);
-        qemu_co_queue_restart_all(&s->overwrapping_queue);
+        qemu_co_queue_restart_all(&s->overlapping_queue);
         qemu_aio_unref(acb);
         return ret;
     }
@@ -2606,7 +2702,7 @@ retry:
     qemu_coroutine_yield();
 
     QLIST_REMOVE(acb, aiocb_siblings);
-    qemu_co_queue_restart_all(&s->overwrapping_queue);
+    qemu_co_queue_restart_all(&s->overlapping_queue);
 
     return acb->ret;
 }
@@ -2703,6 +2799,9 @@ static BlockDriver bdrv_sheepdog = {
     .instance_size  = sizeof(BDRVSheepdogState),
     .bdrv_needs_filename = true,
     .bdrv_file_open = sd_open,
+    .bdrv_reopen_prepare    = sd_reopen_prepare,
+    .bdrv_reopen_commit     = sd_reopen_commit,
+    .bdrv_reopen_abort      = sd_reopen_abort,
     .bdrv_close     = sd_close,
     .bdrv_create    = sd_create,
     .bdrv_has_zero_init = bdrv_has_zero_init_1,
@@ -2736,6 +2835,9 @@ static BlockDriver bdrv_sheepdog_tcp = {
     .instance_size  = sizeof(BDRVSheepdogState),
     .bdrv_needs_filename = true,
     .bdrv_file_open = sd_open,
+    .bdrv_reopen_prepare    = sd_reopen_prepare,
+    .bdrv_reopen_commit     = sd_reopen_commit,
+    .bdrv_reopen_abort      = sd_reopen_abort,
     .bdrv_close     = sd_close,
     .bdrv_create    = sd_create,
     .bdrv_has_zero_init = bdrv_has_zero_init_1,
@@ -2769,6 +2871,9 @@ static BlockDriver bdrv_sheepdog_unix = {
     .instance_size  = sizeof(BDRVSheepdogState),
     .bdrv_needs_filename = true,
     .bdrv_file_open = sd_open,
+    .bdrv_reopen_prepare    = sd_reopen_prepare,
+    .bdrv_reopen_commit     = sd_reopen_commit,
+    .bdrv_reopen_abort      = sd_reopen_abort,
     .bdrv_close     = sd_close,
     .bdrv_create    = sd_create,
     .bdrv_has_zero_init = bdrv_has_zero_init_1,
diff --git a/block/snapshot.c b/block/snapshot.c
index 49e143e99..6e9fa8da9 100644
--- a/block/snapshot.c
+++ b/block/snapshot.c
@@ -149,7 +149,7 @@ int bdrv_can_snapshot(BlockDriverState *bs)
 
     if (!drv->bdrv_snapshot_create) {
         if (bs->file != NULL) {
-            return bdrv_can_snapshot(bs->file);
+            return bdrv_can_snapshot(bs->file->bs);
         }
         return 0;
     }
@@ -168,7 +168,7 @@ int bdrv_snapshot_create(BlockDriverState *bs,
         return drv->bdrv_snapshot_create(bs, sn_info);
     }
     if (bs->file) {
-        return bdrv_snapshot_create(bs->file, sn_info);
+        return bdrv_snapshot_create(bs->file->bs, sn_info);
     }
     return -ENOTSUP;
 }
@@ -188,10 +188,10 @@ int bdrv_snapshot_goto(BlockDriverState *bs,
 
     if (bs->file) {
         drv->bdrv_close(bs);
-        ret = bdrv_snapshot_goto(bs->file, snapshot_id);
+        ret = bdrv_snapshot_goto(bs->file->bs, snapshot_id);
         open_ret = drv->bdrv_open(bs, NULL, bs->open_flags, NULL);
         if (open_ret < 0) {
-            bdrv_unref(bs->file);
+            bdrv_unref(bs->file->bs);
             bs->drv = NULL;
             return open_ret;
         }
@@ -245,7 +245,7 @@ int bdrv_snapshot_delete(BlockDriverState *bs,
         return drv->bdrv_snapshot_delete(bs, snapshot_id, name, errp);
     }
     if (bs->file) {
-        return bdrv_snapshot_delete(bs->file, snapshot_id, name, errp);
+        return bdrv_snapshot_delete(bs->file->bs, snapshot_id, name, errp);
     }
     error_setg(errp, "Block format '%s' used by device '%s' "
                "does not support internal snapshot deletion",
@@ -253,9 +253,9 @@ int bdrv_snapshot_delete(BlockDriverState *bs,
     return -ENOTSUP;
 }
 
-void bdrv_snapshot_delete_by_id_or_name(BlockDriverState *bs,
-                                        const char *id_or_name,
-                                        Error **errp)
+int bdrv_snapshot_delete_by_id_or_name(BlockDriverState *bs,
+                                       const char *id_or_name,
+                                       Error **errp)
 {
     int ret;
     Error *local_err = NULL;
@@ -270,6 +270,7 @@ void bdrv_snapshot_delete_by_id_or_name(BlockDriverState *bs,
     if (ret < 0) {
         error_propagate(errp, local_err);
     }
+    return ret;
 }
 
 int bdrv_snapshot_list(BlockDriverState *bs,
@@ -283,7 +284,7 @@ int bdrv_snapshot_list(BlockDriverState *bs,
         return drv->bdrv_snapshot_list(bs, psn_info);
     }
     if (bs->file) {
-        return bdrv_snapshot_list(bs->file, psn_info);
+        return bdrv_snapshot_list(bs->file->bs, psn_info);
     }
     return -ENOTSUP;
 }
@@ -356,3 +357,130 @@ int bdrv_snapshot_load_tmp_by_id_or_name(BlockDriverState *bs,
 
     return ret;
 }
+
+
+/* Group operations. All block drivers are involved.
+ * These functions will properly handle dataplane (take aio_context_acquire
+ * when appropriate for appropriate block drivers) */
+
+bool bdrv_all_can_snapshot(BlockDriverState **first_bad_bs)
+{
+    bool ok = true;
+    BlockDriverState *bs = NULL;
+
+    while (ok && (bs = bdrv_next(bs))) {
+        AioContext *ctx = bdrv_get_aio_context(bs);
+
+        aio_context_acquire(ctx);
+        if (bdrv_is_inserted(bs) && !bdrv_is_read_only(bs)) {
+            ok = bdrv_can_snapshot(bs);
+        }
+        aio_context_release(ctx);
+    }
+
+    *first_bad_bs = bs;
+    return ok;
+}
+
+int bdrv_all_delete_snapshot(const char *name, BlockDriverState **first_bad_bs,
+                             Error **err)
+{
+    int ret = 0;
+    BlockDriverState *bs = NULL;
+    QEMUSnapshotInfo sn1, *snapshot = &sn1;
+
+    while (ret == 0 && (bs = bdrv_next(bs))) {
+        AioContext *ctx = bdrv_get_aio_context(bs);
+
+        aio_context_acquire(ctx);
+        if (bdrv_can_snapshot(bs) &&
+                bdrv_snapshot_find(bs, snapshot, name) >= 0) {
+            ret = bdrv_snapshot_delete_by_id_or_name(bs, name, err);
+        }
+        aio_context_release(ctx);
+    }
+
+    *first_bad_bs = bs;
+    return ret;
+}
+
+
+int bdrv_all_goto_snapshot(const char *name, BlockDriverState **first_bad_bs)
+{
+    int err = 0;
+    BlockDriverState *bs = NULL;
+
+    while (err == 0 && (bs = bdrv_next(bs))) {
+        AioContext *ctx = bdrv_get_aio_context(bs);
+
+        aio_context_acquire(ctx);
+        if (bdrv_can_snapshot(bs)) {
+            err = bdrv_snapshot_goto(bs, name);
+        }
+        aio_context_release(ctx);
+    }
+
+    *first_bad_bs = bs;
+    return err;
+}
+
+int bdrv_all_find_snapshot(const char *name, BlockDriverState **first_bad_bs)
+{
+    QEMUSnapshotInfo sn;
+    int err = 0;
+    BlockDriverState *bs = NULL;
+
+    while (err == 0 && (bs = bdrv_next(bs))) {
+        AioContext *ctx = bdrv_get_aio_context(bs);
+
+        aio_context_acquire(ctx);
+        if (bdrv_can_snapshot(bs)) {
+            err = bdrv_snapshot_find(bs, &sn, name);
+        }
+        aio_context_release(ctx);
+    }
+
+    *first_bad_bs = bs;
+    return err;
+}
+
+int bdrv_all_create_snapshot(QEMUSnapshotInfo *sn,
+                             BlockDriverState *vm_state_bs,
+                             uint64_t vm_state_size,
+                             BlockDriverState **first_bad_bs)
+{
+    int err = 0;
+    BlockDriverState *bs = NULL;
+
+    while (err == 0 && (bs = bdrv_next(bs))) {
+        AioContext *ctx = bdrv_get_aio_context(bs);
+
+        aio_context_acquire(ctx);
+        if (bs == vm_state_bs) {
+            sn->vm_state_size = vm_state_size;
+            err = bdrv_snapshot_create(bs, sn);
+        } else if (bdrv_can_snapshot(bs)) {
+            sn->vm_state_size = 0;
+            err = bdrv_snapshot_create(bs, sn);
+        }
+        aio_context_release(ctx);
+    }
+
+    *first_bad_bs = bs;
+    return err;
+}
+
+BlockDriverState *bdrv_all_find_vmstate_bs(void)
+{
+    bool not_found = true;
+    BlockDriverState *bs = NULL;
+
+    while (not_found && (bs = bdrv_next(bs))) {
+        AioContext *ctx = bdrv_get_aio_context(bs);
+
+        aio_context_acquire(ctx);
+        not_found = !bdrv_can_snapshot(bs);
+        aio_context_release(ctx);
+    }
+    return bs;
+}
diff --git a/block/ssh.c b/block/ssh.c
index 8d0673903..af025c08a 100644
--- a/block/ssh.c
+++ b/block/ssh.c
@@ -193,7 +193,7 @@ sftp_error_report(BDRVSSHState *s, const char *fs, ...)
 static int parse_uri(const char *filename, QDict *options, Error **errp)
 {
     URI *uri = NULL;
-    QueryParams *qp = NULL;
+    QueryParams *qp;
     int i;
 
     uri = uri_parse(filename);
@@ -249,9 +249,6 @@ static int parse_uri(const char *filename, QDict *options, Error **errp)
     return 0;
 
  err:
-    if (qp) {
-      query_params_free(qp);
-    }
     if (uri) {
       uri_free(uri);
     }
@@ -803,14 +800,15 @@ static coroutine_fn void set_fd_handler(BDRVSSHState *s, BlockDriverState *bs)
             rd_handler, wr_handler);
 
     aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock,
-                       rd_handler, wr_handler, co);
+                       false, rd_handler, wr_handler, co);
 }
 
 static coroutine_fn void clear_fd_handler(BDRVSSHState *s,
                                           BlockDriverState *bs)
 {
     DPRINTF("s->sock=%d", s->sock);
-    aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock, NULL, NULL, NULL);
+    aio_set_fd_handler(bdrv_get_aio_context(bs), s->sock,
+                       false, NULL, NULL, NULL);
 }
 
 /* A non-blocking call returned EAGAIN, so yield, ensuring the
diff --git a/block/stream.c b/block/stream.c
index ab0bd057f..25af7eff6 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -16,6 +16,7 @@
 #include "block/blockjob.h"
 #include "qapi/qmp/qerror.h"
 #include "qemu/ratelimit.h"
+#include "sysemu/block-backend.h"
 
 enum {
     /*
@@ -52,34 +53,6 @@ static int coroutine_fn stream_populate(BlockDriverState *bs,
     return bdrv_co_copy_on_readv(bs, sector_num, nb_sectors, &qiov);
 }
 
-static void close_unused_images(BlockDriverState *top, BlockDriverState *base,
-                                const char *base_id)
-{
-    BlockDriverState *intermediate;
-    intermediate = top->backing_hd;
-
-    /* Must assign before bdrv_delete() to prevent traversing dangling pointer
-     * while we delete backing image instances.
-     */
-    bdrv_set_backing_hd(top, base);
-
-    while (intermediate) {
-        BlockDriverState *unused;
-
-        /* reached base */
-        if (intermediate == base) {
-            break;
-        }
-
-        unused = intermediate;
-        intermediate = intermediate->backing_hd;
-        bdrv_set_backing_hd(unused, NULL);
-        bdrv_unref(unused);
-    }
-
-    bdrv_refresh_limits(top, NULL);
-}
-
 typedef struct {
     int ret;
     bool reached_end;
@@ -101,7 +74,7 @@ static void stream_complete(BlockJob *job, void *opaque)
             }
         }
         data->ret = bdrv_change_backing_file(job->bs, base_id, base_fmt);
-        close_unused_images(job->bs, base, base_id);
+        bdrv_set_backing_hd(job->bs, base);
     }
 
     g_free(s->backing_file_str);
@@ -121,7 +94,7 @@ static void coroutine_fn stream_run(void *opaque)
     int n = 0;
     void *buf;
 
-    if (!bs->backing_hd) {
+    if (!bs->backing) {
         block_job_completed(&s->common, 0);
         return;
     }
@@ -166,7 +139,7 @@ wait:
         } else if (ret >= 0) {
             /* Copy if allocated in the intermediate images.  Limit to the
              * known-unallocated area [sector_num, sector_num+n).  */
-            ret = bdrv_is_allocated_above(bs->backing_hd, base,
+            ret = bdrv_is_allocated_above(backing_bs(bs), base,
                                           sector_num, n, &n);
 
             /* Finish early if end of backing file has been reached */
@@ -250,7 +223,7 @@ void stream_start(BlockDriverState *bs, BlockDriverState *base,
 
     if ((on_error == BLOCKDEV_ON_ERROR_STOP ||
          on_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
-        !bdrv_iostatus_is_enabled(bs)) {
+        (!bs->blk || !blk_iostatus_is_enabled(bs->blk))) {
         error_setg(errp, QERR_INVALID_PARAMETER, "on-error");
         return;
     }
diff --git a/block/throttle-groups.c b/block/throttle-groups.c
index 1abc6fcae..13b5baa5d 100644
--- a/block/throttle-groups.c
+++ b/block/throttle-groups.c
@@ -33,8 +33,7 @@
  * its own locking.
  *
  * This locking is however handled internally in this file, so it's
- * mostly transparent to outside users (but see the documentation in
- * throttle_groups_lock()).
+ * transparent to outside users.
  *
  * The whole ThrottleGroup structure is private and invisible to
  * outside users, that only use it through its ThrottleState.
@@ -76,9 +75,9 @@ static QTAILQ_HEAD(, ThrottleGroup) throttle_groups =
  * created.
  *
  * @name: the name of the ThrottleGroup
- * @ret:  the ThrottleGroup
+ * @ret:  the ThrottleState member of the ThrottleGroup
  */
-static ThrottleGroup *throttle_group_incref(const char *name)
+ThrottleState *throttle_group_incref(const char *name)
 {
     ThrottleGroup *tg = NULL;
     ThrottleGroup *iter;
@@ -108,7 +107,7 @@ static ThrottleGroup *throttle_group_incref(const char *name)
 
     qemu_mutex_unlock(&throttle_groups_lock);
 
-    return tg;
+    return &tg->ts;
 }
 
 /* Decrease the reference count of a ThrottleGroup.
@@ -116,10 +115,12 @@ static ThrottleGroup *throttle_group_incref(const char *name)
  * When the reference count reaches zero the ThrottleGroup is
  * destroyed.
  *
- * @tg:  The ThrottleGroup to unref
+ * @ts:  The ThrottleGroup to unref, given by its ThrottleState member
  */
-static void throttle_group_unref(ThrottleGroup *tg)
+void throttle_group_unref(ThrottleState *ts)
 {
+    ThrottleGroup *tg = container_of(ts, ThrottleGroup, ts);
+
     qemu_mutex_lock(&throttle_groups_lock);
     if (--tg->refcount == 0) {
         QTAILQ_REMOVE(&throttle_groups, tg, list);
@@ -401,7 +402,8 @@ static void write_timer_cb(void *opaque)
 void throttle_group_register_bs(BlockDriverState *bs, const char *groupname)
 {
     int i;
-    ThrottleGroup *tg = throttle_group_incref(groupname);
+    ThrottleState *ts = throttle_group_incref(groupname);
+    ThrottleGroup *tg = container_of(ts, ThrottleGroup, ts);
     int clock_type = QEMU_CLOCK_REALTIME;
 
     if (qtest_enabled()) {
@@ -409,7 +411,7 @@ void throttle_group_register_bs(BlockDriverState *bs, const char *groupname)
         clock_type = QEMU_CLOCK_VIRTUAL;
     }
 
-    bs->throttle_state = &tg->ts;
+    bs->throttle_state = ts;
 
     qemu_mutex_lock(&tg->lock);
     /* If the ThrottleGroup is new set this BlockDriverState as the token */
@@ -435,6 +437,9 @@ void throttle_group_register_bs(BlockDriverState *bs, const char *groupname)
  * list, destroying the timers and setting the throttle_state pointer
  * to NULL.
  *
+ * The BlockDriverState must not have pending throttled requests, so
+ * the caller has to drain them first.
+ *
  * The group will be destroyed if it's empty after this operation.
  *
  * @bs: the BlockDriverState to remove
@@ -444,6 +449,10 @@ void throttle_group_unregister_bs(BlockDriverState *bs)
     ThrottleGroup *tg = container_of(bs->throttle_state, ThrottleGroup, ts);
     int i;
 
+    assert(bs->pending_reqs[0] == 0 && bs->pending_reqs[1] == 0);
+    assert(qemu_co_queue_empty(&bs->throttled_reqs[0]));
+    assert(qemu_co_queue_empty(&bs->throttled_reqs[1]));
+
     qemu_mutex_lock(&tg->lock);
     for (i = 0; i < 2; i++) {
         if (tg->tokens[i] == bs) {
@@ -461,38 +470,10 @@ void throttle_group_unregister_bs(BlockDriverState *bs)
     throttle_timers_destroy(&bs->throttle_timers);
     qemu_mutex_unlock(&tg->lock);
 
-    throttle_group_unref(tg);
+    throttle_group_unref(&tg->ts);
     bs->throttle_state = NULL;
 }
 
-/* Acquire the lock of this throttling group.
- *
- * You won't normally need to use this. None of the functions from the
- * ThrottleGroup API require you to acquire the lock since all of them
- * deal with it internally.
- *
- * This should only be used in exceptional cases when you want to
- * access the protected fields of a BlockDriverState directly
- * (e.g. bdrv_swap()).
- *
- * @bs: a BlockDriverState that is member of the group
- */
-void throttle_group_lock(BlockDriverState *bs)
-{
-    ThrottleGroup *tg = container_of(bs->throttle_state, ThrottleGroup, ts);
-    qemu_mutex_lock(&tg->lock);
-}
-
-/* Release the lock of this throttling group.
- *
- * See the comments in throttle_group_lock().
- */
-void throttle_group_unlock(BlockDriverState *bs)
-{
-    ThrottleGroup *tg = container_of(bs->throttle_state, ThrottleGroup, ts);
-    qemu_mutex_unlock(&tg->lock);
-}
-
 static void throttle_groups_init(void)
 {
     qemu_mutex_init(&throttle_groups_lock);
diff --git a/block/vdi.c b/block/vdi.c
index 7642ef359..17f435fad 100644
--- a/block/vdi.c
+++ b/block/vdi.c
@@ -53,7 +53,7 @@
 #include "block/block_int.h"
 #include "qemu/module.h"
 #include "migration/migration.h"
-#include "block/coroutine.h"
+#include "qemu/coroutine.h"
 
 #if defined(CONFIG_UUID)
 #include <uuid/uuid.h>
@@ -399,7 +399,7 @@ static int vdi_open(BlockDriverState *bs, QDict *options, int flags,
 
     logout("\n");
 
-    ret = bdrv_read(bs->file, 0, (uint8_t *)&header, 1);
+    ret = bdrv_read(bs->file->bs, 0, (uint8_t *)&header, 1);
     if (ret < 0) {
         goto fail;
     }
@@ -490,13 +490,14 @@ static int vdi_open(BlockDriverState *bs, QDict *options, int flags,
 
     bmap_size = header.blocks_in_image * sizeof(uint32_t);
     bmap_size = DIV_ROUND_UP(bmap_size, SECTOR_SIZE);
-    s->bmap = qemu_try_blockalign(bs->file, bmap_size * SECTOR_SIZE);
+    s->bmap = qemu_try_blockalign(bs->file->bs, bmap_size * SECTOR_SIZE);
     if (s->bmap == NULL) {
         ret = -ENOMEM;
         goto fail;
     }
 
-    ret = bdrv_read(bs->file, s->bmap_sector, (uint8_t *)s->bmap, bmap_size);
+    ret = bdrv_read(bs->file->bs, s->bmap_sector, (uint8_t *)s->bmap,
+                    bmap_size);
     if (ret < 0) {
         goto fail_free_bmap;
     }
@@ -585,7 +586,7 @@ static int vdi_co_read(BlockDriverState *bs,
             uint64_t offset = s->header.offset_data / SECTOR_SIZE +
                               (uint64_t)bmap_entry * s->block_sectors +
                               sector_in_block;
-            ret = bdrv_read(bs->file, offset, buf, n_sectors);
+            ret = bdrv_read(bs->file->bs, offset, buf, n_sectors);
         }
         logout("%u sectors read\n", n_sectors);
 
@@ -653,7 +654,7 @@ static int vdi_co_write(BlockDriverState *bs,
              * acquire the lock and thus the padded cluster is written before
              * the other coroutines can write to the affected area. */
             qemu_co_mutex_lock(&s->write_lock);
-            ret = bdrv_write(bs->file, offset, block, s->block_sectors);
+            ret = bdrv_write(bs->file->bs, offset, block, s->block_sectors);
             qemu_co_mutex_unlock(&s->write_lock);
         } else {
             uint64_t offset = s->header.offset_data / SECTOR_SIZE +
@@ -669,7 +670,7 @@ static int vdi_co_write(BlockDriverState *bs,
              * that that write operation has returned (there may be other writes
              * in flight, but they do not concern this very operation). */
             qemu_co_mutex_unlock(&s->write_lock);
-            ret = bdrv_write(bs->file, offset, buf, n_sectors);
+            ret = bdrv_write(bs->file->bs, offset, buf, n_sectors);
         }
 
         nb_sectors -= n_sectors;
@@ -694,7 +695,7 @@ static int vdi_co_write(BlockDriverState *bs,
         assert(VDI_IS_ALLOCATED(bmap_first));
         *header = s->header;
         vdi_header_to_le(header);
-        ret = bdrv_write(bs->file, 0, block, 1);
+        ret = bdrv_write(bs->file->bs, 0, block, 1);
         g_free(block);
         block = NULL;
 
@@ -712,7 +713,7 @@ static int vdi_co_write(BlockDriverState *bs,
         base = ((uint8_t *)&s->bmap[0]) + bmap_first * SECTOR_SIZE;
         logout("will write %u block map sectors starting from entry %u\n",
                n_sectors, bmap_first);
-        ret = bdrv_write(bs->file, offset, base, n_sectors);
+        ret = bdrv_write(bs->file->bs, offset, base, n_sectors);
     }
 
     return ret;
@@ -764,7 +765,7 @@ static int vdi_create(const char *filename, QemuOpts *opts, Error **errp)
         goto exit;
     }
     ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
-                    NULL, &local_err);
+                    &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto exit;
diff --git a/block/vhdx-log.c b/block/vhdx-log.c
index 47fec63c6..47ae4b135 100644
--- a/block/vhdx-log.c
+++ b/block/vhdx-log.c
@@ -81,7 +81,7 @@ static int vhdx_log_peek_hdr(BlockDriverState *bs, VHDXLogEntries *log,
 
     offset = log->offset + read;
 
-    ret = bdrv_pread(bs->file, offset, hdr, sizeof(VHDXLogEntryHeader));
+    ret = bdrv_pread(bs->file->bs, offset, hdr, sizeof(VHDXLogEntryHeader));
     if (ret < 0) {
         goto exit;
     }
@@ -141,7 +141,7 @@ static int vhdx_log_read_sectors(BlockDriverState *bs, VHDXLogEntries *log,
         }
         offset = log->offset + read;
 
-        ret = bdrv_pread(bs->file, offset, buffer, VHDX_LOG_SECTOR_SIZE);
+        ret = bdrv_pread(bs->file->bs, offset, buffer, VHDX_LOG_SECTOR_SIZE);
         if (ret < 0) {
             goto exit;
         }
@@ -191,7 +191,8 @@ static int vhdx_log_write_sectors(BlockDriverState *bs, VHDXLogEntries *log,
             /* full */
             break;
         }
-        ret = bdrv_pwrite(bs->file, offset, buffer_tmp, VHDX_LOG_SECTOR_SIZE);
+        ret = bdrv_pwrite(bs->file->bs, offset, buffer_tmp,
+                          VHDX_LOG_SECTOR_SIZE);
         if (ret < 0) {
             goto exit;
         }
@@ -353,7 +354,7 @@ static int vhdx_log_read_desc(BlockDriverState *bs, BDRVVHDXState *s,
     }
 
     desc_sectors = vhdx_compute_desc_sectors(hdr.descriptor_count);
-    desc_entries = qemu_try_blockalign(bs->file,
+    desc_entries = qemu_try_blockalign(bs->file->bs,
                                        desc_sectors * VHDX_LOG_SECTOR_SIZE);
     if (desc_entries == NULL) {
         ret = -ENOMEM;
@@ -462,7 +463,7 @@ static int vhdx_log_flush_desc(BlockDriverState *bs, VHDXLogDescriptor *desc,
 
     /* count is only > 1 if we are writing zeroes */
     for (i = 0; i < count; i++) {
-        ret = bdrv_pwrite_sync(bs->file, file_offset, buffer,
+        ret = bdrv_pwrite_sync(bs->file->bs, file_offset, buffer,
                                VHDX_LOG_SECTOR_SIZE);
         if (ret < 0) {
             goto exit;
@@ -509,7 +510,7 @@ static int vhdx_log_flush(BlockDriverState *bs, BDRVVHDXState *s,
         /* if the log shows a FlushedFileOffset larger than our current file
          * size, then that means the file has been truncated / corrupted, and
          * we must refused to open it / use it */
-        if (hdr_tmp.flushed_file_offset > bdrv_getlength(bs->file)) {
+        if (hdr_tmp.flushed_file_offset > bdrv_getlength(bs->file->bs)) {
             ret = -EINVAL;
             goto exit;
         }
@@ -539,12 +540,12 @@ static int vhdx_log_flush(BlockDriverState *bs, BDRVVHDXState *s,
                 goto exit;
             }
         }
-        if (bdrv_getlength(bs->file) < desc_entries->hdr.last_file_offset) {
+        if (bdrv_getlength(bs->file->bs) < desc_entries->hdr.last_file_offset) {
             new_file_size = desc_entries->hdr.last_file_offset;
             if (new_file_size % (1024*1024)) {
                 /* round up to nearest 1MB boundary */
                 new_file_size = ((new_file_size >> 20) + 1) << 20;
-                bdrv_truncate(bs->file, new_file_size);
+                bdrv_truncate(bs->file->bs, new_file_size);
             }
         }
         qemu_vfree(desc_entries);
@@ -908,8 +909,8 @@ static int vhdx_log_write(BlockDriverState *bs, BDRVVHDXState *s,
                 .sequence_number     = s->log.sequence,
                 .descriptor_count    = sectors,
                 .reserved            = 0,
-                .flushed_file_offset = bdrv_getlength(bs->file),
-                .last_file_offset    = bdrv_getlength(bs->file),
+                .flushed_file_offset = bdrv_getlength(bs->file->bs),
+                .last_file_offset    = bdrv_getlength(bs->file->bs),
               };
 
     new_hdr.log_guid = header->log_guid;
@@ -940,7 +941,7 @@ static int vhdx_log_write(BlockDriverState *bs, BDRVVHDXState *s,
 
         if (i == 0 && leading_length) {
             /* partial sector at the front of the buffer */
-            ret = bdrv_pread(bs->file, file_offset, merged_sector,
+            ret = bdrv_pread(bs->file->bs, file_offset, merged_sector,
                              VHDX_LOG_SECTOR_SIZE);
             if (ret < 0) {
                 goto exit;
@@ -950,7 +951,7 @@ static int vhdx_log_write(BlockDriverState *bs, BDRVVHDXState *s,
             sector_write = merged_sector;
         } else if (i == sectors - 1 && trailing_length) {
             /* partial sector at the end of the buffer */
-            ret = bdrv_pread(bs->file,
+            ret = bdrv_pread(bs->file->bs,
                             file_offset,
                             merged_sector + trailing_length,
                             VHDX_LOG_SECTOR_SIZE - trailing_length);
diff --git a/block/vhdx.c b/block/vhdx.c
index 0776de717..2fe9a5e0c 100644
--- a/block/vhdx.c
+++ b/block/vhdx.c
@@ -375,7 +375,7 @@ static int vhdx_update_header(BlockDriverState *bs, BDRVVHDXState *s,
         inactive_header->log_guid = *log_guid;
     }
 
-    ret = vhdx_write_header(bs->file, inactive_header, header_offset, true);
+    ret = vhdx_write_header(bs->file->bs, inactive_header, header_offset, true);
     if (ret < 0) {
         goto exit;
     }
@@ -427,7 +427,8 @@ static void vhdx_parse_header(BlockDriverState *bs, BDRVVHDXState *s,
     /* We have to read the whole VHDX_HEADER_SIZE instead of
      * sizeof(VHDXHeader), because the checksum is over the whole
      * region */
-    ret = bdrv_pread(bs->file, VHDX_HEADER1_OFFSET, buffer, VHDX_HEADER_SIZE);
+    ret = bdrv_pread(bs->file->bs, VHDX_HEADER1_OFFSET, buffer,
+                     VHDX_HEADER_SIZE);
     if (ret < 0) {
         goto fail;
     }
@@ -443,7 +444,8 @@ static void vhdx_parse_header(BlockDriverState *bs, BDRVVHDXState *s,
         }
     }
 
-    ret = bdrv_pread(bs->file, VHDX_HEADER2_OFFSET, buffer, VHDX_HEADER_SIZE);
+    ret = bdrv_pread(bs->file->bs, VHDX_HEADER2_OFFSET, buffer,
+                     VHDX_HEADER_SIZE);
     if (ret < 0) {
         goto fail;
     }
@@ -516,7 +518,7 @@ static int vhdx_open_region_tables(BlockDriverState *bs, BDRVVHDXState *s)
      * whole block */
     buffer = qemu_blockalign(bs, VHDX_HEADER_BLOCK_SIZE);
 
-    ret = bdrv_pread(bs->file, VHDX_REGION_TABLE_OFFSET, buffer,
+    ret = bdrv_pread(bs->file->bs, VHDX_REGION_TABLE_OFFSET, buffer,
                      VHDX_HEADER_BLOCK_SIZE);
     if (ret < 0) {
         goto fail;
@@ -629,7 +631,7 @@ static int vhdx_parse_metadata(BlockDriverState *bs, BDRVVHDXState *s)
 
     buffer = qemu_blockalign(bs, VHDX_METADATA_TABLE_MAX_SIZE);
 
-    ret = bdrv_pread(bs->file, s->metadata_rt.file_offset, buffer,
+    ret = bdrv_pread(bs->file->bs, s->metadata_rt.file_offset, buffer,
                      VHDX_METADATA_TABLE_MAX_SIZE);
     if (ret < 0) {
         goto exit;
@@ -732,7 +734,7 @@ static int vhdx_parse_metadata(BlockDriverState *bs, BDRVVHDXState *s)
         goto exit;
     }
 
-    ret = bdrv_pread(bs->file,
+    ret = bdrv_pread(bs->file->bs,
                      s->metadata_entries.file_parameters_entry.offset
                                          + s->metadata_rt.file_offset,
                      &s->params,
@@ -767,7 +769,7 @@ static int vhdx_parse_metadata(BlockDriverState *bs, BDRVVHDXState *s)
     /* determine virtual disk size, logical sector size,
      * and phys sector size */
 
-    ret = bdrv_pread(bs->file,
+    ret = bdrv_pread(bs->file->bs,
                      s->metadata_entries.virtual_disk_size_entry.offset
                                            + s->metadata_rt.file_offset,
                      &s->virtual_disk_size,
@@ -775,7 +777,7 @@ static int vhdx_parse_metadata(BlockDriverState *bs, BDRVVHDXState *s)
     if (ret < 0) {
         goto exit;
     }
-    ret = bdrv_pread(bs->file,
+    ret = bdrv_pread(bs->file->bs,
                      s->metadata_entries.logical_sector_size_entry.offset
                                              + s->metadata_rt.file_offset,
                      &s->logical_sector_size,
@@ -783,7 +785,7 @@ static int vhdx_parse_metadata(BlockDriverState *bs, BDRVVHDXState *s)
     if (ret < 0) {
         goto exit;
     }
-    ret = bdrv_pread(bs->file,
+    ret = bdrv_pread(bs->file->bs,
                      s->metadata_entries.phys_sector_size_entry.offset
                                           + s->metadata_rt.file_offset,
                      &s->physical_sector_size,
@@ -906,7 +908,7 @@ static int vhdx_open(BlockDriverState *bs, QDict *options, int flags,
     QLIST_INIT(&s->regions);
 
     /* validate the file signature */
-    ret = bdrv_pread(bs->file, 0, &signature, sizeof(uint64_t));
+    ret = bdrv_pread(bs->file->bs, 0, &signature, sizeof(uint64_t));
     if (ret < 0) {
         goto fail;
     }
@@ -959,13 +961,13 @@ static int vhdx_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     /* s->bat is freed in vhdx_close() */
-    s->bat = qemu_try_blockalign(bs->file, s->bat_rt.length);
+    s->bat = qemu_try_blockalign(bs->file->bs, s->bat_rt.length);
     if (s->bat == NULL) {
         ret = -ENOMEM;
         goto fail;
     }
 
-    ret = bdrv_pread(bs->file, s->bat_offset, s->bat, s->bat_rt.length);
+    ret = bdrv_pread(bs->file->bs, s->bat_offset, s->bat, s->bat_rt.length);
     if (ret < 0) {
         goto fail;
     }
@@ -1118,7 +1120,7 @@ static coroutine_fn int vhdx_co_readv(BlockDriverState *bs, int64_t sector_num,
                 break;
             case PAYLOAD_BLOCK_FULLY_PRESENT:
                 qemu_co_mutex_unlock(&s->lock);
-                ret = bdrv_co_readv(bs->file,
+                ret = bdrv_co_readv(bs->file->bs,
                                     sinfo.file_offset >> BDRV_SECTOR_BITS,
                                     sinfo.sectors_avail, &hd_qiov);
                 qemu_co_mutex_lock(&s->lock);
@@ -1156,12 +1158,12 @@ exit:
 static int vhdx_allocate_block(BlockDriverState *bs, BDRVVHDXState *s,
                                     uint64_t *new_offset)
 {
-    *new_offset = bdrv_getlength(bs->file);
+    *new_offset = bdrv_getlength(bs->file->bs);
 
     /* per the spec, the address for a block is in units of 1MB */
     *new_offset = ROUND_UP(*new_offset, 1024 * 1024);
 
-    return bdrv_truncate(bs->file, *new_offset + s->block_size);
+    return bdrv_truncate(bs->file->bs, *new_offset + s->block_size);
 }
 
 /*
@@ -1260,7 +1262,7 @@ static coroutine_fn int vhdx_co_writev(BlockDriverState *bs, int64_t sector_num,
                 /* Queue another write of zero buffers if the underlying file
                  * does not zero-fill on file extension */
 
-                if (bdrv_has_zero_init(bs->file) == 0) {
+                if (bdrv_has_zero_init(bs->file->bs) == 0) {
                     use_zero_buffers = true;
 
                     /* zero fill the front, if any */
@@ -1327,7 +1329,7 @@ static coroutine_fn int vhdx_co_writev(BlockDriverState *bs, int64_t sector_num,
                 }
                 /* block exists, so we can just overwrite it */
                 qemu_co_mutex_unlock(&s->lock);
-                ret = bdrv_co_writev(bs->file,
+                ret = bdrv_co_writev(bs->file->bs,
                                     sinfo.file_offset >> BDRV_SECTOR_BITS,
                                     sectors_to_write, &hd_qiov);
                 qemu_co_mutex_lock(&s->lock);
@@ -1454,7 +1456,7 @@ static int vhdx_create_new_metadata(BlockDriverState *bs,
     uint32_t offset = 0;
     void *buffer = NULL;
     void *entry_buffer;
-    VHDXMetadataTableHeader *md_table;;
+    VHDXMetadataTableHeader *md_table;
     VHDXMetadataTableEntry  *md_table_entry;
 
     /* Metadata entries */
@@ -1842,7 +1844,7 @@ static int vhdx_create(const char *filename, QemuOpts *opts, Error **errp)
 
     bs = NULL;
     ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
-                    NULL, &local_err);
+                    &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto exit;
diff --git a/block/vmdk.c b/block/vmdk.c
index fbaab67c8..e46271a80 100644
--- a/block/vmdk.c
+++ b/block/vmdk.c
@@ -87,7 +87,7 @@ typedef struct {
 #define L2_CACHE_SIZE 16
 
 typedef struct VmdkExtent {
-    BlockDriverState *file;
+    BdrvChild *file;
     bool flat;
     bool compressed;
     bool has_marker;
@@ -222,7 +222,7 @@ static void vmdk_free_extents(BlockDriverState *bs)
         g_free(e->l1_backup_table);
         g_free(e->type);
         if (e->file != bs->file) {
-            bdrv_unref(e->file);
+            bdrv_unref_child(bs, e->file);
         }
     }
     g_free(s->extents);
@@ -248,7 +248,7 @@ static uint32_t vmdk_read_cid(BlockDriverState *bs, int parent)
     BDRVVmdkState *s = bs->opaque;
     int ret;
 
-    ret = bdrv_pread(bs->file, s->desc_offset, desc, DESC_SIZE);
+    ret = bdrv_pread(bs->file->bs, s->desc_offset, desc, DESC_SIZE);
     if (ret < 0) {
         return 0;
     }
@@ -278,7 +278,7 @@ static int vmdk_write_cid(BlockDriverState *bs, uint32_t cid)
     BDRVVmdkState *s = bs->opaque;
     int ret;
 
-    ret = bdrv_pread(bs->file, s->desc_offset, desc, DESC_SIZE);
+    ret = bdrv_pread(bs->file->bs, s->desc_offset, desc, DESC_SIZE);
     if (ret < 0) {
         return ret;
     }
@@ -297,7 +297,7 @@ static int vmdk_write_cid(BlockDriverState *bs, uint32_t cid)
         pstrcat(desc, sizeof(desc), tmp_desc);
     }
 
-    ret = bdrv_pwrite_sync(bs->file, s->desc_offset, desc, DESC_SIZE);
+    ret = bdrv_pwrite_sync(bs->file->bs, s->desc_offset, desc, DESC_SIZE);
     if (ret < 0) {
         return ret;
     }
@@ -308,10 +308,11 @@ static int vmdk_write_cid(BlockDriverState *bs, uint32_t cid)
 static int vmdk_is_cid_valid(BlockDriverState *bs)
 {
     BDRVVmdkState *s = bs->opaque;
-    BlockDriverState *p_bs = bs->backing_hd;
     uint32_t cur_pcid;
 
-    if (!s->cid_checked && p_bs) {
+    if (!s->cid_checked && bs->backing) {
+        BlockDriverState *p_bs = bs->backing->bs;
+
         cur_pcid = vmdk_read_cid(p_bs, 0);
         if (s->parent_cid != cur_pcid) {
             /* CID not valid */
@@ -340,7 +341,7 @@ static int vmdk_parent_open(BlockDriverState *bs)
     int ret;
 
     desc[DESC_SIZE] = '\0';
-    ret = bdrv_pread(bs->file, s->desc_offset, desc, DESC_SIZE);
+    ret = bdrv_pread(bs->file->bs, s->desc_offset, desc, DESC_SIZE);
     if (ret < 0) {
         return ret;
     }
@@ -367,7 +368,7 @@ static int vmdk_parent_open(BlockDriverState *bs)
 /* Create and append extent to the extent array. Return the added VmdkExtent
  * address. return NULL if allocation failed. */
 static int vmdk_add_extent(BlockDriverState *bs,
-                           BlockDriverState *file, bool flat, int64_t sectors,
+                           BdrvChild *file, bool flat, int64_t sectors,
                            int64_t l1_offset, int64_t l1_backup_offset,
                            uint32_t l1_size,
                            int l2_size, uint64_t cluster_sectors,
@@ -392,7 +393,7 @@ static int vmdk_add_extent(BlockDriverState *bs,
         return -EFBIG;
     }
 
-    nb_sectors = bdrv_nb_sectors(file);
+    nb_sectors = bdrv_nb_sectors(file->bs);
     if (nb_sectors < 0) {
         return nb_sectors;
     }
@@ -439,14 +440,14 @@ static int vmdk_init_tables(BlockDriverState *bs, VmdkExtent *extent,
         return -ENOMEM;
     }
 
-    ret = bdrv_pread(extent->file,
+    ret = bdrv_pread(extent->file->bs,
                      extent->l1_table_offset,
                      extent->l1_table,
                      l1_size);
     if (ret < 0) {
         error_setg_errno(errp, -ret,
                          "Could not read l1 table from extent '%s'",
-                         extent->file->filename);
+                         extent->file->bs->filename);
         goto fail_l1;
     }
     for (i = 0; i < extent->l1_size; i++) {
@@ -459,14 +460,14 @@ static int vmdk_init_tables(BlockDriverState *bs, VmdkExtent *extent,
             ret = -ENOMEM;
             goto fail_l1;
         }
-        ret = bdrv_pread(extent->file,
+        ret = bdrv_pread(extent->file->bs,
                          extent->l1_backup_table_offset,
                          extent->l1_backup_table,
                          l1_size);
         if (ret < 0) {
             error_setg_errno(errp, -ret,
                              "Could not read l1 backup table from extent '%s'",
-                             extent->file->filename);
+                             extent->file->bs->filename);
             goto fail_l1b;
         }
         for (i = 0; i < extent->l1_size; i++) {
@@ -485,7 +486,7 @@ static int vmdk_init_tables(BlockDriverState *bs, VmdkExtent *extent,
 }
 
 static int vmdk_open_vmfs_sparse(BlockDriverState *bs,
-                                 BlockDriverState *file,
+                                 BdrvChild *file,
                                  int flags, Error **errp)
 {
     int ret;
@@ -493,11 +494,11 @@ static int vmdk_open_vmfs_sparse(BlockDriverState *bs,
     VMDK3Header header;
     VmdkExtent *extent;
 
-    ret = bdrv_pread(file, sizeof(magic), &header, sizeof(header));
+    ret = bdrv_pread(file->bs, sizeof(magic), &header, sizeof(header));
     if (ret < 0) {
         error_setg_errno(errp, -ret,
                          "Could not read header from file '%s'",
-                         file->filename);
+                         file->bs->filename);
         return ret;
     }
     ret = vmdk_add_extent(bs, file, false,
@@ -559,7 +560,7 @@ static char *vmdk_read_desc(BlockDriverState *file, uint64_t desc_offset,
 }
 
 static int vmdk_open_vmdk4(BlockDriverState *bs,
-                           BlockDriverState *file,
+                           BdrvChild *file,
                            int flags, QDict *options, Error **errp)
 {
     int ret;
@@ -569,18 +570,19 @@ static int vmdk_open_vmdk4(BlockDriverState *bs,
     VmdkExtent *extent;
     BDRVVmdkState *s = bs->opaque;
     int64_t l1_backup_offset = 0;
+    bool compressed;
 
-    ret = bdrv_pread(file, sizeof(magic), &header, sizeof(header));
+    ret = bdrv_pread(file->bs, sizeof(magic), &header, sizeof(header));
     if (ret < 0) {
         error_setg_errno(errp, -ret,
                          "Could not read header from file '%s'",
-                         file->filename);
+                         file->bs->filename);
         return -EINVAL;
     }
     if (header.capacity == 0) {
         uint64_t desc_offset = le64_to_cpu(header.desc_offset);
         if (desc_offset) {
-            char *buf = vmdk_read_desc(file, desc_offset << 9, errp);
+            char *buf = vmdk_read_desc(file->bs, desc_offset << 9, errp);
             if (!buf) {
                 return -EINVAL;
             }
@@ -620,8 +622,8 @@ static int vmdk_open_vmdk4(BlockDriverState *bs,
             } QEMU_PACKED eos_marker;
         } QEMU_PACKED footer;
 
-        ret = bdrv_pread(file,
-            bs->file->total_sectors * 512 - 1536,
+        ret = bdrv_pread(file->bs,
+            bs->file->bs->total_sectors * 512 - 1536,
             &footer, sizeof(footer));
         if (ret < 0) {
             error_setg_errno(errp, -ret, "Failed to read footer");
@@ -643,6 +645,8 @@ static int vmdk_open_vmdk4(BlockDriverState *bs,
         header = footer.header;
     }
 
+    compressed =
+        le16_to_cpu(header.compressAlgorithm) == VMDK4_COMPRESSION_DEFLATE;
     if (le32_to_cpu(header.version) > 3) {
         char buf[64];
         snprintf(buf, sizeof(buf), "VMDK version %" PRId32,
@@ -650,7 +654,8 @@ static int vmdk_open_vmdk4(BlockDriverState *bs,
         error_setg(errp, QERR_UNKNOWN_BLOCK_FORMAT_FEATURE,
                    bdrv_get_device_or_node_name(bs), "vmdk", buf);
         return -ENOTSUP;
-    } else if (le32_to_cpu(header.version) == 3 && (flags & BDRV_O_RDWR)) {
+    } else if (le32_to_cpu(header.version) == 3 && (flags & BDRV_O_RDWR) &&
+               !compressed) {
         /* VMware KB 2064959 explains that version 3 added support for
          * persistent changed block tracking (CBT), and backup software can
          * read it as version=1 if it doesn't care about the changed area
@@ -675,7 +680,7 @@ static int vmdk_open_vmdk4(BlockDriverState *bs,
     if (le32_to_cpu(header.flags) & VMDK4_FLAG_RGD) {
         l1_backup_offset = le64_to_cpu(header.rgd_offset) << 9;
     }
-    if (bdrv_nb_sectors(file) < le64_to_cpu(header.grain_offset)) {
+    if (bdrv_nb_sectors(file->bs) < le64_to_cpu(header.grain_offset)) {
         error_setg(errp, "File truncated, expecting at least %" PRId64 " bytes",
                    (int64_t)(le64_to_cpu(header.grain_offset)
                              * BDRV_SECTOR_SIZE));
@@ -739,8 +744,7 @@ static int vmdk_parse_description(const char *desc, const char *opt_name,
 }
 
 /* Open an extent file and append to bs array */
-static int vmdk_open_sparse(BlockDriverState *bs,
-                            BlockDriverState *file, int flags,
+static int vmdk_open_sparse(BlockDriverState *bs, BdrvChild *file, int flags,
                             char *buf, QDict *options, Error **errp)
 {
     uint32_t magic;
@@ -773,10 +777,11 @@ static int vmdk_parse_extents(const char *desc, BlockDriverState *bs,
     int64_t sectors = 0;
     int64_t flat_offset;
     char *extent_path;
-    BlockDriverState *extent_file;
+    BdrvChild *extent_file;
     BDRVVmdkState *s = bs->opaque;
     VmdkExtent *extent;
     char extent_opt_prefix[32];
+    Error *local_err = NULL;
 
     while (*p) {
         /* parse extent line in one of below formats:
@@ -819,22 +824,22 @@ static int vmdk_parse_extents(const char *desc, BlockDriverState *bs,
             !desc_file_path[0])
         {
             error_setg(errp, "Cannot use relative extent paths with VMDK "
-                       "descriptor file '%s'", bs->file->filename);
+                       "descriptor file '%s'", bs->file->bs->filename);
             return -EINVAL;
         }
 
         extent_path = g_malloc0(PATH_MAX);
         path_combine(extent_path, PATH_MAX, desc_file_path, fname);
-        extent_file = NULL;
 
         ret = snprintf(extent_opt_prefix, 32, "extents.%d", s->num_extents);
         assert(ret < 32);
 
-        ret = bdrv_open_image(&extent_file, extent_path, options,
-                              extent_opt_prefix, bs, &child_file, false, errp);
+        extent_file = bdrv_open_child(extent_path, options, extent_opt_prefix,
+                                      bs, &child_file, false, &local_err);
         g_free(extent_path);
-        if (ret) {
-            return ret;
+        if (local_err) {
+            error_propagate(errp, local_err);
+            return -EINVAL;
         }
 
         /* save to extents array */
@@ -844,13 +849,13 @@ static int vmdk_parse_extents(const char *desc, BlockDriverState *bs,
             ret = vmdk_add_extent(bs, extent_file, true, sectors,
                             0, 0, 0, 0, 0, &extent, errp);
             if (ret < 0) {
-                bdrv_unref(extent_file);
+                bdrv_unref_child(bs, extent_file);
                 return ret;
             }
             extent->flat_start_offset = flat_offset << 9;
         } else if (!strcmp(type, "SPARSE") || !strcmp(type, "VMFSSPARSE")) {
             /* SPARSE extent and VMFSSPARSE extent are both "COWD" sparse file*/
-            char *buf = vmdk_read_desc(extent_file, 0, errp);
+            char *buf = vmdk_read_desc(extent_file->bs, 0, errp);
             if (!buf) {
                 ret = -EINVAL;
             } else {
@@ -859,13 +864,13 @@ static int vmdk_parse_extents(const char *desc, BlockDriverState *bs,
             }
             g_free(buf);
             if (ret) {
-                bdrv_unref(extent_file);
+                bdrv_unref_child(bs, extent_file);
                 return ret;
             }
             extent = &s->extents[s->num_extents - 1];
         } else {
             error_setg(errp, "Unsupported extent type '%s'", type);
-            bdrv_unref(extent_file);
+            bdrv_unref_child(bs, extent_file);
             return -ENOTSUP;
         }
         extent->type = g_strdup(type);
@@ -905,7 +910,8 @@ static int vmdk_open_desc_file(BlockDriverState *bs, int flags, char *buf,
     }
     s->create_type = g_strdup(ct);
     s->desc_offset = 0;
-    ret = vmdk_parse_extents(buf, bs, bs->file->exact_filename, options, errp);
+    ret = vmdk_parse_extents(buf, bs, bs->file->bs->exact_filename, options,
+                             errp);
 exit:
     return ret;
 }
@@ -918,7 +924,7 @@ static int vmdk_open(BlockDriverState *bs, QDict *options, int flags,
     BDRVVmdkState *s = bs->opaque;
     uint32_t magic;
 
-    buf = vmdk_read_desc(bs->file, 0, errp);
+    buf = vmdk_read_desc(bs->file->bs, 0, errp);
     if (!buf) {
         return -EINVAL;
     }
@@ -927,7 +933,8 @@ static int vmdk_open(BlockDriverState *bs, QDict *options, int flags,
     switch (magic) {
         case VMDK3_MAGIC:
         case VMDK4_MAGIC:
-            ret = vmdk_open_sparse(bs, bs->file, flags, buf, options, errp);
+            ret = vmdk_open_sparse(bs, bs->file, flags, buf, options,
+                                   errp);
             s->desc_offset = 0x200;
             break;
         default:
@@ -1004,7 +1011,7 @@ static int get_whole_cluster(BlockDriverState *bs,
     cluster_bytes = extent->cluster_sectors << BDRV_SECTOR_BITS;
     whole_grain = qemu_blockalign(bs, cluster_bytes);
 
-    if (!bs->backing_hd) {
+    if (!bs->backing) {
         memset(whole_grain, 0,  skip_start_sector << BDRV_SECTOR_BITS);
         memset(whole_grain + (skip_end_sector << BDRV_SECTOR_BITS), 0,
                cluster_bytes - (skip_end_sector << BDRV_SECTOR_BITS));
@@ -1013,22 +1020,22 @@ static int get_whole_cluster(BlockDriverState *bs,
     assert(skip_end_sector <= extent->cluster_sectors);
     /* we will be here if it's first write on non-exist grain(cluster).
      * try to read from parent image, if exist */
-    if (bs->backing_hd && !vmdk_is_cid_valid(bs)) {
+    if (bs->backing && !vmdk_is_cid_valid(bs)) {
         ret = VMDK_ERROR;
         goto exit;
     }
 
     /* Read backing data before skip range */
     if (skip_start_sector > 0) {
-        if (bs->backing_hd) {
-            ret = bdrv_read(bs->backing_hd, sector_num,
+        if (bs->backing) {
+            ret = bdrv_read(bs->backing->bs, sector_num,
                             whole_grain, skip_start_sector);
             if (ret < 0) {
                 ret = VMDK_ERROR;
                 goto exit;
             }
         }
-        ret = bdrv_write(extent->file, cluster_sector_num, whole_grain,
+        ret = bdrv_write(extent->file->bs, cluster_sector_num, whole_grain,
                          skip_start_sector);
         if (ret < 0) {
             ret = VMDK_ERROR;
@@ -1037,8 +1044,8 @@ static int get_whole_cluster(BlockDriverState *bs,
     }
     /* Read backing data after skip range */
     if (skip_end_sector < extent->cluster_sectors) {
-        if (bs->backing_hd) {
-            ret = bdrv_read(bs->backing_hd, sector_num + skip_end_sector,
+        if (bs->backing) {
+            ret = bdrv_read(bs->backing->bs, sector_num + skip_end_sector,
                             whole_grain + (skip_end_sector << BDRV_SECTOR_BITS),
                             extent->cluster_sectors - skip_end_sector);
             if (ret < 0) {
@@ -1046,7 +1053,7 @@ static int get_whole_cluster(BlockDriverState *bs,
                 goto exit;
             }
         }
-        ret = bdrv_write(extent->file, cluster_sector_num + skip_end_sector,
+        ret = bdrv_write(extent->file->bs, cluster_sector_num + skip_end_sector,
                          whole_grain + (skip_end_sector << BDRV_SECTOR_BITS),
                          extent->cluster_sectors - skip_end_sector);
         if (ret < 0) {
@@ -1066,7 +1073,7 @@ static int vmdk_L2update(VmdkExtent *extent, VmdkMetaData *m_data,
     offset = cpu_to_le32(offset);
     /* update L2 table */
     if (bdrv_pwrite_sync(
-                extent->file,
+                extent->file->bs,
                 ((int64_t)m_data->l2_offset * 512)
                     + (m_data->l2_index * sizeof(offset)),
                 &offset, sizeof(offset)) < 0) {
@@ -1076,7 +1083,7 @@ static int vmdk_L2update(VmdkExtent *extent, VmdkMetaData *m_data,
     if (extent->l1_backup_table_offset != 0) {
         m_data->l2_offset = extent->l1_backup_table[m_data->l1_index];
         if (bdrv_pwrite_sync(
-                    extent->file,
+                    extent->file->bs,
                     ((int64_t)m_data->l2_offset * 512)
                         + (m_data->l2_index * sizeof(offset)),
                     &offset, sizeof(offset)) < 0) {
@@ -1166,7 +1173,7 @@ static int get_cluster_offset(BlockDriverState *bs,
     }
     l2_table = extent->l2_cache + (min_index * extent->l2_size);
     if (bdrv_pread(
-                extent->file,
+                extent->file->bs,
                 (int64_t)l2_offset * 512,
                 l2_table,
                 extent->l2_size * sizeof(uint32_t)
@@ -1320,12 +1327,16 @@ static int vmdk_write_extent(VmdkExtent *extent, int64_t cluster_offset,
         write_len = buf_len + sizeof(VmdkGrainMarker);
     }
     write_offset = cluster_offset + offset_in_cluster,
-    ret = bdrv_pwrite(extent->file, write_offset, write_buf, write_len);
+    ret = bdrv_pwrite(extent->file->bs, write_offset, write_buf, write_len);
 
     write_end_sector = DIV_ROUND_UP(write_offset + write_len, BDRV_SECTOR_SIZE);
 
-    extent->next_cluster_sector = MAX(extent->next_cluster_sector,
-                                      write_end_sector);
+    if (extent->compressed) {
+        extent->next_cluster_sector = write_end_sector;
+    } else {
+        extent->next_cluster_sector = MAX(extent->next_cluster_sector,
+                                          write_end_sector);
+    }
 
     if (ret != write_len) {
         ret = ret < 0 ? ret : -EIO;
@@ -1351,7 +1362,7 @@ static int vmdk_read_extent(VmdkExtent *extent, int64_t cluster_offset,
 
 
     if (!extent->compressed) {
-        ret = bdrv_pread(extent->file,
+        ret = bdrv_pread(extent->file->bs,
                           cluster_offset + offset_in_cluster,
                           buf, nb_sectors * 512);
         if (ret == nb_sectors * 512) {
@@ -1365,7 +1376,7 @@ static int vmdk_read_extent(VmdkExtent *extent, int64_t cluster_offset,
     buf_bytes = cluster_bytes * 2;
     cluster_buf = g_malloc(buf_bytes);
     uncomp_buf = g_malloc(cluster_bytes);
-    ret = bdrv_pread(extent->file,
+    ret = bdrv_pread(extent->file->bs,
                 cluster_offset,
                 cluster_buf, buf_bytes);
     if (ret < 0) {
@@ -1427,11 +1438,11 @@ static int vmdk_read(BlockDriverState *bs, int64_t sector_num,
         }
         if (ret != VMDK_OK) {
             /* if not allocated, try to read from parent image, if exist */
-            if (bs->backing_hd && ret != VMDK_ZEROED) {
+            if (bs->backing && ret != VMDK_ZEROED) {
                 if (!vmdk_is_cid_valid(bs)) {
                     return -EINVAL;
                 }
-                ret = bdrv_read(bs->backing_hd, sector_num, buf, n);
+                ret = bdrv_read(bs->backing->bs, sector_num, buf, n);
                 if (ret < 0) {
                     return ret;
                 }
@@ -1632,7 +1643,7 @@ static int vmdk_create_extent(const char *filename, int64_t filesize,
 
     assert(bs == NULL);
     ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
-                    NULL, &local_err);
+                    &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto exit;
@@ -1647,7 +1658,13 @@ static int vmdk_create_extent(const char *filename, int64_t filesize,
     }
     magic = cpu_to_be32(VMDK4_MAGIC);
     memset(&header, 0, sizeof(header));
-    header.version = zeroed_grain ? 2 : 1;
+    if (compress) {
+        header.version = 3;
+    } else if (zeroed_grain) {
+        header.version = 2;
+    } else {
+        header.version = 1;
+    }
     header.flags = VMDK4_FLAG_RGD | VMDK4_FLAG_NL_DETECT
                    | (compress ? VMDK4_FLAG_COMPRESS | VMDK4_FLAG_MARKER : 0)
                    | (zeroed_grain ? VMDK4_FLAG_ZERO_GRAIN : 0);
@@ -1905,8 +1922,7 @@ static int vmdk_create(const char *filename, QemuOpts *opts, Error **errp)
             ret = -ENOENT;
             goto exit;
         }
-        ret = bdrv_open(&bs, full_backing, NULL, NULL, BDRV_O_NO_BACKING, NULL,
-                        errp);
+        ret = bdrv_open(&bs, full_backing, NULL, NULL, BDRV_O_NO_BACKING, errp);
         g_free(full_backing);
         if (ret != 0) {
             goto exit;
@@ -1977,7 +1993,7 @@ static int vmdk_create(const char *filename, QemuOpts *opts, Error **errp)
     }
     assert(new_bs == NULL);
     ret = bdrv_open(&new_bs, filename, NULL, NULL,
-                    BDRV_O_RDWR | BDRV_O_PROTOCOL, NULL, &local_err);
+                    BDRV_O_RDWR | BDRV_O_PROTOCOL, &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto exit;
@@ -2032,7 +2048,7 @@ static coroutine_fn int vmdk_co_flush(BlockDriverState *bs)
     int ret = 0;
 
     for (i = 0; i < s->num_extents; i++) {
-        err = bdrv_co_flush(s->extents[i].file);
+        err = bdrv_co_flush(s->extents[i].file->bs);
         if (err < 0) {
             ret = err;
         }
@@ -2047,7 +2063,7 @@ static int64_t vmdk_get_allocated_file_size(BlockDriverState *bs)
     int64_t r;
     BDRVVmdkState *s = bs->opaque;
 
-    ret = bdrv_get_allocated_file_size(bs->file);
+    ret = bdrv_get_allocated_file_size(bs->file->bs);
     if (ret < 0) {
         return ret;
     }
@@ -2055,7 +2071,7 @@ static int64_t vmdk_get_allocated_file_size(BlockDriverState *bs)
         if (s->extents[i].file == bs->file) {
             continue;
         }
-        r = bdrv_get_allocated_file_size(s->extents[i].file);
+        r = bdrv_get_allocated_file_size(s->extents[i].file->bs);
         if (r < 0) {
             return r;
         }
@@ -2073,7 +2089,7 @@ static int vmdk_has_zero_init(BlockDriverState *bs)
      * return 0. */
     for (i = 0; i < s->num_extents; i++) {
         if (s->extents[i].flat) {
-            if (!bdrv_has_zero_init(s->extents[i].file)) {
+            if (!bdrv_has_zero_init(s->extents[i].file->bs)) {
                 return 0;
             }
         }
@@ -2086,7 +2102,7 @@ static ImageInfo *vmdk_get_extent_info(VmdkExtent *extent)
     ImageInfo *info = g_new0(ImageInfo, 1);
 
     *info = (ImageInfo){
-        .filename         = g_strdup(extent->file->filename),
+        .filename         = g_strdup(extent->file->bs->filename),
         .format           = g_strdup(extent->type),
         .virtual_size     = extent->sectors * BDRV_SECTOR_SIZE,
         .compressed       = extent->compressed,
@@ -2132,7 +2148,9 @@ static int vmdk_check(BlockDriverState *bs, BdrvCheckResult *result,
                     PRId64 "\n", sector_num);
             break;
         }
-        if (ret == VMDK_OK && cluster_offset >= bdrv_getlength(extent->file)) {
+        if (ret == VMDK_OK &&
+            cluster_offset >= bdrv_getlength(extent->file->bs))
+        {
             fprintf(stderr,
                     "ERROR: cluster offset for sector %"
                     PRId64 " points after EOF\n", sector_num);
@@ -2153,19 +2171,19 @@ static ImageInfoSpecific *vmdk_get_specific_info(BlockDriverState *bs)
     ImageInfoList **next;
 
     *spec_info = (ImageInfoSpecific){
-        .kind = IMAGE_INFO_SPECIFIC_KIND_VMDK,
+        .type = IMAGE_INFO_SPECIFIC_KIND_VMDK,
         {
             .vmdk = g_new0(ImageInfoSpecificVmdk, 1),
         },
     };
 
-    *spec_info->vmdk = (ImageInfoSpecificVmdk) {
+    *spec_info->u.vmdk = (ImageInfoSpecificVmdk) {
         .create_type = g_strdup(s->create_type),
         .cid = s->cid,
         .parent_cid = s->parent_cid,
     };
 
-    next = &spec_info->vmdk->extents;
+    next = &spec_info->u.vmdk->extents;
     for (i = 0; i < s->num_extents; i++) {
         *next = g_new0(ImageInfoList, 1);
         (*next)->value = vmdk_get_extent_info(&s->extents[i]);
@@ -2208,7 +2226,7 @@ static void vmdk_detach_aio_context(BlockDriverState *bs)
     int i;
 
     for (i = 0; i < s->num_extents; i++) {
-        bdrv_detach_aio_context(s->extents[i].file);
+        bdrv_detach_aio_context(s->extents[i].file->bs);
     }
 }
 
@@ -2219,7 +2237,7 @@ static void vmdk_attach_aio_context(BlockDriverState *bs,
     int i;
 
     for (i = 0; i < s->num_extents; i++) {
-        bdrv_attach_aio_context(s->extents[i].file, new_context);
+        bdrv_attach_aio_context(s->extents[i].file->bs, new_context);
     }
 }
 
diff --git a/block/vpc.c b/block/vpc.c
index 3e385d9fb..299d37309 100644
--- a/block/vpc.c
+++ b/block/vpc.c
@@ -172,14 +172,14 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
     int disk_type = VHD_DYNAMIC;
     int ret;
 
-    ret = bdrv_pread(bs->file, 0, s->footer_buf, HEADER_SIZE);
+    ret = bdrv_pread(bs->file->bs, 0, s->footer_buf, HEADER_SIZE);
     if (ret < 0) {
         goto fail;
     }
 
     footer = (VHDFooter *) s->footer_buf;
     if (strncmp(footer->creator, "conectix", 8)) {
-        int64_t offset = bdrv_getlength(bs->file);
+        int64_t offset = bdrv_getlength(bs->file->bs);
         if (offset < 0) {
             ret = offset;
             goto fail;
@@ -189,7 +189,7 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
         }
 
         /* If a fixed disk, the footer is found only at the end of the file */
-        ret = bdrv_pread(bs->file, offset-HEADER_SIZE, s->footer_buf,
+        ret = bdrv_pread(bs->file->bs, offset-HEADER_SIZE, s->footer_buf,
                          HEADER_SIZE);
         if (ret < 0) {
             goto fail;
@@ -232,7 +232,7 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
     }
 
     if (disk_type == VHD_DYNAMIC) {
-        ret = bdrv_pread(bs->file, be64_to_cpu(footer->data_offset), buf,
+        ret = bdrv_pread(bs->file->bs, be64_to_cpu(footer->data_offset), buf,
                          HEADER_SIZE);
         if (ret < 0) {
             goto fail;
@@ -280,7 +280,7 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
 
         pagetable_size = (uint64_t) s->max_table_entries * 4;
 
-        s->pagetable = qemu_try_blockalign(bs->file, pagetable_size);
+        s->pagetable = qemu_try_blockalign(bs->file->bs, pagetable_size);
         if (s->pagetable == NULL) {
             ret = -ENOMEM;
             goto fail;
@@ -288,7 +288,8 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
 
         s->bat_offset = be64_to_cpu(dyndisk_header->table_offset);
 
-        ret = bdrv_pread(bs->file, s->bat_offset, s->pagetable, pagetable_size);
+        ret = bdrv_pread(bs->file->bs, s->bat_offset, s->pagetable,
+                         pagetable_size);
         if (ret < 0) {
             goto fail;
         }
@@ -308,7 +309,7 @@ static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
             }
         }
 
-        if (s->free_data_block_offset > bdrv_getlength(bs->file)) {
+        if (s->free_data_block_offset > bdrv_getlength(bs->file->bs)) {
             error_setg(errp, "block-vpc: free_data_block_offset points after "
                              "the end of file. The image has been truncated.");
             ret = -EINVAL;
@@ -383,7 +384,7 @@ static inline int64_t get_sector_offset(BlockDriverState *bs,
 
         s->last_bitmap_offset = bitmap_offset;
         memset(bitmap, 0xff, s->bitmap_size);
-        bdrv_pwrite_sync(bs->file, bitmap_offset, bitmap, s->bitmap_size);
+        bdrv_pwrite_sync(bs->file->bs, bitmap_offset, bitmap, s->bitmap_size);
     }
 
     return block_offset;
@@ -401,7 +402,7 @@ static int rewrite_footer(BlockDriverState* bs)
     BDRVVPCState *s = bs->opaque;
     int64_t offset = s->free_data_block_offset;
 
-    ret = bdrv_pwrite_sync(bs->file, offset, s->footer_buf, HEADER_SIZE);
+    ret = bdrv_pwrite_sync(bs->file->bs, offset, s->footer_buf, HEADER_SIZE);
     if (ret < 0)
         return ret;
 
@@ -436,7 +437,7 @@ static int64_t alloc_block(BlockDriverState* bs, int64_t sector_num)
 
     // Initialize the block's bitmap
     memset(bitmap, 0xff, s->bitmap_size);
-    ret = bdrv_pwrite_sync(bs->file, s->free_data_block_offset, bitmap,
+    ret = bdrv_pwrite_sync(bs->file->bs, s->free_data_block_offset, bitmap,
         s->bitmap_size);
     if (ret < 0) {
         return ret;
@@ -451,7 +452,7 @@ static int64_t alloc_block(BlockDriverState* bs, int64_t sector_num)
     // Write BAT entry to disk
     bat_offset = s->bat_offset + (4 * index);
     bat_value = cpu_to_be32(s->pagetable[index]);
-    ret = bdrv_pwrite_sync(bs->file, bat_offset, &bat_value, 4);
+    ret = bdrv_pwrite_sync(bs->file->bs, bat_offset, &bat_value, 4);
     if (ret < 0)
         goto fail;
 
@@ -485,7 +486,7 @@ static int vpc_read(BlockDriverState *bs, int64_t sector_num,
     VHDFooter *footer = (VHDFooter *) s->footer_buf;
 
     if (be32_to_cpu(footer->type) == VHD_FIXED) {
-        return bdrv_read(bs->file, sector_num, buf, nb_sectors);
+        return bdrv_read(bs->file->bs, sector_num, buf, nb_sectors);
     }
     while (nb_sectors > 0) {
         offset = get_sector_offset(bs, sector_num, 0);
@@ -499,7 +500,7 @@ static int vpc_read(BlockDriverState *bs, int64_t sector_num,
         if (offset == -1) {
             memset(buf, 0, sectors * BDRV_SECTOR_SIZE);
         } else {
-            ret = bdrv_pread(bs->file, offset, buf,
+            ret = bdrv_pread(bs->file->bs, offset, buf,
                 sectors * BDRV_SECTOR_SIZE);
             if (ret != sectors * BDRV_SECTOR_SIZE) {
                 return -1;
@@ -534,7 +535,7 @@ static int vpc_write(BlockDriverState *bs, int64_t sector_num,
     VHDFooter *footer =  (VHDFooter *) s->footer_buf;
 
     if (be32_to_cpu(footer->type) == VHD_FIXED) {
-        return bdrv_write(bs->file, sector_num, buf, nb_sectors);
+        return bdrv_write(bs->file->bs, sector_num, buf, nb_sectors);
     }
     while (nb_sectors > 0) {
         offset = get_sector_offset(bs, sector_num, 1);
@@ -551,7 +552,8 @@ static int vpc_write(BlockDriverState *bs, int64_t sector_num,
                 return -1;
         }
 
-        ret = bdrv_pwrite(bs->file, offset, buf, sectors * BDRV_SECTOR_SIZE);
+        ret = bdrv_pwrite(bs->file->bs, offset, buf,
+                          sectors * BDRV_SECTOR_SIZE);
         if (ret != sectors * BDRV_SECTOR_SIZE) {
             return -1;
         }
@@ -794,7 +796,7 @@ static int vpc_create(const char *filename, QemuOpts *opts, Error **errp)
         goto out;
     }
     ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
-                    NULL, &local_err);
+                    &local_err);
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto out;
@@ -878,7 +880,7 @@ static int vpc_has_zero_init(BlockDriverState *bs)
     VHDFooter *footer =  (VHDFooter *) s->footer_buf;
 
     if (be32_to_cpu(footer->type) == VHD_FIXED) {
-        return bdrv_has_zero_init(bs->file);
+        return bdrv_has_zero_init(bs->file->bs);
     } else {
         return 1;
     }
diff --git a/block/vvfat.c b/block/vvfat.c
index 206869712..b184eca6f 100644
--- a/block/vvfat.c
+++ b/block/vvfat.c
@@ -985,12 +985,6 @@ static BDRVVVFATState *vvv = NULL;
 static int enable_write_target(BDRVVVFATState *s, Error **errp);
 static int is_consistent(BDRVVVFATState *s);
 
-static void vvfat_rebind(BlockDriverState *bs)
-{
-    BDRVVVFATState *s = bs->opaque;
-    s->bs = bs;
-}
-
 static QemuOptsList runtime_opts = {
     .name = "vvfat",
     .head = QTAILQ_HEAD_INITIALIZER(runtime_opts.head),
@@ -2923,9 +2917,12 @@ static BlockDriver vvfat_write_target = {
 static int enable_write_target(BDRVVVFATState *s, Error **errp)
 {
     BlockDriver *bdrv_qcow = NULL;
+    BlockDriverState *backing;
     QemuOpts *opts = NULL;
     int ret;
     int size = sector2cluster(s, s->sector_count);
+    QDict *options;
+
     s->used_clusters = calloc(size, 1);
 
     array_init(&(s->commits), sizeof(commit_t));
@@ -2956,9 +2953,11 @@ static int enable_write_target(BDRVVVFATState *s, Error **errp)
     }
 
     s->qcow = NULL;
-    ret = bdrv_open(&s->qcow, s->qcow_filename, NULL, NULL,
+    options = qdict_new();
+    qdict_put(options, "driver", qstring_from_str("qcow"));
+    ret = bdrv_open(&s->qcow, s->qcow_filename, NULL, options,
                     BDRV_O_RDWR | BDRV_O_CACHE_WB | BDRV_O_NO_FLUSH,
-                    bdrv_qcow, errp);
+                    errp);
     if (ret < 0) {
         goto err;
     }
@@ -2967,10 +2966,13 @@ static int enable_write_target(BDRVVVFATState *s, Error **errp)
     unlink(s->qcow_filename);
 #endif
 
-    bdrv_set_backing_hd(s->bs, bdrv_new());
-    s->bs->backing_hd->drv = &vvfat_write_target;
-    s->bs->backing_hd->opaque = g_new(void *, 1);
-    *(void**)s->bs->backing_hd->opaque = s;
+    backing = bdrv_new();
+    bdrv_set_backing_hd(s->bs, backing);
+    bdrv_unref(backing);
+
+    s->bs->backing->bs->drv = &vvfat_write_target;
+    s->bs->backing->bs->opaque = g_new(void *, 1);
+    *(void**)s->bs->backing->bs->opaque = s;
 
     return 0;
 
@@ -3004,7 +3006,6 @@ static BlockDriver bdrv_vvfat = {
     .bdrv_parse_filename    = vvfat_parse_filename,
     .bdrv_file_open         = vvfat_open,
     .bdrv_close             = vvfat_close,
-    .bdrv_rebind            = vvfat_rebind,
 
     .bdrv_read              = vvfat_co_read,
     .bdrv_write             = vvfat_co_write,
diff --git a/block/win32-aio.c b/block/win32-aio.c
index 64e86827b..bbf2f01c1 100644
--- a/block/win32-aio.c
+++ b/block/win32-aio.c
@@ -174,7 +174,7 @@ int win32_aio_attach(QEMUWin32AIOState *aio, HANDLE hfile)
 void win32_aio_detach_aio_context(QEMUWin32AIOState *aio,
                                   AioContext *old_context)
 {
-    aio_set_event_notifier(old_context, &aio->e, NULL);
+    aio_set_event_notifier(old_context, &aio->e, false, NULL);
     aio->is_aio_context_attached = false;
 }
 
@@ -182,7 +182,8 @@ void win32_aio_attach_aio_context(QEMUWin32AIOState *aio,
                                   AioContext *new_context)
 {
     aio->is_aio_context_attached = true;
-    aio_set_event_notifier(new_context, &aio->e, win32_aio_completion_cb);
+    aio_set_event_notifier(new_context, &aio->e, false,
+                           win32_aio_completion_cb);
 }
 
 QEMUWin32AIOState *win32_aio_init(void)
diff --git a/block/write-threshold.c b/block/write-threshold.c
index a53c1f5e6..0fe38917c 100644
--- a/block/write-threshold.c
+++ b/block/write-threshold.c
@@ -11,7 +11,7 @@
  */
 
 #include "block/block_int.h"
-#include "block/coroutine.h"
+#include "qemu/coroutine.h"
 #include "block/write-threshold.h"
 #include "qemu/notify.h"
 #include "qapi-event.h"