Merge remote-tracking branch 'kwolf/tags/for-anthony' into staging

Block patches

# gpg: Signature made Fri 24 Jan 2014 08:40:53 AM PST using RSA key ID C88F2FD6
# gpg: Can't check signature: public key not found

* kwolf/tags/for-anthony: (93 commits)
  block: Switch bdrv_io_limits_intercept() to byte granularity
  qemu-iotests: Test pwritev RMW logic
  qemu-io: New command 'sleep'
  blkdebug: Make required alignment configurable
  iscsi: Set bs->request_alignment
  block: Make bdrv_pwrite() a bdrv_prwv_co() wrapper
  block: Make bdrv_pread() a bdrv_prwv_co() wrapper
  block: Change coroutine wrapper to byte granularity
  block: Assert serialisation assumptions in pwritev
  block: Align requests in bdrv_co_do_pwritev()
  block: Allow wait_serialising_requests() at any point
  block: Make overlap range for serialisation dynamic
  block: Generalise and optimise COR serialisation
  block: Make zero-after-EOF work with larger alignment
  block: Allow waiting for overlapping requests between begin/end
  block: Switch BdrvTrackedRequest to byte granularity
  block: Introduce bdrv_co_do_pwritev()
  block: write: Handle COR dependency after I/O throttling
  block: Introduce bdrv_aligned_pwritev()
  block: Introduce bdrv_co_do_preadv()
  ...

Message-id: 1390584136-24703-1-git-send-email-kwolf@redhat.com
Signed-off-by: Anthony Liguori <aliguori@amazon.com>

1 files changed, 778 insertions, 246 deletions

diff --git a/block.c b/block.c
index 64e7d22..cb21a5f 100644
--- a/block.c
+++ b/block.c
@@ -32,6 +32,7 @@
 #include "sysemu/sysemu.h"
 #include "qemu/notify.h"
 #include "block/coroutine.h"
+#include "block/qapi.h"
 #include "qmp-commands.h"
 #include "qemu/timer.h"
 
@@ -69,11 +70,11 @@ static int coroutine_fn bdrv_co_readv_em(BlockDriverState *bs,
 static int coroutine_fn bdrv_co_writev_em(BlockDriverState *bs,
                                          int64_t sector_num, int nb_sectors,
                                          QEMUIOVector *iov);
-static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
-    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
+    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
     BdrvRequestFlags flags);
-static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
-    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
+    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
     BdrvRequestFlags flags);
 static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                                int64_t sector_num,
@@ -90,6 +91,9 @@ static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
 static QTAILQ_HEAD(, BlockDriverState) bdrv_states =
     QTAILQ_HEAD_INITIALIZER(bdrv_states);
 
+static QTAILQ_HEAD(, BlockDriverState) graph_bdrv_states =
+    QTAILQ_HEAD_INITIALIZER(graph_bdrv_states);
+
 static QLIST_HEAD(, BlockDriver) bdrv_drivers =
     QLIST_HEAD_INITIALIZER(bdrv_drivers);
 
@@ -188,7 +192,7 @@ void bdrv_io_limits_enable(BlockDriverState *bs)
  * @is_write:   is the IO a write
  */
 static void bdrv_io_limits_intercept(BlockDriverState *bs,
-                                     int nb_sectors,
+                                     unsigned int bytes,
                                      bool is_write)
 {
     /* does this io must wait */
@@ -201,9 +205,8 @@ static void bdrv_io_limits_intercept(BlockDriverState *bs,
     }
 
     /* the IO will be executed, do the accounting */
-    throttle_account(&bs->throttle_state,
-                     is_write,
-                     nb_sectors * BDRV_SECTOR_SIZE);
+    throttle_account(&bs->throttle_state, is_write, bytes);
+
 
     /* if the next request must wait -> do nothing */
     if (throttle_schedule_timer(&bs->throttle_state, is_write)) {
@@ -214,6 +217,16 @@ static void bdrv_io_limits_intercept(BlockDriverState *bs,
     qemu_co_queue_next(&bs->throttled_reqs[is_write]);
 }
 
+size_t bdrv_opt_mem_align(BlockDriverState *bs)
+{
+    if (!bs || !bs->drv) {
+        /* 4k should be on the safe side */
+        return 4096;
+    }
+
+    return bs->bl.opt_mem_alignment;
+}
+
 /* check if the path starts with "<protocol>:" */
 static int path_has_protocol(const char *path)
 {
@@ -327,7 +340,7 @@ BlockDriverState *bdrv_new(const char *device_name)
     QLIST_INIT(&bs->dirty_bitmaps);
     pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
     if (device_name[0] != '\0') {
-        QTAILQ_INSERT_TAIL(&bdrv_states, bs, list);
+        QTAILQ_INSERT_TAIL(&bdrv_states, bs, device_list);
     }
     bdrv_iostatus_disable(bs);
     notifier_list_init(&bs->close_notifiers);
@@ -479,6 +492,43 @@ int bdrv_create_file(const char* filename, QEMUOptionParameter *options,
     return ret;
 }
 
+int bdrv_refresh_limits(BlockDriverState *bs)
+{
+    BlockDriver *drv = bs->drv;
+
+    memset(&bs->bl, 0, sizeof(bs->bl));
+
+    if (!drv) {
+        return 0;
+    }
+
+    /* Take some limits from the children as a default */
+    if (bs->file) {
+        bdrv_refresh_limits(bs->file);
+        bs->bl.opt_transfer_length = bs->file->bl.opt_transfer_length;
+        bs->bl.opt_mem_alignment = bs->file->bl.opt_mem_alignment;
+    } else {
+        bs->bl.opt_mem_alignment = 512;
+    }
+
+    if (bs->backing_hd) {
+        bdrv_refresh_limits(bs->backing_hd);
+        bs->bl.opt_transfer_length =
+            MAX(bs->bl.opt_transfer_length,
+                bs->backing_hd->bl.opt_transfer_length);
+        bs->bl.opt_mem_alignment =
+            MAX(bs->bl.opt_mem_alignment,
+                bs->backing_hd->bl.opt_mem_alignment);
+    }
+
+    /* Then let the driver override it */
+    if (drv->bdrv_refresh_limits) {
+        return drv->bdrv_refresh_limits(bs);
+    }
+
+    return 0;
+}
+
 /*
  * Create a uniquely-named empty temporary file.
  * Return 0 upon success, otherwise a negative errno value.
@@ -732,6 +782,33 @@ static int bdrv_open_flags(BlockDriverState *bs, int flags)
     return open_flags;
 }
 
+static int bdrv_assign_node_name(BlockDriverState *bs,
+                                 const char *node_name,
+                                 Error **errp)
+{
+    if (!node_name) {
+        return 0;
+    }
+
+    /* empty string node name is invalid */
+    if (node_name[0] == '\0') {
+        error_setg(errp, "Empty node name");
+        return -EINVAL;
+    }
+
+    /* takes care of avoiding duplicates node names */
+    if (bdrv_find_node(node_name)) {
+        error_setg(errp, "Duplicate node name");
+        return -EINVAL;
+    }
+
+    /* copy node name into the bs and insert it into the graph list */
+    pstrcpy(bs->node_name, sizeof(bs->node_name), node_name);
+    QTAILQ_INSERT_TAIL(&graph_bdrv_states, bs, node_list);
+
+    return 0;
+}
+
 /*
  * Common part for opening disk images and files
  *
@@ -742,6 +819,7 @@ static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
 {
     int ret, open_flags;
     const char *filename;
+    const char *node_name = NULL;
     Error *local_err = NULL;
 
     assert(drv != NULL);
@@ -756,6 +834,13 @@ static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
 
     trace_bdrv_open_common(bs, filename ?: "", flags, drv->format_name);
 
+    node_name = qdict_get_try_str(options, "node-name");
+    ret = bdrv_assign_node_name(bs, node_name, errp);
+    if (ret < 0) {
+        return ret;
+    }
+    qdict_del(options, "node-name");
+
     /* bdrv_open() with directly using a protocol as drv. This layer is already
      * opened, so assign it to bs (while file becomes a closed BlockDriverState)
      * and return immediately. */
@@ -765,7 +850,8 @@ static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
     }
 
     bs->open_flags = flags;
-    bs->buffer_alignment = 512;
+    bs->guest_block_size = 512;
+    bs->request_alignment = 512;
     bs->zero_beyond_eof = true;
     open_flags = bdrv_open_flags(bs, flags);
     bs->read_only = !(open_flags & BDRV_O_RDWR);
@@ -833,6 +919,10 @@ static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
         goto free_and_fail;
     }
 
+    bdrv_refresh_limits(bs);
+    assert(bdrv_opt_mem_align(bs) != 0);
+    assert(bs->request_alignment != 0);
+
 #ifndef _WIN32
     if (bs->is_temporary) {
         assert(bs->filename[0] != '\0');
@@ -858,9 +948,10 @@ free_and_fail:
  * dictionary, it needs to use QINCREF() before calling bdrv_file_open.
  */
 int bdrv_file_open(BlockDriverState **pbs, const char *filename,
-                   QDict *options, int flags, Error **errp)
+                   const char *reference, QDict *options, int flags,
+                   Error **errp)
 {
-    BlockDriverState *bs;
+    BlockDriverState *bs = NULL;
     BlockDriver *drv;
     const char *drvname;
     bool allow_protocol_prefix = false;
@@ -872,6 +963,24 @@ int bdrv_file_open(BlockDriverState **pbs, const char *filename,
         options = qdict_new();
     }
 
+    if (reference) {
+        if (filename || qdict_size(options)) {
+            error_setg(errp, "Cannot reference an existing block device with "
+                       "additional options or a new filename");
+            return -EINVAL;
+        }
+        QDECREF(options);
+
+        bs = bdrv_find(reference);
+        if (!bs) {
+            error_setg(errp, "Cannot find block device '%s'", reference);
+            return -ENODEV;
+        }
+        bdrv_ref(bs);
+        *pbs = bs;
+        return 0;
+    }
+
     bs = bdrv_new("");
     bs->options = options;
     options = qdict_clone_shallow(options);
@@ -929,14 +1038,19 @@ int bdrv_file_open(BlockDriverState **pbs, const char *filename,
         goto fail;
     }
 
-    ret = bdrv_open_common(bs, NULL, options, flags, drv, &local_err);
+    if (!drv->bdrv_file_open) {
+        ret = bdrv_open(bs, filename, options, flags, drv, &local_err);
+        options = NULL;
+    } else {
+        ret = bdrv_open_common(bs, NULL, options, flags, drv, &local_err);
+    }
     if (ret < 0) {
         error_propagate(errp, local_err);
         goto fail;
     }
 
     /* Check if any unknown options were used */
-    if (qdict_size(options) != 0) {
+    if (options && (qdict_size(options) != 0)) {
         const QDictEntry *entry = qdict_first(options);
         error_setg(errp, "Block protocol '%s' doesn't support the option '%s'",
                    drv->format_name, entry->key);
@@ -1016,12 +1130,92 @@ int bdrv_open_backing_file(BlockDriverState *bs, QDict *options, Error **errp)
         error_free(local_err);
         return ret;
     }
-    pstrcpy(bs->backing_file, sizeof(bs->backing_file),
-            bs->backing_hd->file->filename);
+
+    if (bs->backing_hd->file) {
+        pstrcpy(bs->backing_file, sizeof(bs->backing_file),
+                bs->backing_hd->file->filename);
+    }
+
+    /* Recalculate the BlockLimits with the backing file */
+    bdrv_refresh_limits(bs);
+
     return 0;
 }
 
 /*
+ * Opens a disk image whose options are given as BlockdevRef in another block
+ * device's options.
+ *
+ * If force_raw is true, bdrv_file_open() will be used, thereby preventing any
+ * image format auto-detection. If it is false and a filename is given,
+ * bdrv_open() will be used for auto-detection.
+ *
+ * If allow_none is true, no image will be opened if filename is false and no
+ * BlockdevRef is given. *pbs will remain unchanged and 0 will be returned.
+ *
+ * bdrev_key specifies the key for the image's BlockdevRef in the options QDict.
+ * That QDict has to be flattened; therefore, if the BlockdevRef is a QDict
+ * itself, all options starting with "${bdref_key}." are considered part of the
+ * BlockdevRef.
+ *
+ * The BlockdevRef will be removed from the options QDict.
+ */
+int bdrv_open_image(BlockDriverState **pbs, const char *filename,
+                    QDict *options, const char *bdref_key, int flags,
+                    bool force_raw, bool allow_none, Error **errp)
+{
+    QDict *image_options;
+    int ret;
+    char *bdref_key_dot;
+    const char *reference;
+
+    bdref_key_dot = g_strdup_printf("%s.", bdref_key);
+    qdict_extract_subqdict(options, &image_options, bdref_key_dot);
+    g_free(bdref_key_dot);
+
+    reference = qdict_get_try_str(options, bdref_key);
+    if (!filename && !reference && !qdict_size(image_options)) {
+        if (allow_none) {
+            ret = 0;
+        } else {
+            error_setg(errp, "A block device must be specified for \"%s\"",
+                       bdref_key);
+            ret = -EINVAL;
+        }
+        goto done;
+    }
+
+    if (filename && !force_raw) {
+        /* If a filename is given and the block driver should be detected
+           automatically (instead of using none), use bdrv_open() in order to do
+           that auto-detection. */
+        BlockDriverState *bs;
+
+        if (reference) {
+            error_setg(errp, "Cannot reference an existing block device while "
+                       "giving a filename");
+            ret = -EINVAL;
+            goto done;
+        }
+
+        bs = bdrv_new("");
+        ret = bdrv_open(bs, filename, image_options, flags, NULL, errp);
+        if (ret < 0) {
+            bdrv_unref(bs);
+        } else {
+            *pbs = bs;
+        }
+    } else {
+        ret = bdrv_file_open(pbs, filename, reference, image_options, flags,
+                             errp);
+    }
+
+done:
+    qdict_del(options, bdref_key);
+    return ret;
+}
+
+/*
  * Opens a disk image (raw, qcow2, vmdk, ...)
  *
  * options is a QDict of options to pass to the block drivers, or NULL for an
@@ -1036,7 +1230,6 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
     /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
     char tmp_filename[PATH_MAX + 1];
     BlockDriverState *file = NULL;
-    QDict *file_options = NULL;
     const char *drvname;
     Error *local_err = NULL;
 
@@ -1122,10 +1315,9 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
         flags |= BDRV_O_ALLOW_RDWR;
     }
 
-    qdict_extract_subqdict(options, &file_options, "file.");
-
-    ret = bdrv_file_open(&file, filename, file_options,
-                         bdrv_open_flags(bs, flags | BDRV_O_UNMAP), &local_err);
+    ret = bdrv_open_image(&file, filename, options, "file",
+                          bdrv_open_flags(bs, flags | BDRV_O_UNMAP), true, true,
+                          &local_err);
     if (ret < 0) {
         goto fail;
     }
@@ -1143,7 +1335,13 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
     }
 
     if (!drv) {
-        ret = find_image_format(file, filename, &drv, &local_err);
+        if (file) {
+            ret = find_image_format(file, filename, &drv, &local_err);
+        } else {
+            error_setg(errp, "Must specify either driver or file");
+            ret = -EINVAL;
+            goto unlink_and_fail;
+        }
     }
 
     if (!drv) {
@@ -1156,7 +1354,7 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
         goto unlink_and_fail;
     }
 
-    if (bs->file != file) {
+    if (file && (bs->file != file)) {
         bdrv_unref(file);
         file = NULL;
     }
@@ -1427,6 +1625,8 @@ void bdrv_reopen_commit(BDRVReopenState *reopen_state)
     reopen_state->bs->enable_write_cache = !!(reopen_state->flags &
                                               BDRV_O_CACHE_WB);
     reopen_state->bs->read_only = !(reopen_state->flags & BDRV_O_RDWR);
+
+    bdrv_refresh_limits(reopen_state->bs);
 }
 
 /*
@@ -1501,7 +1701,7 @@ void bdrv_close_all(void)
 {
     BlockDriverState *bs;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         bdrv_close(bs);
     }
 }
@@ -1530,7 +1730,7 @@ static bool bdrv_requests_pending(BlockDriverState *bs)
 static bool bdrv_requests_pending_all(void)
 {
     BlockDriverState *bs;
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         if (bdrv_requests_pending(bs)) {
             return true;
         }
@@ -1557,7 +1757,7 @@ void bdrv_drain_all(void)
     BlockDriverState *bs;
 
     while (busy) {
-        QTAILQ_FOREACH(bs, &bdrv_states, list) {
+        QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
             bdrv_start_throttled_reqs(bs);
         }
 
@@ -1566,14 +1766,19 @@ void bdrv_drain_all(void)
     }
 }
 
-/* make a BlockDriverState anonymous by removing from bdrv_state list.
+/* make a BlockDriverState anonymous by removing from bdrv_state and
+ * graph_bdrv_state list.
    Also, NULL terminate the device_name to prevent double remove */
 void bdrv_make_anon(BlockDriverState *bs)
 {
     if (bs->device_name[0] != '\0') {
-        QTAILQ_REMOVE(&bdrv_states, bs, list);
+        QTAILQ_REMOVE(&bdrv_states, bs, device_list);
     }
     bs->device_name[0] = '\0';
+    if (bs->node_name[0] != '\0') {
+        QTAILQ_REMOVE(&graph_bdrv_states, bs, node_list);
+    }
+    bs->node_name[0] = '\0';
 }
 
 static void bdrv_rebind(BlockDriverState *bs)
@@ -1593,7 +1798,7 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
     bs_dest->dev_ops            = bs_src->dev_ops;
     bs_dest->dev_opaque         = bs_src->dev_opaque;
     bs_dest->dev                = bs_src->dev;
-    bs_dest->buffer_alignment   = bs_src->buffer_alignment;
+    bs_dest->guest_block_size   = bs_src->guest_block_size;
     bs_dest->copy_on_read       = bs_src->copy_on_read;
 
     bs_dest->enable_write_cache = bs_src->enable_write_cache;
@@ -1627,7 +1832,12 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
     /* keep the same entry in bdrv_states */
     pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
             bs_src->device_name);
-    bs_dest->list = bs_src->list;
+    bs_dest->device_list = bs_src->device_list;
+
+    /* keep the same entry in graph_bdrv_states
+     * We do want to swap name but don't want to swap linked list entries
+     */
+    bs_dest->node_list   = bs_src->node_list;
 }
 
 /*
@@ -1745,7 +1955,7 @@ void bdrv_detach_dev(BlockDriverState *bs, void *dev)
     bs->dev = NULL;
     bs->dev_ops = NULL;
     bs->dev_opaque = NULL;
-    bs->buffer_alignment = 512;
+    bs->guest_block_size = 512;
 }
 
 /* TODO change to return DeviceState * when all users are qdevified */
@@ -1876,10 +2086,10 @@ int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
 int bdrv_commit(BlockDriverState *bs)
 {
     BlockDriver *drv = bs->drv;
-    int64_t sector, total_sectors;
+    int64_t sector, total_sectors, length, backing_length;
     int n, ro, open_flags;
     int ret = 0;
-    uint8_t *buf;
+    uint8_t *buf = NULL;
     char filename[PATH_MAX];
 
     if (!drv)
@@ -1904,7 +2114,29 @@ int bdrv_commit(BlockDriverState *bs)
         }
     }
 
-    total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
+    length = bdrv_getlength(bs);
+    if (length < 0) {
+        ret = length;
+        goto ro_cleanup;
+    }
+
+    backing_length = bdrv_getlength(bs->backing_hd);
+    if (backing_length < 0) {
+        ret = backing_length;
+        goto ro_cleanup;
+    }
+
+    /* If our top snapshot is larger than the backing file image,
+     * grow the backing file image if possible.  If not possible,
+     * we must return an error */
+    if (length > backing_length) {
+        ret = bdrv_truncate(bs->backing_hd, length);
+        if (ret < 0) {
+            goto ro_cleanup;
+        }
+    }
+
+    total_sectors = length >> BDRV_SECTOR_BITS;
     buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
 
     for (sector = 0; sector < total_sectors; sector += n) {
@@ -1913,13 +2145,13 @@ int bdrv_commit(BlockDriverState *bs)
             goto ro_cleanup;
         }
         if (ret) {
-            if (bdrv_read(bs, sector, buf, n) != 0) {
-                ret = -EIO;
+            ret = bdrv_read(bs, sector, buf, n);
+            if (ret < 0) {
                 goto ro_cleanup;
             }
 
-            if (bdrv_write(bs->backing_hd, sector, buf, n) != 0) {
-                ret = -EIO;
+            ret = bdrv_write(bs->backing_hd, sector, buf, n);
+            if (ret < 0) {
                 goto ro_cleanup;
             }
         }
@@ -1927,6 +2159,9 @@ int bdrv_commit(BlockDriverState *bs)
 
     if (drv->bdrv_make_empty) {
         ret = drv->bdrv_make_empty(bs);
+        if (ret < 0) {
+            goto ro_cleanup;
+        }
         bdrv_flush(bs);
     }
 
@@ -1934,9 +2169,11 @@ int bdrv_commit(BlockDriverState *bs)
      * Make sure all data we wrote to the backing device is actually
      * stable on disk.
      */
-    if (bs->backing_hd)
+    if (bs->backing_hd) {
         bdrv_flush(bs->backing_hd);
+    }
 
+    ret = 0;
 ro_cleanup:
     g_free(buf);
 
@@ -1952,7 +2189,7 @@ int bdrv_commit_all(void)
 {
     BlockDriverState *bs;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         if (bs->drv && bs->backing_hd) {
             int ret = bdrv_commit(bs);
             if (ret < 0) {
@@ -1970,6 +2207,10 @@ int bdrv_commit_all(void)
  */
 static void tracked_request_end(BdrvTrackedRequest *req)
 {
+    if (req->serialising) {
+        req->bs->serialising_in_flight--;
+    }
+
     QLIST_REMOVE(req, list);
     qemu_co_queue_restart_all(&req->wait_queue);
 }
@@ -1979,15 +2220,18 @@ static void tracked_request_end(BdrvTrackedRequest *req)
  */
 static void tracked_request_begin(BdrvTrackedRequest *req,
                                   BlockDriverState *bs,
-                                  int64_t sector_num,
-                                  int nb_sectors, bool is_write)
+                                  int64_t offset,
+                                  unsigned int bytes, bool is_write)
 {
     *req = (BdrvTrackedRequest){
         .bs = bs,
-        .sector_num = sector_num,
-        .nb_sectors = nb_sectors,
-        .is_write = is_write,
-        .co = qemu_coroutine_self(),
+        .offset         = offset,
+        .bytes          = bytes,
+        .is_write       = is_write,
+        .co             = qemu_coroutine_self(),
+        .serialising    = false,
+        .overlap_offset = offset,
+        .overlap_bytes  = bytes,
     };
 
     qemu_co_queue_init(&req->wait_queue);
@@ -1995,6 +2239,21 @@ static void tracked_request_begin(BdrvTrackedRequest *req,
     QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
 }
 
+static void mark_request_serialising(BdrvTrackedRequest *req, size_t align)
+{
+    int64_t overlap_offset = req->offset & ~(align - 1);
+    int overlap_bytes = ROUND_UP(req->offset + req->bytes, align)
+                      - overlap_offset;
+
+    if (!req->serialising) {
+        req->bs->serialising_in_flight++;
+        req->serialising = true;
+    }
+
+    req->overlap_offset = MIN(req->overlap_offset, overlap_offset);
+    req->overlap_bytes = MAX(req->overlap_bytes, overlap_bytes);
+}
+
 /**
  * Round a region to cluster boundaries
  */
@@ -2016,53 +2275,75 @@ void bdrv_round_to_clusters(BlockDriverState *bs,
     }
 }
 
+static int bdrv_get_cluster_size(BlockDriverState *bs)
+{
+    BlockDriverInfo bdi;
+    int ret;
+
+    ret = bdrv_get_info(bs, &bdi);
+    if (ret < 0 || bdi.cluster_size == 0) {
+        return bs->request_alignment;
+    } else {
+        return bdi.cluster_size;
+    }
+}
+
 static bool tracked_request_overlaps(BdrvTrackedRequest *req,
-                                     int64_t sector_num, int nb_sectors) {
+                                     int64_t offset, unsigned int bytes)
+{
     /*        aaaa   bbbb */
-    if (sector_num >= req->sector_num + req->nb_sectors) {
+    if (offset >= req->overlap_offset + req->overlap_bytes) {
         return false;
     }
     /* bbbb   aaaa        */
-    if (req->sector_num >= sector_num + nb_sectors) {
+    if (req->overlap_offset >= offset + bytes) {
         return false;
     }
     return true;
 }
 
-static void coroutine_fn wait_for_overlapping_requests(BlockDriverState *bs,
-        int64_t sector_num, int nb_sectors)
+static bool coroutine_fn wait_serialising_requests(BdrvTrackedRequest *self)
 {
+    BlockDriverState *bs = self->bs;
     BdrvTrackedRequest *req;
-    int64_t cluster_sector_num;
-    int cluster_nb_sectors;
     bool retry;
+    bool waited = false;
 
-    /* If we touch the same cluster it counts as an overlap.  This guarantees
-     * that allocating writes will be serialized and not race with each other
-     * for the same cluster.  For example, in copy-on-read it ensures that the
-     * CoR read and write operations are atomic and guest writes cannot
-     * interleave between them.
-     */
-    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
-                           &cluster_sector_num, &cluster_nb_sectors);
+    if (!bs->serialising_in_flight) {
+        return false;
+    }
 
     do {
         retry = false;
         QLIST_FOREACH(req, &bs->tracked_requests, list) {
-            if (tracked_request_overlaps(req, cluster_sector_num,
-                                         cluster_nb_sectors)) {
+            if (req == self || (!req->serialising && !self->serialising)) {
+                continue;
+            }
+            if (tracked_request_overlaps(req, self->overlap_offset,
+                                         self->overlap_bytes))
+            {
                 /* Hitting this means there was a reentrant request, for
                  * example, a block driver issuing nested requests.  This must
                  * never happen since it means deadlock.
                  */
                 assert(qemu_coroutine_self() != req->co);
 
-                qemu_co_queue_wait(&req->wait_queue);
-                retry = true;
-                break;
+                /* If the request is already (indirectly) waiting for us, or
+                 * will wait for us as soon as it wakes up, then just go on
+                 * (instead of producing a deadlock in the former case). */
+                if (!req->waiting_for) {
+                    self->waiting_for = req;
+                    qemu_co_queue_wait(&req->wait_queue);
+                    self->waiting_for = NULL;
+                    retry = true;
+                    waited = true;
+                    break;
+                }
             }
         }
     } while (retry);
+
+    return waited;
 }
 
 /*
@@ -2224,6 +2505,7 @@ int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
     }
     new_top_bs->backing_hd = base_bs;
 
+    bdrv_refresh_limits(new_top_bs);
 
     QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
         /* so that bdrv_close() does not recursively close the chain */
@@ -2271,8 +2553,7 @@ static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
 
 typedef struct RwCo {
     BlockDriverState *bs;
-    int64_t sector_num;
-    int nb_sectors;
+    int64_t offset;
     QEMUIOVector *qiov;
     bool is_write;
     int ret;
@@ -2284,34 +2565,32 @@ static void coroutine_fn bdrv_rw_co_entry(void *opaque)
     RwCo *rwco = opaque;
 
     if (!rwco->is_write) {
-        rwco->ret = bdrv_co_do_readv(rwco->bs, rwco->sector_num,
-                                     rwco->nb_sectors, rwco->qiov,
-                                     rwco->flags);
-    } else {
-        rwco->ret = bdrv_co_do_writev(rwco->bs, rwco->sector_num,
-                                      rwco->nb_sectors, rwco->qiov,
+        rwco->ret = bdrv_co_do_preadv(rwco->bs, rwco->offset,
+                                      rwco->qiov->size, rwco->qiov,
                                       rwco->flags);
+    } else {
+        rwco->ret = bdrv_co_do_pwritev(rwco->bs, rwco->offset,
+                                       rwco->qiov->size, rwco->qiov,
+                                       rwco->flags);
     }
 }
 
 /*
  * Process a vectored synchronous request using coroutines
  */
-static int bdrv_rwv_co(BlockDriverState *bs, int64_t sector_num,
-                       QEMUIOVector *qiov, bool is_write,
-                       BdrvRequestFlags flags)
+static int bdrv_prwv_co(BlockDriverState *bs, int64_t offset,
+                        QEMUIOVector *qiov, bool is_write,
+                        BdrvRequestFlags flags)
 {
     Coroutine *co;
     RwCo rwco = {
         .bs = bs,
-        .sector_num = sector_num,
-        .nb_sectors = qiov->size >> BDRV_SECTOR_BITS,
+        .offset = offset,
         .qiov = qiov,
         .is_write = is_write,
         .ret = NOT_DONE,
         .flags = flags,
     };
-    assert((qiov->size & (BDRV_SECTOR_SIZE - 1)) == 0);
 
     /**
      * In sync call context, when the vcpu is blocked, this throttling timer
@@ -2350,7 +2629,8 @@ static int bdrv_rw_co(BlockDriverState *bs, int64_t sector_num, uint8_t *buf,
     };
 
     qemu_iovec_init_external(&qiov, &iov, 1);
-    return bdrv_rwv_co(bs, sector_num, &qiov, is_write, flags);
+    return bdrv_prwv_co(bs, sector_num << BDRV_SECTOR_BITS,
+                        &qiov, is_write, flags);
 }
 
 /* return < 0 if error. See bdrv_write() for the return codes */
@@ -2386,11 +2666,6 @@ int bdrv_write(BlockDriverState *bs, int64_t sector_num,
     return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true, 0);
 }
 
-int bdrv_writev(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov)
-{
-    return bdrv_rwv_co(bs, sector_num, qiov, true, 0);
-}
-
 int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num,
                       int nb_sectors, BdrvRequestFlags flags)
 {
@@ -2440,117 +2715,53 @@ int bdrv_make_zero(BlockDriverState *bs, BdrvRequestFlags flags)
     }
 }
 
-int bdrv_pread(BlockDriverState *bs, int64_t offset,
-               void *buf, int count1)
+int bdrv_pread(BlockDriverState *bs, int64_t offset, void *buf, int bytes)
 {
-    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
-    int len, nb_sectors, count;
-    int64_t sector_num;
+    QEMUIOVector qiov;
+    struct iovec iov = {
+        .iov_base = (void *)buf,
+        .iov_len = bytes,
+    };
     int ret;
 
-    count = count1;
-    /* first read to align to sector start */
-    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
-    if (len > count)
-        len = count;
-    sector_num = offset >> BDRV_SECTOR_BITS;
-    if (len > 0) {
-        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
-            return ret;
-        memcpy(buf, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), len);
-        count -= len;
-        if (count == 0)
-            return count1;
-        sector_num++;
-        buf += len;
-    }
-
-    /* read the sectors "in place" */
-    nb_sectors = count >> BDRV_SECTOR_BITS;
-    if (nb_sectors > 0) {
-        if ((ret = bdrv_read(bs, sector_num, buf, nb_sectors)) < 0)
-            return ret;
-        sector_num += nb_sectors;
-        len = nb_sectors << BDRV_SECTOR_BITS;
-        buf += len;
-        count -= len;
+    if (bytes < 0) {
+        return -EINVAL;
     }
 
-    /* add data from the last sector */
-    if (count > 0) {
-        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
-            return ret;
-        memcpy(buf, tmp_buf, count);
+    qemu_iovec_init_external(&qiov, &iov, 1);
+    ret = bdrv_prwv_co(bs, offset, &qiov, false, 0);
+    if (ret < 0) {
+        return ret;
     }
-    return count1;
+
+    return bytes;
 }
 
 int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
 {
-    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
-    int len, nb_sectors, count;
-    int64_t sector_num;
     int ret;
 
-    count = qiov->size;
-
-    /* first write to align to sector start */
-    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
-    if (len > count)
-        len = count;
-    sector_num = offset >> BDRV_SECTOR_BITS;
-    if (len > 0) {
-        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
-            return ret;
-        qemu_iovec_to_buf(qiov, 0, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)),
-                          len);
-        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
-            return ret;
-        count -= len;
-        if (count == 0)
-            return qiov->size;
-        sector_num++;
-    }
-
-    /* write the sectors "in place" */
-    nb_sectors = count >> BDRV_SECTOR_BITS;
-    if (nb_sectors > 0) {
-        QEMUIOVector qiov_inplace;
-
-        qemu_iovec_init(&qiov_inplace, qiov->niov);
-        qemu_iovec_concat(&qiov_inplace, qiov, len,
-                          nb_sectors << BDRV_SECTOR_BITS);
-        ret = bdrv_writev(bs, sector_num, &qiov_inplace);
-        qemu_iovec_destroy(&qiov_inplace);
-        if (ret < 0) {
-            return ret;
-        }
-
-        sector_num += nb_sectors;
-        len = nb_sectors << BDRV_SECTOR_BITS;
-        count -= len;
+    ret = bdrv_prwv_co(bs, offset, qiov, true, 0);
+    if (ret < 0) {
+        return ret;
     }
 
-    /* add data from the last sector */
-    if (count > 0) {
-        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
-            return ret;
-        qemu_iovec_to_buf(qiov, qiov->size - count, tmp_buf, count);
-        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
-            return ret;
-    }
     return qiov->size;
 }
 
 int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
-                const void *buf, int count1)
+                const void *buf, int bytes)
 {
     QEMUIOVector qiov;
     struct iovec iov = {
         .iov_base   = (void *) buf,
-        .iov_len    = count1,
+        .iov_len    = bytes,
     };
 
+    if (bytes < 0) {
+        return -EINVAL;
+    }
+
     qemu_iovec_init_external(&qiov, &iov, 1);
     return bdrv_pwritev(bs, offset, &qiov);
 }
@@ -2646,40 +2857,34 @@ err:
 }
 
 /*
- * Handle a read request in coroutine context
+ * Forwards an already correctly aligned request to the BlockDriver. This
+ * handles copy on read and zeroing after EOF; any other features must be
+ * implemented by the caller.
  */
-static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
-    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
-    BdrvRequestFlags flags)
+static int coroutine_fn bdrv_aligned_preadv(BlockDriverState *bs,
+    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
+    int64_t align, QEMUIOVector *qiov, int flags)
 {
     BlockDriver *drv = bs->drv;
-    BdrvTrackedRequest req;
     int ret;
 
-    if (!drv) {
-        return -ENOMEDIUM;
-    }
-    if (bdrv_check_request(bs, sector_num, nb_sectors)) {
-        return -EIO;
-    }
+    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
+    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
 
-    if (bs->copy_on_read) {
-        flags |= BDRV_REQ_COPY_ON_READ;
-    }
-    if (flags & BDRV_REQ_COPY_ON_READ) {
-        bs->copy_on_read_in_flight++;
-    }
+    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
+    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
 
-    if (bs->copy_on_read_in_flight) {
-        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
-    }
-
-    /* throttling disk I/O */
-    if (bs->io_limits_enabled) {
-        bdrv_io_limits_intercept(bs, nb_sectors, false);
+    /* Handle Copy on Read and associated serialisation */
+    if (flags & BDRV_REQ_COPY_ON_READ) {
+        /* If we touch the same cluster it counts as an overlap.  This
+         * guarantees that allocating writes will be serialized and not race
+         * with each other for the same cluster.  For example, in copy-on-read
+         * it ensures that the CoR read and write operations are atomic and
+         * guest writes cannot interleave between them. */
+        mark_request_serialising(req, bdrv_get_cluster_size(bs));
     }
 
-    tracked_request_begin(&req, bs, sector_num, nb_sectors, false);
+    wait_serialising_requests(req);
 
     if (flags & BDRV_REQ_COPY_ON_READ) {
         int pnum;
@@ -2695,6 +2900,7 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
         }
     }
 
+    /* Forward the request to the BlockDriver */
     if (!(bs->zero_beyond_eof && bs->growable)) {
         ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
     } else {
@@ -2708,7 +2914,8 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
         }
 
         total_sectors = DIV_ROUND_UP(len, BDRV_SECTOR_SIZE);
-        max_nb_sectors = MAX(0, total_sectors - sector_num);
+        max_nb_sectors = MAX(0, ROUND_UP(total_sectors - sector_num,
+                                         align >> BDRV_SECTOR_BITS));
         if (max_nb_sectors > 0) {
             ret = drv->bdrv_co_readv(bs, sector_num,
                                      MIN(nb_sectors, max_nb_sectors), qiov);
@@ -2726,15 +2933,95 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
     }
 
 out:
+    return ret;
+}
+
+/*
+ * Handle a read request in coroutine context
+ */
+static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
+    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
+    BdrvRequestFlags flags)
+{
+    BlockDriver *drv = bs->drv;
+    BdrvTrackedRequest req;
+
+    /* TODO Lift BDRV_SECTOR_SIZE restriction in BlockDriver interface */
+    uint64_t align = MAX(BDRV_SECTOR_SIZE, bs->request_alignment);
+    uint8_t *head_buf = NULL;
+    uint8_t *tail_buf = NULL;
+    QEMUIOVector local_qiov;
+    bool use_local_qiov = false;
+    int ret;
+
+    if (!drv) {
+        return -ENOMEDIUM;
+    }
+    if (bdrv_check_byte_request(bs, offset, bytes)) {
+        return -EIO;
+    }
+
+    if (bs->copy_on_read) {
+        flags |= BDRV_REQ_COPY_ON_READ;
+    }
+
+    /* throttling disk I/O */
+    if (bs->io_limits_enabled) {
+        bdrv_io_limits_intercept(bs, bytes, false);
+    }
+
+    /* Align read if necessary by padding qiov */
+    if (offset & (align - 1)) {
+        head_buf = qemu_blockalign(bs, align);
+        qemu_iovec_init(&local_qiov, qiov->niov + 2);
+        qemu_iovec_add(&local_qiov, head_buf, offset & (align - 1));
+        qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
+        use_local_qiov = true;
+
+        bytes += offset & (align - 1);
+        offset = offset & ~(align - 1);
+    }
+
+    if ((offset + bytes) & (align - 1)) {
+        if (!use_local_qiov) {
+            qemu_iovec_init(&local_qiov, qiov->niov + 1);
+            qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
+            use_local_qiov = true;
+        }
+        tail_buf = qemu_blockalign(bs, align);
+        qemu_iovec_add(&local_qiov, tail_buf,
+                       align - ((offset + bytes) & (align - 1)));
+
+        bytes = ROUND_UP(bytes, align);
+    }
+
+    tracked_request_begin(&req, bs, offset, bytes, false);
+    ret = bdrv_aligned_preadv(bs, &req, offset, bytes, align,
+                              use_local_qiov ? &local_qiov : qiov,
+                              flags);
     tracked_request_end(&req);
 
-    if (flags & BDRV_REQ_COPY_ON_READ) {
-        bs->copy_on_read_in_flight--;
+    if (use_local_qiov) {
+        qemu_iovec_destroy(&local_qiov);
+        qemu_vfree(head_buf);
+        qemu_vfree(tail_buf);
     }
 
     return ret;
 }
 
+static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+    BdrvRequestFlags flags)
+{
+    if (nb_sectors < 0 || nb_sectors > (UINT_MAX >> BDRV_SECTOR_BITS)) {
+        return -EINVAL;
+    }
+
+    return bdrv_co_do_preadv(bs, sector_num << BDRV_SECTOR_BITS,
+                             nb_sectors << BDRV_SECTOR_BITS, qiov, flags);
+}
+
 int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
     int nb_sectors, QEMUIOVector *qiov)
 {
@@ -2828,46 +3115,37 @@ static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
 }
 
 /*
- * Handle a write request in coroutine context
+ * Forwards an already correctly aligned write request to the BlockDriver.
  */
-static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
-    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
-    BdrvRequestFlags flags)
+static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
+    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
+    QEMUIOVector *qiov, int flags)
 {
     BlockDriver *drv = bs->drv;
-    BdrvTrackedRequest req;
+    bool waited;
     int ret;
 
-    if (!bs->drv) {
-        return -ENOMEDIUM;
-    }
-    if (bs->read_only) {
-        return -EACCES;
-    }
-    if (bdrv_check_request(bs, sector_num, nb_sectors)) {
-        return -EIO;
-    }
+    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
+    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
 
-    if (bs->copy_on_read_in_flight) {
-        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
-    }
-
-    /* throttling disk I/O */
-    if (bs->io_limits_enabled) {
-        bdrv_io_limits_intercept(bs, nb_sectors, true);
-    }
+    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
+    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
 
-    tracked_request_begin(&req, bs, sector_num, nb_sectors, true);
+    waited = wait_serialising_requests(req);
+    assert(!waited || !req->serialising);
 
-    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, &req);
+    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, req);
 
     if (ret < 0) {
         /* Do nothing, write notifier decided to fail this request */
     } else if (flags & BDRV_REQ_ZERO_WRITE) {
+        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_ZERO);
         ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
     } else {
+        BLKDBG_EVENT(bs, BLKDBG_PWRITEV);
         ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
     }
+    BLKDBG_EVENT(bs, BLKDBG_PWRITEV_DONE);
 
     if (ret == 0 && !bs->enable_write_cache) {
         ret = bdrv_co_flush(bs);
@@ -2882,11 +3160,143 @@ static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
         bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
     }
 
+    return ret;
+}
+
+/*
+ * Handle a write request in coroutine context
+ */
+static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
+    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
+    BdrvRequestFlags flags)
+{
+    BdrvTrackedRequest req;
+    /* TODO Lift BDRV_SECTOR_SIZE restriction in BlockDriver interface */
+    uint64_t align = MAX(BDRV_SECTOR_SIZE, bs->request_alignment);
+    uint8_t *head_buf = NULL;
+    uint8_t *tail_buf = NULL;
+    QEMUIOVector local_qiov;
+    bool use_local_qiov = false;
+    int ret;
+
+    if (!bs->drv) {
+        return -ENOMEDIUM;
+    }
+    if (bs->read_only) {
+        return -EACCES;
+    }
+    if (bdrv_check_byte_request(bs, offset, bytes)) {
+        return -EIO;
+    }
+
+    /* throttling disk I/O */
+    if (bs->io_limits_enabled) {
+        bdrv_io_limits_intercept(bs, bytes, true);
+    }
+
+    /*
+     * Align write if necessary by performing a read-modify-write cycle.
+     * 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);
+
+    if (offset & (align - 1)) {
+        QEMUIOVector head_qiov;
+        struct iovec head_iov;
+
+        mark_request_serialising(&req, align);
+        wait_serialising_requests(&req);
+
+        head_buf = qemu_blockalign(bs, align);
+        head_iov = (struct iovec) {
+            .iov_base   = head_buf,
+            .iov_len    = align,
+        };
+        qemu_iovec_init_external(&head_qiov, &head_iov, 1);
+
+        BLKDBG_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);
+
+        qemu_iovec_init(&local_qiov, qiov->niov + 2);
+        qemu_iovec_add(&local_qiov, head_buf, offset & (align - 1));
+        qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
+        use_local_qiov = true;
+
+        bytes += offset & (align - 1);
+        offset = offset & ~(align - 1);
+    }
+
+    if ((offset + bytes) & (align - 1)) {
+        QEMUIOVector tail_qiov;
+        struct iovec tail_iov;
+        size_t tail_bytes;
+        bool waited;
+
+        mark_request_serialising(&req, align);
+        waited = wait_serialising_requests(&req);
+        assert(!waited || !use_local_qiov);
+
+        tail_buf = qemu_blockalign(bs, align);
+        tail_iov = (struct iovec) {
+            .iov_base   = tail_buf,
+            .iov_len    = align,
+        };
+        qemu_iovec_init_external(&tail_qiov, &tail_iov, 1);
+
+        BLKDBG_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);
+
+        if (!use_local_qiov) {
+            qemu_iovec_init(&local_qiov, qiov->niov + 1);
+            qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
+            use_local_qiov = true;
+        }
+
+        tail_bytes = (offset + bytes) & (align - 1);
+        qemu_iovec_add(&local_qiov, tail_buf + tail_bytes, align - tail_bytes);
+
+        bytes = ROUND_UP(bytes, align);
+    }
+
+    ret = bdrv_aligned_pwritev(bs, &req, offset, bytes,
+                               use_local_qiov ? &local_qiov : qiov,
+                               flags);
+
+fail:
     tracked_request_end(&req);
 
+    if (use_local_qiov) {
+        qemu_iovec_destroy(&local_qiov);
+        qemu_vfree(head_buf);
+        qemu_vfree(tail_buf);
+    }
+
     return ret;
 }
 
+static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
+    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
+    BdrvRequestFlags flags)
+{
+    if (nb_sectors < 0 || nb_sectors > (INT_MAX >> BDRV_SECTOR_BITS)) {
+        return -EINVAL;
+    }
+
+    return bdrv_co_do_pwritev(bs, sector_num << BDRV_SECTOR_BITS,
+                              nb_sectors << BDRV_SECTOR_BITS, qiov, flags);
+}
+
 int coroutine_fn bdrv_co_writev(BlockDriverState *bs, int64_t sector_num,
     int nb_sectors, QEMUIOVector *qiov)
 {
@@ -3110,11 +3520,12 @@ void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
     }
 }
 
+/* This function is to find block backend bs */
 BlockDriverState *bdrv_find(const char *name)
 {
     BlockDriverState *bs;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         if (!strcmp(name, bs->device_name)) {
             return bs;
         }
@@ -3122,19 +3533,83 @@ BlockDriverState *bdrv_find(const char *name)
     return NULL;
 }
 
+/* This function is to find a node in the bs graph */
+BlockDriverState *bdrv_find_node(const char *node_name)
+{
+    BlockDriverState *bs;
+
+    assert(node_name);
+
+    QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
+        if (!strcmp(node_name, bs->node_name)) {
+            return bs;
+        }
+    }
+    return NULL;
+}
+
+/* Put this QMP function here so it can access the static graph_bdrv_states. */
+BlockDeviceInfoList *bdrv_named_nodes_list(void)
+{
+    BlockDeviceInfoList *list, *entry;
+    BlockDriverState *bs;
+
+    list = NULL;
+    QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
+        entry = g_malloc0(sizeof(*entry));
+        entry->value = bdrv_block_device_info(bs);
+        entry->next = list;
+        list = entry;
+    }
+
+    return list;
+}
+
+BlockDriverState *bdrv_lookup_bs(const char *device,
+                                 const char *node_name,
+                                 Error **errp)
+{
+    BlockDriverState *bs = NULL;
+
+    if ((!device && !node_name) || (device && node_name)) {
+        error_setg(errp, "Use either device or node-name but not both");
+        return NULL;
+    }
+
+    if (device) {
+        bs = bdrv_find(device);
+
+        if (!bs) {
+            error_set(errp, QERR_DEVICE_NOT_FOUND, device);
+            return NULL;
+        }
+
+        return bs;
+    }
+
+    bs = bdrv_find_node(node_name);
+
+    if (!bs) {
+        error_set(errp, QERR_DEVICE_NOT_FOUND, node_name);
+        return NULL;
+    }
+
+    return bs;
+}
+
 BlockDriverState *bdrv_next(BlockDriverState *bs)
 {
     if (!bs) {
         return QTAILQ_FIRST(&bdrv_states);
     }
-    return QTAILQ_NEXT(bs, list);
+    return QTAILQ_NEXT(bs, device_list);
 }
 
 void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
 {
     BlockDriverState *bs;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         it(opaque, bs);
     }
 }
@@ -3154,7 +3629,7 @@ int bdrv_flush_all(void)
     BlockDriverState *bs;
     int result = 0;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         int ret = bdrv_flush(bs);
         if (ret < 0 && !result) {
             result = ret;
@@ -4278,7 +4753,7 @@ void bdrv_invalidate_cache_all(void)
 {
     BlockDriverState *bs;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         bdrv_invalidate_cache(bs);
     }
 }
@@ -4287,7 +4762,7 @@ void bdrv_clear_incoming_migration_all(void)
 {
     BlockDriverState *bs;
 
-    QTAILQ_FOREACH(bs, &bdrv_states, list) {
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
         bs->open_flags = bs->open_flags & ~(BDRV_O_INCOMING);
     }
 }
@@ -4314,9 +4789,15 @@ int bdrv_flush(BlockDriverState *bs)
     return rwco.ret;
 }
 
+typedef struct DiscardCo {
+    BlockDriverState *bs;
+    int64_t sector_num;
+    int nb_sectors;
+    int ret;
+} DiscardCo;
 static void coroutine_fn bdrv_discard_co_entry(void *opaque)
 {
-    RwCo *rwco = opaque;
+    DiscardCo *rwco = opaque;
 
     rwco->ret = bdrv_co_discard(rwco->bs, rwco->sector_num, rwco->nb_sectors);
 }
@@ -4400,7 +4881,7 @@ int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
 int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
 {
     Coroutine *co;
-    RwCo rwco = {
+    DiscardCo rwco = {
         .bs = bs,
         .sector_num = sector_num,
         .nb_sectors = nb_sectors,
@@ -4505,14 +4986,14 @@ BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
     return NULL;
 }
 
-void bdrv_set_buffer_alignment(BlockDriverState *bs, int align)
+void bdrv_set_guest_block_size(BlockDriverState *bs, int align)
 {
-    bs->buffer_alignment = align;
+    bs->guest_block_size = align;
 }
 
 void *qemu_blockalign(BlockDriverState *bs, size_t size)
 {
-    return qemu_memalign((bs && bs->buffer_alignment) ? bs->buffer_alignment : 512, size);
+    return qemu_memalign(bdrv_opt_mem_align(bs), size);
 }
 
 /*
@@ -4521,9 +5002,13 @@ void *qemu_blockalign(BlockDriverState *bs, size_t size)
 bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
 {
     int i;
+    size_t alignment = bdrv_opt_mem_align(bs);
 
     for (i = 0; i < qiov->niov; i++) {
-        if ((uintptr_t) qiov->iov[i].iov_base % bs->buffer_alignment) {
+        if ((uintptr_t) qiov->iov[i].iov_base % alignment) {
+            return false;
+        }
+        if (qiov->iov[i].iov_len % alignment) {
             return false;
         }
     }
@@ -4875,21 +5360,68 @@ int bdrv_amend_options(BlockDriverState *bs, QEMUOptionParameter *options)
     return bs->drv->bdrv_amend_options(bs, options);
 }
 
-ExtSnapshotPerm bdrv_check_ext_snapshot(BlockDriverState *bs)
+/* Used to recurse on single child block filters.
+ * Single child block filter will store their child in bs->file.
+ */
+bool bdrv_generic_is_first_non_filter(BlockDriverState *bs,
+                                      BlockDriverState *candidate)
 {
-    if (bs->drv->bdrv_check_ext_snapshot) {
-        return bs->drv->bdrv_check_ext_snapshot(bs);
+    if (!bs->drv) {
+        return false;
     }
 
-    if (bs->file && bs->file->drv && bs->file->drv->bdrv_check_ext_snapshot) {
-        return bs->file->drv->bdrv_check_ext_snapshot(bs);
+    if (!bs->drv->authorizations[BS_IS_A_FILTER]) {
+        if (bs == candidate) {
+            return true;
+        } else {
+            return false;
+        }
+    }
+
+    if (!bs->drv->authorizations[BS_FILTER_PASS_DOWN]) {
+        return false;
+    }
+
+    if (!bs->file) {
+        return false;
     }
 
-    /* external snapshots are allowed by default */
-    return EXT_SNAPSHOT_ALLOWED;
+    return bdrv_recurse_is_first_non_filter(bs->file, candidate);
 }
 
-ExtSnapshotPerm bdrv_check_ext_snapshot_forbidden(BlockDriverState *bs)
+bool bdrv_recurse_is_first_non_filter(BlockDriverState *bs,
+                                      BlockDriverState *candidate)
+{
+    if (bs->drv && bs->drv->bdrv_recurse_is_first_non_filter) {
+        return bs->drv->bdrv_recurse_is_first_non_filter(bs, candidate);
+    }
+
+    return bdrv_generic_is_first_non_filter(bs, candidate);
+}
+
+/* This function checks if the candidate is the first non filter bs down it's
+ * bs chain. Since we don't have pointers to parents it explore all bs chains
+ * from the top. Some filters can choose not to pass down the recursion.
+ */
+bool bdrv_is_first_non_filter(BlockDriverState *candidate)
 {
-    return EXT_SNAPSHOT_FORBIDDEN;
+    BlockDriverState *bs;
+
+    /* walk down the bs forest recursively */
+    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
+        bool perm;
+
+        if (!bs->file) {
+            continue;
+        }
+
+        perm = bdrv_recurse_is_first_non_filter(bs->file, candidate);
+
+        /* candidate is the first non filter */
+        if (perm) {
+            return true;
+        }
+    }
+
+    return false;
 }