diff --git a/block/block-backend.c b/block/block-backend.c
index 6880659665..730b8a9949 100644
--- a/block/block-backend.c
+++ b/block/block-backend.c
@@ -716,6 +716,11 @@ static int coroutine_fn blk_co_preadv(BlockBackend *blk, int64_t offset,
         return ret;
     }
 
+    /* throttling disk I/O */
+    if (blk->public.throttle_state) {
+        throttle_group_co_io_limits_intercept(blk, bytes, false);
+    }
+
     return bdrv_co_preadv(blk_bs(blk), offset, bytes, qiov, flags);
 }
 
@@ -730,6 +735,11 @@ static int coroutine_fn blk_co_pwritev(BlockBackend *blk, int64_t offset,
         return ret;
     }
 
+    /* throttling disk I/O */
+    if (blk->public.throttle_state) {
+        throttle_group_co_io_limits_intercept(blk, bytes, true);
+    }
+
     if (!blk->enable_write_cache) {
         flags |= BDRV_REQ_FUA;
     }
diff --git a/block/io.c b/block/io.c
index bdbaa1c0da..cf2ac4cca5 100644
--- a/block/io.c
+++ b/block/io.c
@@ -1083,11 +1083,6 @@ int coroutine_fn bdrv_co_preadv(BlockDriverState *bs,
         flags |= BDRV_REQ_COPY_ON_READ;
     }
 
-    /* throttling disk I/O */
-    if (bs->blk && blk_get_public(bs->blk)->throttle_state) {
-        throttle_group_co_io_limits_intercept(bs, bytes, false);
-    }
-
     /* Align read if necessary by padding qiov */
     if (offset & (align - 1)) {
         head_buf = qemu_blockalign(bs, align);
@@ -1444,11 +1439,6 @@ int coroutine_fn bdrv_co_pwritev(BlockDriverState *bs,
         return ret;
     }
 
-    /* throttling disk I/O */
-    if (bs->blk && blk_get_public(bs->blk)->throttle_state) {
-        throttle_group_co_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
diff --git a/block/throttle-groups.c b/block/throttle-groups.c
index 56dc311867..3db8cf74a5 100644
--- a/block/throttle-groups.c
+++ b/block/throttle-groups.c
@@ -284,18 +284,17 @@ static void schedule_next_request(BlockBackend *blk, bool is_write)
  * if necessary, and schedule the next request using a round robin
  * algorithm.
  *
- * @bs:        the current BlockDriverState
+ * @blk:       the current BlockBackend
  * @bytes:     the number of bytes for this I/O
  * @is_write:  the type of operation (read/write)
  */
-void coroutine_fn throttle_group_co_io_limits_intercept(BlockDriverState *bs,
+void coroutine_fn throttle_group_co_io_limits_intercept(BlockBackend *blk,
                                                         unsigned int bytes,
                                                         bool is_write)
 {
     bool must_wait;
     BlockBackend *token;
 
-    BlockBackend *blk = bs->blk;
     BlockBackendPublic *blkp = blk_get_public(blk);
     ThrottleGroup *tg = container_of(blkp->throttle_state, ThrottleGroup, ts);
     qemu_mutex_lock(&tg->lock);
diff --git a/include/block/throttle-groups.h b/include/block/throttle-groups.h
index 840ba443ae..ac4224835b 100644
--- a/include/block/throttle-groups.h
+++ b/include/block/throttle-groups.h
@@ -40,7 +40,7 @@ void throttle_group_register_blk(BlockBackend *blk, const char *groupname);
 void throttle_group_unregister_blk(BlockBackend *blk);
 void throttle_group_restart_blk(BlockBackend *blk);
 
-void coroutine_fn throttle_group_co_io_limits_intercept(BlockDriverState *bs,
+void coroutine_fn throttle_group_co_io_limits_intercept(BlockBackend *blk,
                                                         unsigned int bytes,
                                                         bool is_write);