1 files changed, 61 insertions, 15 deletions

diff --git a/block/qcow2.c b/block/qcow2.c
index 470734be9f..62febfc386 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1757,13 +1757,6 @@ static void qcow2_invalidate_cache(BlockDriverState *bs, Error **errp)
 
     qcow2_close(bs);
 
-    bdrv_invalidate_cache(bs->file->bs, &local_err);
-    if (local_err) {
-        error_propagate(errp, local_err);
-        bs->drv = NULL;
-        return;
-    }
-
     memset(s, 0, sizeof(BDRVQcow2State));
     options = qdict_clone_shallow(bs->options);
 
@@ -2207,7 +2200,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
             cpu_to_be64(QCOW2_COMPAT_LAZY_REFCOUNTS);
     }
 
-    ret = blk_pwrite(blk, 0, header, cluster_size);
+    ret = blk_pwrite(blk, 0, header, cluster_size, 0);
     g_free(header);
     if (ret < 0) {
         error_setg_errno(errp, -ret, "Could not write qcow2 header");
@@ -2217,7 +2210,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
     /* Write a refcount table with one refcount block */
     refcount_table = g_malloc0(2 * cluster_size);
     refcount_table[0] = cpu_to_be64(2 * cluster_size);
-    ret = blk_pwrite(blk, cluster_size, refcount_table, 2 * cluster_size);
+    ret = blk_pwrite(blk, cluster_size, refcount_table, 2 * cluster_size, 0);
     g_free(refcount_table);
 
     if (ret < 0) {
@@ -2411,21 +2404,74 @@ finish:
     return ret;
 }
 
+
+static bool is_zero_cluster(BlockDriverState *bs, int64_t start)
+{
+    BDRVQcow2State *s = bs->opaque;
+    int nr;
+    BlockDriverState *file;
+    int64_t res = bdrv_get_block_status_above(bs, NULL, start,
+                                              s->cluster_sectors, &nr, &file);
+    return res >= 0 && ((res & BDRV_BLOCK_ZERO) || !(res & BDRV_BLOCK_DATA));
+}
+
+static bool is_zero_cluster_top_locked(BlockDriverState *bs, int64_t start)
+{
+    BDRVQcow2State *s = bs->opaque;
+    int nr = s->cluster_sectors;
+    uint64_t off;
+    int ret;
+
+    ret = qcow2_get_cluster_offset(bs, start << BDRV_SECTOR_BITS, &nr, &off);
+    return ret == QCOW2_CLUSTER_UNALLOCATED || ret == QCOW2_CLUSTER_ZERO;
+}
+
 static coroutine_fn int qcow2_co_write_zeroes(BlockDriverState *bs,
     int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
 {
     int ret;
     BDRVQcow2State *s = bs->opaque;
 
-    /* Emulate misaligned zero writes */
-    if (sector_num % s->cluster_sectors || nb_sectors % s->cluster_sectors) {
-        return -ENOTSUP;
+    int head = sector_num % s->cluster_sectors;
+    int tail = (sector_num + nb_sectors) % s->cluster_sectors;
+
+    if (head != 0 || tail != 0) {
+        int64_t cl_end = -1;
+
+        sector_num -= head;
+        nb_sectors += head;
+
+        if (tail != 0) {
+            nb_sectors += s->cluster_sectors - tail;
+        }
+
+        if (!is_zero_cluster(bs, sector_num)) {
+            return -ENOTSUP;
+        }
+
+        if (nb_sectors > s->cluster_sectors) {
+            /* Technically the request can cover 2 clusters, f.e. 4k write
+               at s->cluster_sectors - 2k offset. One of these cluster can
+               be zeroed, one unallocated */
+            cl_end = sector_num + nb_sectors - s->cluster_sectors;
+            if (!is_zero_cluster(bs, cl_end)) {
+                return -ENOTSUP;
+            }
+        }
+
+        qemu_co_mutex_lock(&s->lock);
+        /* We can have new write after previous check */
+        if (!is_zero_cluster_top_locked(bs, sector_num) ||
+                (cl_end > 0 && !is_zero_cluster_top_locked(bs, cl_end))) {
+            qemu_co_mutex_unlock(&s->lock);
+            return -ENOTSUP;
+        }
+    } else {
+        qemu_co_mutex_lock(&s->lock);
     }
 
     /* Whatever is left can use real zero clusters */
-    qemu_co_mutex_lock(&s->lock);
-    ret = qcow2_zero_clusters(bs, sector_num << BDRV_SECTOR_BITS,
-        nb_sectors);
+    ret = qcow2_zero_clusters(bs, sector_num << BDRV_SECTOR_BITS, nb_sectors);
     qemu_co_mutex_unlock(&s->lock);
 
     return ret;