Consider discard option when writing zeros

When opening an image with discard=off, we punch hole in the image when
writing zeroes, making the image sparse. This breaks users that want to
ensure that writes cannot fail with ENOSPACE by using fully allocated
images[1].

bdrv_co_pwrite_zeroes() correctly disables BDRV_REQ_MAY_UNMAP if we
opened the child without discard=unmap or discard=on. But we don't go
through this function when accessing the top node. Move the check down
to bdrv_co_do_pwrite_zeroes() which seems to be used in all code paths.

This change implements the documented behavior, punching holes only when
opening the image with discard=on or discard=unmap. This may not be the
best default but can improve it later.

The test depends on a file system supporting discard, deallocating the
entire file when punching hole with the length of the entire file.
Tested with xfs, ext4, and tmpfs.

[1] https://lists.nongnu.org/archive/html/qemu-discuss/2024-06/msg00003.html

Signed-off-by: Nir Soffer <nsoffer@redhat.com>
Message-id: 20240628202058.1964986-3-nsoffer@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/io.c

@@ -1862,6 +1862,11 @@ bdrv_co_do_pwrite_zeroes(BlockDriverState *bs, int64_t offset, int64_t bytes,
         return -EINVAL;
     }
 
+    /* If opened with discard=off we should never unmap. */
+    if (!(bs->open_flags & BDRV_O_UNMAP)) {
+        flags &= ~BDRV_REQ_MAY_UNMAP;
+    }
+
     /* Invalidate the cached block-status data range if this write overlaps */
     bdrv_bsc_invalidate_range(bs, offset, bytes);
 
@@ -2315,10 +2320,6 @@ int coroutine_fn bdrv_co_pwrite_zeroes(BdrvChild *child, int64_t offset,
     trace_bdrv_co_pwrite_zeroes(child->bs, offset, bytes, flags);
     assert_bdrv_graph_readable();
 
-    if (!(child->bs->open_flags & BDRV_O_UNMAP)) {
-        flags &= ~BDRV_REQ_MAY_UNMAP;
-    }
-
     return bdrv_co_pwritev(child, offset, bytes, NULL,
                            BDRV_REQ_ZERO_WRITE | flags);
 }