On 08/03/2017 10:03 AM, Kevin Wolf wrote:
> This serves as a regression test for the bugs that were just fixed for
> bdrv_reopen() between read-only and read-write mode.

If I'm right that this also fixes the difference between intermediate
vs. live commit to an initial read-write image that can't be reopened,
can we add that to this test?

> 
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
>  tests/qemu-iotests/187     | 69 ++++++++++++++++++++++++++++++++++++++++++++++
>  tests/qemu-iotests/187.out | 18 ++++++++++++
>  tests/qemu-iotests/group   |  1 +
>  3 files changed, 88 insertions(+)
>  create mode 100755 tests/qemu-iotests/187
>  create mode 100644 tests/qemu-iotests/187.out
> 

> +
> +_cleanup()
> +{
> +	_cleanup_test_img
> +    rm -f "$TEST_IMG.2"
> +    rm -f "$TEST_IMG.3"
> +}
> +trap "_cleanup; exit \$status" 0 1 2 3 15

There are pending patches for 2.11 that will need to tweak this. But for
2.10, this is fine.


> +++ b/tests/qemu-iotests/187.out
> @@ -0,0 +1,18 @@
> +QA output created by 187
> +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
> +
> +Start from read-only
> +
> +Block node is read-only
> +wrote 65536/65536 bytes at offset 0
> +64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +Block node is read-only
> +

Why the difference in error messages, when starting read-only,

> +Start from read-write
> +
> +wrote 65536/65536 bytes at offset 0
> +64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +write failed: Operation not permitted

vs. when reopened read-only?

I don't see it as a flaw in the test, so much as another odd difference
in code paths that we may later want to improve.

So, as written,
Reviewed-by: Eric Blake <eblake@redhat.com>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org