On 04/13/2013 05:11 AM, Wenchao Xia wrote: > Now qmp_transaction() can be extended with other operation, > external snapshot or backing chain creation, is just one case it.
This read a bit awkwardly. Might I suggest:
block: use callbacks in qmp_transaction()
Make it easier to add other operations to qmp_transaction() by using
callbacks, with external snapshots serving as an example implementation
of the callbacks.
>
> Signed-off-by: Wenchao Xia <xiaw...@linux.vnet.ibm.com>
> ---
> blockdev.c | 68 +++++++++++++++++++++++++++++++++++++++++++++++++++--------
> 1 files changed, 58 insertions(+), 10 deletions(-)
>
> diff --git a/blockdev.c b/blockdev.c
> index 3e69569..9f0acfb 100644
> --- a/blockdev.c
> +++ b/blockdev.c
> @@ -779,14 +779,35 @@ void qmp_blockdev_snapshot_sync(const char *device,
> const char *snapshot_file,
>
>
> /* New and old BlockDriverState structs for group snapshots */
> +
> +/* Only in prepare() it may fail, so roll back just need to take care
> + what is done in prepare(). */
Better might be:
/* Only prepare() may fail. In a single transaction, only one of
commit() or rollback() will be called. */
> +typedef struct BdrvActionOps {
> + /* Prepare the work, must NOT be NULL. */
> + int (*prepare)(BlockdevAction *action, void **p_opaque, Error **errp);
Based on the comments on 1/5, should prepare have a void signature and
just let errp serve to indicate failure?
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
