On 13.02.2017 18:22, Kevin Wolf wrote:
> Now that all block drivers with children tell us what permissions they
> need from each of their children, bdrv_attach_child() can use this
> information and make the right requirements while trying to attach new
> children.
>
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
> block.c | 17 +++++++++++++++--
> 1 file changed, 15 insertions(+), 2 deletions(-)
>
> diff --git a/block.c b/block.c
> index 682548a..fd06549 100644
> --- a/block.c
> +++ b/block.c
> @@ -1558,10 +1558,21 @@ BdrvChild *bdrv_attach_child(BlockDriverState *parent_bs,
> Error **errp)
> {
> BdrvChild *child;
> + uint64_t perm, shared_perm;
> +
> + /* XXX This check is a workaround for bdrv_open_inherit(), which attaches
> + * bs->file before bs is opened for format probing. The 'else' branch
> + * really shouldn't exist. */
> + if (parent_bs->drv) {
> + parent_bs->drv->bdrv_child_perm(parent_bs, NULL, child_role,
> + 0, BLK_PERM_ALL, &perm, &shared_perm);
> + } else {
> + perm = 0;
> + shared_perm = BLK_PERM_ALL;
> + }
>
> - /* FIXME Use real permissions */
> child = bdrv_root_attach_child(child_bs, child_name, child_role,
> - 0, BLK_PERM_ALL, parent_bs, errp);
> + perm, shared_perm, parent_bs, errp);
> if (child == NULL) {
> return NULL;
> }
> @@ -2041,6 +2052,8 @@ static BlockDriverState *bdrv_open_inherit(const char *filename,
> * probing, the block drivers will do their own bdrv_open_child() for the
> * same BDS, which is why we put the node name back into options. */
> if ((flags & BDRV_O_PROTOCOL) == 0) {
> + /* FIXME Really shouldn't try to open a child for a still closed bs,
> + * see bdrv_attach_child() */
That gives me hope for my remark on patch 1. :-)
Max
> file = bdrv_open_child(filename, options, "file", bs,
> &child_file, true, &local_err);
> if (local_err) {
>