Re: Prevent compiler warning on block.c

[Thomas Huth]

On 05/05/2021 10.05, Vladimir Sementsov-Ogievskiy wrote:
> 05.05.2021 10:59, Miroslav Rezanina wrote:
> > Commit 3108a15cf (block: introduce bdrv_drop_filter()) introduced 
> > uninitialized
> > variable to_cow_parent in bdrv_replace_node_common function that is used 
> > only when
> > detach_subchain is true. It is used in two places. First if block properly 
> > initialize
> > the variable and second block use it.
> >
> > However, compiler treats this two blocks as two independent cases so it 
> > thinks first
> > block can fail test and second one pass (although both use same 
> > condition). This cause
> > warning that variable can be uninitialized in second block.
> >
> > To prevent this warning, initialize the variable with NULL.
> >
> > Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
> > ---
> > diff --git a/block.c b/block.c
> > index 874c22c43e..3ca27bd2d9 100644
> > --- a/block.c
> > +++ b/block.c
> > @@ -4851,7 +4851,7 @@ static int bdrv_replace_node_common(BlockDriverState 
> > *from,
> >       Transaction *tran = tran_new();
> >       g_autoptr(GHashTable) found = NULL;
> >       g_autoptr(GSList) refresh_list = NULL;
> > -    BlockDriverState *to_cow_parent;
> > +    BlockDriverState *to_cow_parent = NULL;
> >       int ret;
> >
> >       if (detach_subchain) {
>
> Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

This just popped up again here:

 https://lists.gnu.org/archive/html/qemu-devel/2021-06/msg02067.html

Kevin, Max, could you please pick it up to get this problem fixed?

 Thomas