On Tue, Jul 13, 2021 at 12:27:48PM +0300, Vladimir Sementsov-Ogievskiy wrote:
> 13.07.2021 00:55, Richard Henderson wrote:
> > From clang-13:
> > nbd/server.c:976:22: error: variable 'bitmaps' set but not used \
> > [-Werror,-Wunused-but-set-variable]
> >
> > +++ b/nbd/server.c
> > @@ -973,7 +973,6 @@ static int nbd_negotiate_meta_queries(NBDClient *client,
> > {
> > int ret;
> > g_autofree char *export_name = NULL;
> > - g_autofree bool *bitmaps = NULL;
> > NBDExportMetaContexts local_meta = {0};
> > uint32_t nb_queries;
> > size_t i;
> > @@ -1007,9 +1006,6 @@ static int nbd_negotiate_meta_queries(NBDClient
> > *client,
> > "export '%s' not present", sane_name);
> > }
> > meta->bitmaps = g_new0(bool, meta->exp->nr_export_bitmaps);
> > - if (client->opt == NBD_OPT_LIST_META_CONTEXT) {
> > - bitmaps = meta->bitmaps;
> > - }
> > ret = nbd_opt_read(client, &nb_queries, sizeof(nb_queries), false,
> > errp);
> > if (ret <= 0) {
> >
>
>
> Hm. I'm afraid, this way meta->bitmaps will be leaked in
> NBD_OPT_LIST_META_CONTEXT case.
>
> Actually, "bitmaps" _is_ used, in cleanup handler, setup by g_autofree. So
> it's a false positive.
>
Correct; this patch is wrong, and would cause a memory leak. This is a
false positive in clang, and a known issue that clang is in general
unable to see that g_autofree variables are used, sometimes for their
intentional side effects such as easier memory cleanup as done here.
I suspect that the definition of g_autofree already uses
__attribute__((unused)) to work around clang's oddities, which means
I'm not sure how to silence clang on this one.
--
Eric Blake, Principal Software Engineer
Red Hat, Inc. +1-919-301-3266
Virtualization: qemu.org | libvirt.org
