From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:37444) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1gyI52-0000Uc-BN for qemu-devel@nongnu.org; Mon, 25 Feb 2019 10:22:11 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1gyI4u-00014A-P0 for qemu-devel@nongnu.org; Mon, 25 Feb 2019 10:22:04 -0500 From: Kevin Wolf Date: Mon, 25 Feb 2019 16:20:04 +0100 Message-Id: <20190225152053.15976-23-kwolf@redhat.com> In-Reply-To: <20190225152053.15976-1-kwolf@redhat.com> References: <20190225152053.15976-1-kwolf@redhat.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Subject: [Qemu-devel] [PULL 22/71] block: improve should_update_child List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-block@nongnu.org Cc: kwolf@redhat.com, peter.maydell@linaro.org, qemu-devel@nongnu.org From: Vladimir Sementsov-Ogievskiy As it already said in the comment, we don't want to create loops in parent->child relations. So, when we try to append @to to @c, we should check that @c is not in @to children subtree, and we should check it recursively, not only the first level. The patch provides BFS-based search, to check the relations. This is needed for further fleecing-hook filter usage: we need to append it to source, when the hook is already a parent of target, and source may be in a backing chain of target (fleecing-scheme). So, on appending, the hook should not became a child (direct or through children subtree) of the target. Signed-off-by: Vladimir Sementsov-Ogievskiy Signed-off-by: Kevin Wolf --- block.c | 43 +++++++++++++++++++++++++++++++++++++------ 1 file changed, 37 insertions(+), 6 deletions(-) diff --git a/block.c b/block.c index 375a216f76..bb4bf1237c 100644 --- a/block.c +++ b/block.c @@ -3542,7 +3542,9 @@ void bdrv_close_all(void) =20 static bool should_update_child(BdrvChild *c, BlockDriverState *to) { - BdrvChild *to_c; + GQueue *queue; + GHashTable *found; + bool ret; =20 if (c->role->stay_at_node) { return false; @@ -3578,14 +3580,43 @@ static bool should_update_child(BdrvChild *c, Blo= ckDriverState *to) * if A is a child of B, that means we cannot replace A by B there * because that would create a loop. Silently detaching A from B * is also not really an option. So overall just leaving A in - * place there is the most sensible choice. */ - QLIST_FOREACH(to_c, &to->children, next) { - if (to_c =3D=3D c) { - return false; + * place there is the most sensible choice. + * + * We would also create a loop in any cases where @c is only + * indirectly referenced by @to. Prevent this by returning false + * if @c is found (by breadth-first search) anywhere in the whole + * subtree of @to. + */ + + ret =3D true; + found =3D g_hash_table_new(NULL, NULL); + g_hash_table_add(found, to); + queue =3D g_queue_new(); + g_queue_push_tail(queue, to); + + while (!g_queue_is_empty(queue)) { + BlockDriverState *v =3D g_queue_pop_head(queue); + BdrvChild *c2; + + QLIST_FOREACH(c2, &v->children, next) { + if (c2 =3D=3D c) { + ret =3D false; + break; + } + + if (g_hash_table_contains(found, c2->bs)) { + continue; + } + + g_queue_push_tail(queue, c2->bs); + g_hash_table_add(found, c2->bs); } } =20 - return true; + g_queue_free(queue); + g_hash_table_destroy(found); + + return ret; } =20 void bdrv_replace_node(BlockDriverState *from, BlockDriverState *to, --=20 2.20.1