On 02.05.23 20:55, Julien Grall wrote:
Hi Juergen,On 05/04/2023 08:03, Juergen Gross wrote:Add the node accounting to the accounting information buffering in order to avoid having to undo it in case of failure. Signed-off-by: Juergen Gross <jgr...@suse.com> --- tools/xenstore/xenstored_core.c | 21 ++------------------- tools/xenstore/xenstored_domain.h | 4 ++-- 2 files changed, 4 insertions(+), 21 deletions(-) diff --git a/tools/xenstore/xenstored_core.c b/tools/xenstore/xenstored_core.c index 84335f5f3d..92a40ccf3f 100644 --- a/tools/xenstore/xenstored_core.c +++ b/tools/xenstore/xenstored_core.c@@ -1452,7 +1452,6 @@ static void destroy_node_rm(struct connection *conn, struct node *node)static int destroy_node(struct connection *conn, struct node *node) { destroy_node_rm(conn, node); - domain_nbentry_dec(conn, get_node_owner(node)); /* * It is not possible to easily revert the changes in a transaction.@@ -1797,27 +1796,11 @@ static int do_set_perms(const void *ctx, struct connection *conn,old_perms = node->perms; domain_nbentry_dec(conn, get_node_owner(node));IIRC, we originally said that domain_nbentry_dec() could never fail in a non-transaction case. But with your current rework, the function can now fail because of an allocation failure.
How would that be possible to happen? domain_nbentry_dec() can only be called if a node is being owned by an already known domain. So allocation is impossible to happen, as this would be a major error in xenstored.
Therefore, shouldn't we now check the error? (Possibly in a patch beforehand).
I don't think so. I can add a comment if you want. Juergen
OpenPGP_0xB0DE9DD628BF132F.asc
Description: OpenPGP public key
OpenPGP_signature
Description: OpenPGP digital signature
