On 05/03/2013 08:53 AM, Michal Privoznik wrote:
--- src/locking/lock_daemon.c | 30 +++++++++++----------------- src/locking/lock_daemon_config.c | 12 +++++------ src/locking/lock_daemon_dispatch.c | 6 ++---- src/locking/lock_driver_lockd.c | 41 ++++++++++++++++---------------------- src/locking/lock_driver_sanlock.c | 12 ++++------- src/locking/lock_manager.c | 4 +--- 6 files changed, 41 insertions(+), 64 deletions(-)
diff --git a/src/locking/lock_daemon.c b/src/locking/lock_daemon.c
@@ -1221,13 +1217,13 @@ int main(int argc, char **argv) {
case 'p': VIR_FREE(pid_file); - if (!(pid_file = strdup(optarg))) + if (VIR_STRDUP(pid_file, optarg) < 0) exit(EXIT_FAILURE);
Hmm, for the similar patch to libvirtd earlier in the series, you used VIR_STRDUP_QUIET and an explicit VIR_WARN when inside main() on the grounds that it's too early to rely on normal error handling being up. Do we need to do that here as well?
break;
case 'f': VIR_FREE(remote_config_file); - if (!(remote_config_file = strdup(optarg))) + if (VIR_STRDUP(remote_config_file, optarg) < 0) exit(EXIT_FAILURE);
Again, a case where immediately calling exit() may eat the error message, compared to VIR_STRDUP_QUIET/VIR_WARN. ACK once you consider that point. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org