Hi guys,
Seeing the https://github.com/kamailio/kamailio/pull/1144 about fixing a memory leak in the tm module due to incorrectly using FL_SHM_CLONE when freing path_vec memory, isn't there a case that a similar patch should be applied to corex/corex_lib.c ?
Inside corex_append_branch(..) function:
/* if this is a cloned message, don't free the path vector as it was copied into shm memory and will be freed as contiguous block*/ if (!(msg->msg_flags&FL_SHM_CLONE)) { if(msg->path_vec.s!=0) pkg_free(msg->path_vec.s); msg->path_vec.s = 0; msg->path_vec.len = 0; }
Maybe we should switch to:
if(msg->path_vec.s!=0) pkg_free(msg->path_vec.s); msg->path_vec.s = 0; msg->path_vec.len = 0;
If you agree this is the case, I will prepare a pull request.
Thank you, Lucian