Understood, the way i was doing it before behaved in the latter method you described. I was just trying to get it to work in another way with only one dispatcher.
But everything should be alright doing it the way i have been where the dispatcher is in the same thread as event structures are setup in. -brian On Fri, Mar 6, 2009 at 11:59 AM, Matt Pearson <404emailnotfo...@gmail.com> wrote: > On Fri, Mar 6, 2009 at 09:16, Brian <ping...@gmail.com> wrote: >> Thanks Guillaume for the feedback. Unfortunately malloc/calloc method >> didn't work for me. I think I will just go back to how it was. Does >> anyone know if there is an inherent flaw, perhaps, with doing the >> event_init()/event_dispatch() within a posix thread? >> >> -brian > > There are two things terribly wrong with what you're trying to do > here. First, you're having libevent access freed memory by giving it > pointers to freed stack variables as Guillaume said. The event struct > is added directly to libevent's internal structures; it is not copied > in any way. You *must* ensure that event structs do not go out of > scope when they're still being used. > > Second, you're calling event_add from one thread while event_dispatch > is running in another. Things may have changed, but last I checked > libevent was not thread-safe in this manner and only supported having > multiple event queues, one for each thread, with add/del being done > only from the same thread event_base_dispatch was running in. > > Matt > _______________________________________________ Libevent-users mailing list Libevent-users@monkey.org http://monkeymail.org/mailman/listinfo/libevent-users
