Simon Morlat @ 2005-08-24 (Wednesday), 23:27 (+0200) > Hello, Hi there, > For now I have no idea. I'll try reproduce the bug with valgrind.
That's an idea I never though of, and silly enough it still chrashes for me when running it through valgrind. :/ I'm not an total expert on valgrind either, but my expection was that a memory access error shouldn't result in a chrash... Spending a few minutes actually stepping through with gdb made me find something suspicious. Applying the attached patch fixes the problem for me, but I guess it's time to do an audit if more things are coded the way that they assume g_list_free() should NULL the pointer. -- /Martin
--- linphone-1.0.1/coreapi/authentication.c 2005-03-10 17:59:41.000000000 +0100 +++ linphone-1.0.1-nocrash/coreapi/authentication.c 2005-08-25 15:37:23.974222897 +0200 @@ -185,6 +185,7 @@ linphone_auth_info_write_config(lc->config,NULL,i); } g_list_free(lc->auth_info); + lc->auth_info = NULL; } gboolean linphone_core_find_or_ask_for_auth_info(LinphoneCore *lc,const char *username,const char* realm, gboolean force_ask)