Hi Allen, 2011/3/25 Allen <alleng...@gmail.com>: > Thanks a lot for your reply. > > I have tried the --no-nat setting, the results are the same. > > Actually the log information I provided in my original email for Visual > studio 2003 release version is already the result with BOTH --no-nat AND > --no-proxy setting, otherwise it halts even earlier with the following > log information:
OK. The problem is, test_nua test cases may suffer from some nasty timing issues, but the newer check-based tests do not work at all on Windows. > -------------------- > TEST NUA-2.6.1: REGISTER b to c > test_nua: testing b_call->nh = nua_handle(b->nua, b_call, (tag_type_t)0, > (tag_value_t)0) > test_nua: ok: (b_call->nh = nua_handle(b->nua, b_call, (tag_type_t)0, > (tag_value_t)0)) > nta: REGISTER (10166550): Connection refused (10061) with > tcp/[192.168.1.5]:1685 > -------------------- > > I have test these release version of test_nua.exe on XP SP3, Vista > business, and Windows 7 Home Preminum, all all have same results as the > log shown in my original email. All these three operation systems have > the latest patches installed. > > I have checked the su_init() function, it's the su_init(void) at the > line 304 of su.c, it DOES include the code to load the _DisconnectEx > function pointer, and I have checked the log information, the loading > function call didn't fail. OK, so that is not an issue. > I am confused why the release version runs differently from debug > version, the debug version just passes all the test cases. The optimization has a tendency to change the work flow slightly and exposing weaknesses and bugs not present in unoptimized versions. Of course, the compilers always have bugs, most of them show up when optimizing. I wonder if the latest MSVCs allow you to debug optimized versions. If so, please try to increase optimization level in debug version and try to see if that helps. > According to > my limited experiences, it might caused by not initializing some > varibles before using them since the debug version always set the > varibles to zero when they are declared. Compiler should warn about not initializing some variables. The project files disable quite a many warnings on MSVC, please see if enabling them reveals something. > Would you please give me some more advices to dig into this issue? Thanks! First, try test_nua with -s option. If it works with that, the issue is most probably in the test suite. You can also run test_nua with various logging options, using -e with --log=a, --log=b or --log=c might reveal something. I'm mostly interested in the logs from the runs that stop at test case 9.1.4 ... -- Pekka.Pessi mail at nokia.com ------------------------------------------------------------------------------ Enable your software for Intel(R) Active Management Technology to meet the growing manageability and security demands of your customers. Businesses are taking advantage of Intel(R) vPro (TM) technology - will your software be a part of the solution? Download the Intel(R) Manageability Checker today! http://p.sf.net/sfu/intel-dev2devmar _______________________________________________ Sofia-sip-devel mailing list Sofia-sip-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sofia-sip-devel
