Hi everyone, I've been using PJSUA-LIB for development of an application that handles/generates a number of simultaneous calls. (I know, PJSUA-LIB might not be the best choice, but it was the easiest way to get something up and running quickly.) Everything seemed to work just fine when testing against regular numbers, but now I've encountered something really odd. As the subject says, it looks like as if PJSUA-LIB enters an infite loop when calling pjsua_call_make_call() after receiving a 480 Temporary Unavailable response from the carrier. CPU usage is 100% in this case. When I send a CANCEL before the carrier has a chance to return the 480, everything is allright. When looking at the backtraces [1], it seems that the previous call isn't cleaned up properly (thread 3), causing the pjsua_call_make_call() to block (thread 11). When VAD is enabled and this situation occurs, to log keeps being filled with messages from the VAD, so probably the RTP stream isn't closed either. The backtrace seems to support this (thread 4). Other (possibly) relevant info: - all PJSUA-LIB calls are done from a single thread - the threads not shown in the backtraces don't use PJSUA-LIB at all - the machine runs FreeBSD 7.1 - I've inluded the log at level 5, but it's cut short when the problem occurs [2] Anyone got any suggestions on how to fix this or why this is occuring? I'm probably forgetting something, but just can't figure it out! Kind regards, Maarten [1] http://www.privateer-software.nl/pjsip/backtraces.txt [2] http://www.privateer-software.nl/pjsip/log.txt -- Privateer Software Development (www.privateer-software.nl) * web usability, web accessibility, web development * cross-platform software development
