We recently received two crashes from one user in pjsip_regc_send, during unregister. It seems it crashes on this line: pj_assert(regc->current_op == REGC_IDLE); I assume this would be related to the same bug, then? Any suggestions how to handle this? Oyatel:0:in `0x0001ffd1 HTHandleSignal + 44' libSystem.B.dylib:1:in `0x33bd3ddf _sigtramp + 42' libSystem.B.dylib:2:in `0x33bd52cb kill + 10' libSystem.B.dylib:3:in `0x33bd52bd raise + 16' libSystem.B.dylib:4:in `0x33be9d79 abort + 56' libSystem.B.dylib:5:in `0x33bd7987 __assert_rtn + 158' Oyatel:6:in `0x00105898 pjsip_regc_send + 492' Oyatel:7:in `0x0010c804 pjsua_acc_set_registration + 244' Oyatel:8:in `0x0001ce41 -[PJSIPHandler unregister] + 176' Oyatel:9:in `0x0001d09f -[PJSIPHandler restart] + 170' Oyatel:10:in `0x0001d349 -[PJSIPHandler reachabilityChanged:] + 224' > Hi, > > I have found what I think is a bug in the sip_regc layer. > > When you encounter an error (different from 401/407/423), the current_op > variable is not reset to REGC_IDLE, preventing to others REGISTER requests > (sip_regc_send checks for current_op being REGC_IDLE). > > This occurs for example when your AOR is already registered for the maximum > number of times allowed, you receive a 403 error, but you can not send an > unregister_all command. > > Philippe Leuba -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20110324/46919782/attachment.html>
