<div class="gmail_quote">Hi, <div><br></div><div>I'm using the latest stable version of libpri 1.4.11.5 ( also reproducible on 1.4.11.4 ), and am noticing the following Issue in ALL of my hangups ( of all duration - 0, 30, 60 seconds and higher )</div>
<div><br></div><div>1. Say, the user ( leg A ) - Application sends a DISCONNECT to the network ( Leg B ). I'm noticing that libpri destroys the call after receiving a DISCONNECT. </div><div><br></div><div>2. Next, A receives RELEASE from Leg B, with a call reference. However, this call reference is not there anymore, and has to be created. It gets created afresh - but as a newcall. </div>
<div> So it has c->newcall = 1.</div><div><br></div><div>3. The post_handle_931_message function sees this c->newcall = 1, and calls q931_release_complete(ctrl, c, Q931_INVALID_CALL_REFERENCE), which leads to RELEASE COMPLETE always being delivered to LEG B with the message Invalid Call Reference (81).</div>
<div><br></div><div><br></div><div>Is this a libpri bug or is there something wrong at my end ? The PRI Q.931 trace of the aforementioned call is at <a href="http://pastebin.com/VQ04hhgR" target="_blank">http://pastebin.com/VQ04hhgR</a> ( the relevant lines are highlighted towards the end ). </div>
<div><br></div><div>Also, according to Q.931, RELEASE COMPLETE with an invalid references are to be ignored and nothing is to be done - so could this lead to stalling of calls for a further few seconds ? ( Leading to a slightly higher bill ) ? Or is the billing meant to be stopped with the DISCONNECT message ( someone helped with this on IRC ) itself, wanted to make sure.</div>
<div><br></div><div><br></div><div>Sending this without attachment this time..</div><div><br></div><div>Thanks, </div><div>Abhinav </div><font color="#888888"><div><br></div>
</font></div><br>