Thanks a lot Nanang. That solved the problem. Case closed :D On Fri, Aug 22, 2008 at 5:06 PM, Nanang Izzuddin <nanang at pjsip.org> wrote: > Hi Joao, > > I just tried to reproduce the problem and I think I found the problem > root/bug, which is miscalculation of conf port buffer size. Please try > the solution by updating your source to the latest SVN or just apply > the modification of conference.c > (http://trac.pjsip.org/repos/changeset/2233). And please give feedback > if the problem still occured. > > Thanks for reporting the problem. > > Btw, I just realize that you build your own app based on pjsua-lib, I > was asking about 'cl' just to make sure the conf ports settings. > > Regards, > nanang > > > On 22/08/2008, Jo?o C?sar <jpcesar at gmail.com> wrote: > > > > Hi Nanang, > > > > Im giving you more details: > > > > >> Those values (256 & 352) seems odd, what is the value of > > >> 'memory_port'->info.samples_per_frame? And also > > conference bridge's > > >> samples_per_frame value? Could you send along the result of 'cl' > > >> command in pjsua prompt? > > > > memory_port->info.samples_per_frame = 256 > > > > this value is achieved by frame_size / 2 / channel_count > > frame_size is being sent with 512, the 2 is dividing the frame size by 2 > (16 > > bit per samples, 2 bytes each sample) and channel_count = 1 > > > > regarding cl command in pjsua, we are running pjsualib on our own > project, i > > don't think i can use that here, right? > > > > >> What is 512 stand for here? > > > > this is passed to the create_memory_port function as explained above and > its > > used to set the port->info.bytes_per_frame value. its the frame size of > > audio data we receive on application. We receive pcm audio frames. > > > > >> Actually I am interested to know what the value of memory > > >> port->info.samples_per_frame and conference's samples_per_frame. > > > > port->info.samples_per_frame = 256 > > conference samples_per_frame i think it has the value 320 as this is the > > value actually on pjsua_var.mconf_cfg.samples_per_frame > > used when we call pjmedia_snd_port_create_player. > > > > > > Thanks in advance > > > > Joao > > > > > > > > On Fri, Aug 22, 2008 at 11:15 AM, Nanang Izzuddin <nanang at pjsip.org> > wrote: > > > > Hi Joao, > > > > Please see inline comments.. > > > > > > 2008/8/22 Jo?o C?sar <jpcesar at gmail.com>: > > > Hi Nanang, > > > > > > these are the last values before the assertion fails: > > > > > > 256 <= 320 > > > 160 <= 160 > > > 352 <= 320 > > > > Those values (256 & 352) seems odd, what is the value of > > 'memory_port'->info.samples_per_frame? And also > > conference bridge's > > samples_per_frame value? Could you send along the result of 'cl' > > command in pjsua prompt? > > > > > > > > Assertion failed: cport->rx_buf_count <= cport->rx_buf_cap, file > > > e:\development\ > > > > > jcesar\pjproject-0.9.0\pjmedia\src\pjmedia\conference.c, > > line 1404 > > > > > >> Which codec and which clock rates are you using for > > >> the sound device/conference and conference ports used such as file > > >>ports if any. And, sometime reclean the whole project may resolve a > > >> problem occured after changing library. > > > > > > we are working with the no sound device option since we only use > PJSIP > > to > > > deliver the packets to a third party application via SIP. we > receive > > audio > > > packets from a socket, if we dont have a call for that client we > make > > the > > > call, else we just feed the memory port with more samples. > > > > > > we are making the call like this: > > > > > > sampleRate = 8000; > > > channels = 1; > > > number_frames = 960; > > > channel_count = 1; > > > > > > // create a media port to generate samples. > > > status = create_memory_port( pool, sampleRate, 512, number_frames, > > > channel_count, > > &callData[sessionId].memory_port_phone_out); > > > > > > > What is 512 stand for here? > > > > > // add port to conference > > > status = pjsua_conf_add_port(pool, > > > callData[sessionId].memory_port_phone_out, > > &callData[sessionId].port_id); > > > if (status != PJ_SUCCESS) error_exit("Error adding memory port", > > status); > > > > > > status = pjsua_call_make_call(acc_id,uri, 0, > > &callData[sessionId].port_id, > > > NULL,NULL); > > > if (status != PJ_SUCCESS) error_exit("Error making call", status); > > > > > > // assign the callid to the current call in callData > > > callData[sessionId].call_id = callId; > > > > > > // increment callId for next call > > > callId++; > > > > > > //creates the sound port > > > status = > > pjmedia_snd_port_create_player(pjsua_var.pool, > > > 0, > > > > > > 8000, > > > > > > pjsua_var.mconf_cfg.channel_count, > > > > > > pjsua_var.mconf_cfg.samples_per_frame, > > > > > > pjsua_var.mconf_cfg.bits_per_sample, > > > 0, > > > > > &sound_port); > > > > > > status = pjmedia_snd_port_connect(sound_port, > > > pjmedia_conf_get_master_port(pjsua_var.mconf)); > > > > > > > > > Each time we receive an audio packet we put that frame into > > memory_port: > > > > > > // put frame > > > > > memory_put_frame(callData[sessionId].memory_port_phone_out, > > &pjframe ); > > > > > > > > > and these are the options on the create memory port function: > > > > > > > > > /* Fill in port info. */ > > > port->info.bits_per_sample = 16; > > > port->info.channel_count = channel_count; > > > port->info.encoding_name = pj_str("pcm"); > > > port->info.has_info = 1; > > > port->info.name = pj_str("memory playback"); > > > port->info.need_info = 0; > > > port->info.pt = 0xFF; > > > port->info.clock_rate = sampling_rate; > > > port->info.samples_per_frame = frame_size / 2 / channel_count ; > > > port->info.bytes_per_frame = frame_size; > > > port->info.type = PJMEDIA_TYPE_AUDIO; > > > > > > > Couldn't find any suspect from the snippet code since it shows mostly > > variables without knowing the 'important' values. Actually I am > > interested to know what the value of memory > > port->info.samples_per_frame and conference's samples_per_frame. > > > > > > > > This was working fine on 0.8. > > > > > > > Conf port's buffer size (rx_buf_cap) calculation has been updated in > > 0.9 due to multichannel/stereo support in conference (and also IIRC > it > > seems rx_buf_cap in 0.8 is a bit over calculated, perhaps that's why > > it was working fine on 0.8, miscalculation was tolerated). > > > > Regards, > > nanang > > > > _______________________________________________ > > Visit our blog: http://blog.pjsip.org > > > > pjsip mailing list > > pjsip at lists.pjsip.org > > > > http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org > > > > > > > > > > -- > > Joao Cesar > > msn: jpcesar at gmail.com > > gtalk: jpcesar at gmail.com > > icq: 13790802 > > > > > > _______________________________________________ > > Visit our blog: http://blog.pjsip.org > > > > pjsip mailing list > > pjsip at lists.pjsip.org > > http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org > > > > > > _______________________________________________ > Visit our blog: http://blog.pjsip.org > > pjsip mailing list > pjsip at lists.pjsip.org > http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org > -- Joao Cesar msn: jpcesar at gmail.com gtalk: jpcesar at gmail.com icq: 13790802 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20080828/c4d43933/attachment.html
