No output when XMOS used as listener

If you have a simple question and just want an answer.
cchandini
Member++
Posts: 28
Joined: Fri Jul 29, 2016 8:25 am

Post by cchandini »

Hi larry, I am not able to find the cause of why 3rd party AVB talker -> XMOS AVB listener is not working. I checked the wireshark logs and couldn't find any mistake. I can send you wireshark log I had captured. Can you look into it and let me know if I am doing anything incorrect? I have been trying to debug this problem since past 3-4 weeks, but couldn't find anything. It's a little urgent. Earliest help would be appreciated! Thanks!


User avatar
larry
Respected Member
Posts: 275
Joined: Fri Mar 12, 2010 6:03 pm

Post by larry »

Wireshark log would be useful to see. Console output from the XMOS endpoint and any information you can get from the third party endpoint and switch command line would help too.
cchandini
Member++
Posts: 28
Joined: Fri Jul 29, 2016 8:25 am

Post by cchandini »

I can share the wireshark logs. Currently, I am not able to get console output from XMOS endpoint since I don't have firmware v5.0 binary files with me. If I use xrun --xscope app_avb_lc_demo.xe, I will need "app_avb_lc_demo.xe" binary file, which I dont have for firmware v5.0. Is there any other way of getting the console output?
cchandini
Member++
Posts: 28
Joined: Fri Jul 29, 2016 8:25 am

Post by cchandini »

Current Setup - 3rd party endpoint (talker) -> Switch <- XMOS Endpoint (both talker and listener running). Wireshark logs are captured using a 3rd port on the switch. Currently, mrp and ptp running on 3rd party. ptp does not actually generate timestamps, hence, the timestamp valid field in 1722 packets sent from 3rd party endpoint is set to 'False'.

1722 packets sent from 3rd party endpoint have VAN ID = 2 and VLAN Priority = 3 (tried to keep it same as XMOS Endpoint).
Source mac address - 00:22:97:00:43:BF is used for 3rd party endpoint. I have got 3 XMOS AVB endpoints. This mac address is of one of the XMOS AVB Endpoints. This was done to keep packets sent from 3rd party endpoint as similar as possible to what an XMOS Endpoint send to another XMOS Endpoint (thought might help in debugging the problem).
cchandini
Member++
Posts: 28
Joined: Fri Jul 29, 2016 8:25 am

Post by cchandini »

I am not able to attach any files. Tried multiple times. Sharing google drive link for the logs -
cchandini
Member++
Posts: 28
Joined: Fri Jul 29, 2016 8:25 am

Post by cchandini »

Also, right now I observed one thing. I am sending 2 channel data to XMOS endpoint as 1722 packets. And no ptp daemon is running on 3rd party endpoint. Whenever I start talker on 3rd party endpoint, some sound is observed on channel 1 & 2 output of XMOS endpoint which looks to me like an underrun/overrun issue at XMOS endpoint end. This sound stops after sometime (varies everytime from around 2s - 30s). With my talker running on 3rd party endpoint, if reset XMOS endpoint (by pressing 'reset' button on XMOS endpoint), after few seconds, the underrun/overrun sound is heard for few seconds (variable duration) and then dies off. Can someone explain me what kind of behaviour is expected from XMOS endpoint during cases of underrun/overrun? I think there might be some clock synchronization issue between 3rd party endpoint and XMOS endpoint. Please correct me if I am wrong.

Thanks,
Chandni
cchandini
Member++
Posts: 28
Joined: Fri Jul 29, 2016 8:25 am

Post by cchandini »

Adding to above mentioned observation, firmware v6.1 source code says that the media clock frequency is 48 kHz. 3rd party endpoint also has a media clock frequency of 48 kHz. When I connect XMOS endpoint (talker, contains firmware v5.0) -> Switch -> 3rd party endpoint (listener), and capture packets using wireshark, to calculate media clock frequency of XMOS endpoint wrt ptp clock, I calculate the difference between timestamp values of 2 consecutive packets sent by XMOS and then since each packet contains 6 samples, divide this difference by 6 and then take the inverse to calculate approximate sampling frequency (media clock frequency) of XMOS endpoint. This value, when calculated at different instances comes out to be around 36 kHz. Is this the expected/correct value?

When I connect XMOS Endpoint (talker) -> 3rd party endpoint (listener), I do not hear any glitches i.e. underrun/overrun issues at 3rd party endpoint. The output audio is fine. My 3rd party endpoint has a media clock frequency of 48 kHz.
Post Reply