04-22-2021 06:42 AM
We are working on the wireless-xorif design. We debugged s0_eth_axis and m000_data_axis ports. Hardware debug shows that there are no packets on m000_data_axis even if ORAN C and U plane packets received on time. The advance time matches with DU. What are the possible reasons of this problem?
The statistics of front-haul interface is something like that:
root@zcu102_om5_exs:~# xorif-app -i -c "get fhi_stats 0" init => status = 0 get fhi_stats 0 => status = 0 total_rx_good_pkt_cnt = 456955 total_rx_bad_pkt_cnt = 0 total_rx_bad_fcs_cnt = 0 total_rx_bit_rate = 652387904 oran_rx_bit_rate = 643084800 oran_rx_total = 456241 oran_rx_on_time = 5629 oran_rx_early = 221717 oran_rx_late = 227360 oran_rx_total_c = 15733 oran_rx_on_time_c = 2493 oran_rx_early_c = 5347 oran_rx_late_c = 7840 oran_rx_corrupt = 0 oran_rx_error_drop = 0 oran_tx_total = 0 oran_tx_total_c = 0 offset_earliest_u_pkt = 70 offset_earliest_c_pkt = 70
04-22-2021 09:09 AM
04-23-2021 01:13 AM - edited 04-23-2021 01:16 AM
I just checked your IP GUI settings. the CONFIG.Xran_Max_Dl_Ctrl_1kWords is not enough. Buffer number is calculated by (Section per CC) (CC number) (Symbol number)/1024, and round up to a integer
Based on your GUI setting 275*1*29/1024 = 7.78，so you need to set it to 8 in IP GUI, but you set it to 5, so buffer is not enough.
For more details, please refer to page 24-29 (Memory Structure) of PG370, and check if other buffer values are large enough.
05-11-2021 03:52 AM
I just realized the PRB numbers are different, so pcap should be okay.
From the get fhi_stats 0 result, following offset values are way too big. It seems the packets are not synced properly.
offset_earliest_u_pkt = 70
offset_earliest_c_pkt = 70
You probably need to re-evaluate the packet arriving time.
Can you refer to line 221 and onwards of https://github.com/Xilinx/wireless-xorif/blob/master/scripts/xil_vivado_build.tcl? and add debug signals to ILA
Please observe how header are missed in m0_packet_in_window, we can get more hints from that.
05-18-2021 04:02 AM
I had a quick look at the file, the app_head_valid looks a bit strange.
I would expect it asserts once for every symbol
Also can you export ILA data, so I can import it at my end, and see ILA waveform?