ERROR_CODE_OVERFLOW (Overflow) with B210

[sayaz...@gmail.com]

Hello,

Exp Link: https://www.powderwireless.net/status.php?uuid=2c5c85fe-f838-11ee-9f39-e4434b2381fc

I am using  indoor OTA lab B210s (working on 5G-NR).

I issue the below command to start gNB (B210 # ota-nuc 3), the node throws error (attached)

sudo numactl --membind=0 --cpubind=0 /var/tmp/oairan/cmake_targets/ran_build/build/nr-softmodem -E -O /var/tmp/etc/oai/gnb.sa.band46.fr1.106PRB.usrpb210.conf --gNBs.[0].min_rxtxtime 6 --sa --MACRLCs.[0].dl_max_mcs 28 --tune-offset 23040000

Issuing the same command on any other nodes (e.g. ota-nuc 2/4) works just fine (no error).

I've taken out nuc 3 from current experiment. But wanted to know what should be done if it throws same error in future. Seems to be an issue between b210-NUC ?

Thank you,

Ayaz

[Kirk Webb]

Taking it out of your experiment makes it hard to troubleshoot; please
leave it in place if you have an issue in the future. It's possible
that ota-nuc3 has some incorrect BIOS settings, or the B210 is
erroneously connecting via USB2 (vs. 3).

> --
> You received this message because you are subscribed to the Google Groups "Powder Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to powder-users...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/561f1f0f-1528-4152-bd12-dd208e6267bbn%40googlegroups.com.

[sayaz...@gmail.com]

Hi Kirk / Team,

I'm having similar error as before, as I recreated the experiment.

Experiment link: https://www.powderwireless.net/status.php?uuid=250a7024-2ce1-11ef-9f39-e4434b2381fc

Below command works fine on nuc-4, but throws error on nuc-3:

sudo numactl --membind=0 --cpubind=0   /var/tmp/oairan/cmake_targets/ran_build/build/nr-uesoftmodem -E -O /var/tmp/etc/oai/ue.conf  -r 106  -C 5754720000  --usrp-args "clock_source=internal,type=b200"  --band 46  --numerology 1  --ue-fo-compensation  --ue-txgain 0   --ue-rxgain 120   --nokrnmod   --dlsch-parallel 4   --sa --ssb 516

Error: 

[HW]   [recv] received 22811 samples out of 23040
[HW]   Time: 3.001 s
ERROR_CODE_OVERFLOW (Overflow)

Assertion (UE->frame_parms.get_samples_per_slot(slot,&UE->frame_parms) == UE->rfdevice.trx_read_func(&UE->rfdevice, timestamp, rxp, UE->frame_parms.get_samples_per_slot(slot,&UE->frame_parms), UE->frame_parms.nb_antennas_rx)) failed!
In readFrame() /var/tmp/oairan/executables/nr-ue.c:687

Can you please let me know what steps to solve this.

Thank you,

Ayaz

[Kirk Webb]

Are you running the same code, with the same configuration on both
ota-nuc3 and ota-nuc4? Segmentation faults are being logged for libuhd
in the syslog:

Jun 17 16:17:03 localhost kernel: [ 193.815456] UEthread[7614]:
segfault at 56071adeb244 ip 00007f569ad65113 sp 00007f569b9f5e70 error
6 in libuhd.so.4.6.0[7f569a7aa000+7bd000]
Jun 17 16:17:03 localhost kernel: [ 193.815468] Code: 7c 24 18 48 89
c6 ff 50 18 48 8b 84 24 98 00 00 00 48 c7 84 24 98 00 00 00 00 00 00
00 48 8b 7d 00 48 89 45 00 48 85 ff 74 0d <f0> 83 6f 08 01 75 06 48 8b
07 ff 50 10 48 8b bc 24 98 00 00 00 48
Jun 17 16:27:19 localhost kernel: [ 810.276686] UEthread[7758]:
segfault at 7f5b45d9c7b2 ip 00007f5cb499a113 sp 00007f5cb562ae70 error
6 in libuhd.so.4.6.0[7f5cb43df000+7bd000]
Jun 17 16:27:19 localhost kernel: [ 810.276699] Code: 7c 24 18 48 89
c6 ff 50 18 48 8b 84 24 98 00 00 00 48 c7 84 24 98 00 00 00 00 00 00
00 48 8b 7d 00 48 89 45 00 48 85 ff 74 0d <f0> 83 6f 08 01 75 06 48 8b
07 ff 50 10 48 8b bc 24 98 00 00 00 48

It also looks like the USRP device may have disconnected from the USB
bus, then immediately reconnected at one point. I have power cycled
ota-nuc4. Please try again and see if that has resolved the issue.

-Kirk

> To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/0fc8ff7e-72f6-409e-aa93-67dd3cbaef82n%40googlegroups.com.

[sayaz...@gmail.com]

Hi Kirk,

No, it does not solve the issue. 

Yes, the configuration and commands i am running are exactly same for nuc-3 and nuc-4. Using both as an UE.

Moreover, I tried to use nuc-3 as gNB. And it throws similar error.

The only difference is the USB port when compared to other nuc's. Both nuc-2 and nuc-4 are connected on USB 3, while nuc-3 is on USB 2.

Thanks

Ayaz

[Kirk Webb]

It's possible a different version of UHD would solve the problem, but
updating is somewhat complicated. If you are comfortable doing this
yourself, please give it a try. Otherwise, I'll consult with the team
to see if there is an OS image with a newer UHD installed. The other
possibility is that a hardware fault has developed, though we have not
had anyone else report issues that would point to this.

-Kirk

> To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/51868d5f-7620-473a-9f41-715ae4c6f811n%40googlegroups.com.

[Alex Orange]

Ayaz,

     Fixed the USB problem, putting nuc3 back in your reservation.

Alex

To view this discussion on the web visit https://groups.google.com/d/msgid/powder-users/CAFKO4CFgxqzb6LfXc%2B%2BUx0az7q_QD_YekX7pKpEfTg5Q2%3DXFxw%40mail.gmail.com.

[Alex Orange]

P.S. In the future, just leave it in your experiment, would make it less work to fix.