[USRP-users] Fifo ctrl timed out looking for acks

Josh Blum josh at ettus.com
Fri Feb 15 12:42:35 EST 2013



On 02/14/2013 01:49 PM, Paul Pruitt wrote:
> I'm getting this exception, thrown from
> usrp2_fifo_ctrl_impl::wait_for_acks, sometimes after several hours of
> operation.
> 
> The device is an N210 with UHD version 003:005.000-26-gb65a3924 on a laptop
> under CentOS 6.3 64-bit. I'm collecting 100 ms duration streams from both
> channels at 10 MSPS (each) every 250 ms.
> 
> As an experiment, I increased ACK_TIMEOUT from 0.5 s to 5.5 s. The
> exception still occurred. No ethernet errors or dropped packets are
> reported by ifconfig. I am running as root and verified the threads are
> running at realtime priority with htop. The frame sizes, recv_frame_size
> and send_frame_size, are set to 4095 bytes. Kernel rx/tx socket buffers are
> 5MB.
> 
> Any idea what might be causing this?
> Is it safe to catch the exception and keep going? If I restart my program
> after the exception, the USRP works fine.
> 

Not sure, the packet is either lost or stuck. My guess is lost. In which
case, the timeout cant help.

I think we can make this recover better. Can you try this diff?
http://pastebin.com/ZMJqvWs5

If you catch the error, the future control requests/responses should
continue to work. Assuming we have lost a packet and something isnt "stuck"

-josh

> Thanks,
> Paul
> 
> 
> 
> _______________________________________________
> USRP-users mailing list
> USRP-users at lists.ettus.com
> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
> 




More information about the USRP-users mailing list