Re: [Discuss-gnuradio] USRP underrun

[Lee Patton]

On Tue, 2006-08-01 at 17:35 -0700, Eric Blossom wrote:
> On Tue, Aug 01, 2006 at 08:26:58PM -0400, Lee Patton wrote:
> > I see a single USRP underrun (uU) about four out of five times that my
> > application starts.  My transmit work() routine couldn't be simpler: a
> > for-loop that moves data from a local buffer to the output buffer.  So,
> > I don't think I can scale it back or anything.
> > 
> > Is there anything I can do to avoid these uU at startup?  Does anyone
> > know what might be causing them? Unfortunately my application can't
> > tolerate any under/overruns.
> > 
> > Thanks,
> >  -Lee
> 
> Have you tried enabling real time scheduling?
> You'll need to be running as root or have CAP_SYS_NICE for it to work.
> 
> See gnuradio-examples/python/gmsk2/tunnel.py
> 
>     # Attempt to enable realtime scheduling
>     r = gr.enable_realtime_scheduling()
>     if r == gr.RT_OK:
>         realtime = True
>     else:
>         realtime = False
>         print "Note: failed to enable realtime scheduling"
> 
> 
> Eric

Thanks, Eric. 

I put that code in my script, and ran as root. Unfortunately, I'm still
seeing uU.  I also messed around with setting block_size and nblocks,
but I don't really know what I'm doing there -- just changing numbers to
see if it works better :)

When I put the above code into my script, I get the following error upon
exit:

Exception exceptions.ReferenceError: 'weakly-referenced object no longer
exists' in <bound method db_flexrf_2400_rx.__del__ of
<gnuradio.db_flexrf.db_flexrf_2400_rx object at 0x40dcbccc>> ignored

Did I incorporate the code properly?

I also tried running my app with "sudo nice -20 make check", but that
didn't fix my problem either.

It is strange that I almost always get one underrun, and it is usually
at the very beginning of my data.

Could my rx block be affecting my tx block?  Again, I'm just
transferring the data in the input buffer into a local buffer.

- Lee