Gr_vmcircbuf_sysv_shm: shmget (1): Invalid argument

GNU Radio
1

./shmget_demo_issue.py -v 1024 -n 2048
gr_vmcircbuf_sysv_shm: shmget (1): Invalid argument
gr_buffer::allocate_buffer: failed to allocate buffer of size 16392 KB
terminate called after throwing an instance of ‘std::bad_alloc’
what(): std::bad_alloc
Aborted

So, I have the latest gnuradio next and ubuntu 10.10. This issue crops
up whenever I try to run the wxgui fftsink with my usrp2 at full rate,
and on several different computers.

I have recreated the problem with a simple python app. No message
queues, no usrp hardware, no graphics, just core gnuradio blocks.
Application attached.

null source -> keep one in n -> null sink

Basically, increasing the N (decimation) to a high enough number causes
this issue consistently. Why should more decimation cause this? Did
something in the tags implementation change this behavior? Any ideas?

Thanks,
-Josh

2

On Thu, Feb 10, 2011 at 2:38 PM, Josh B. [email protected] wrote:

and on several different computers.

Thanks,
-Josh

ACK. Thanks for the report and test case. I have no idea what’s
causing it right now, and I really can’t see how the tags would have
interfered.

What happens if you go back to the master or even the 3.3.0 release
package?

Tom

3

ACK. Thanks for the report and test case. I have no idea what’s
causing it right now, and I really can’t see how the tags would have
interfered.

What happens if you go back to the master or even the 3.3.0 release package?

So the latest master is fine. The stale months old version of next I had
was also fine, but I pulled and forgot to get the hash first.

So, I think there was a regression at some point in the history of next.
Before binary searching for this, any likely candidate feature merges to
try?

-Josh

4

On Thu, Feb 10, 2011 at 11:53 PM, Josh B. [email protected] wrote:

http://gnuradio.org/cgit/gnuradio.git/commit/?h=next&id=649e4387848a5d9273983600be6f35c3e2a88191

Definitely tags related, that change broke it. -josh

Alright, thanks for tracking it down. I’ll see what I can do.

Tom

5

http://gnuradio.org/cgit/gnuradio.git/commit/?h=next&id=649e4387848a5d9273983600be6f35c3e2a88191

Definitely tags related, that change broke it. -josh