script.
Dev PC: Xubuntu 16.04 running in VM Workstation (mentioned because some
python bugs are hypervisor effected only)
Problem: I am getting errors when running flow graphs in GRC after
adding variables to the udp_sink_impl.h file.
I've been struggling all week making changes to an existing project
(https://github.com/ghostop14/gr-grnet) that are alternate versions of
TCP/UDP sink/source. I planned on starting with this project, adding in
some missing bits from the gr-blocks UDP sink/source, and then finally
modifying it for our particular application (raw protocol instead of UDP
with IP).
I'm having problems adding variables to
udp_sink_impl.h/udp_source_impl.h without getting errors when running
the flow graph. I may be overlooking something simple, but I'm really
just trying to append the new variable following the usage of other
existing variables. I think there is either something wrong with my
python 2.7 on my xubuntu 16.04 PC, or there needs to be additional
memory allocated somewhere. Typically, when adding a new variable to
udp_sink_impl.h, I'll get the following, where the "NUM needed" varies
from run to run:
Traceback (most recent call last):
File "/home/mike/top_block.py", line 110, in <module>
main()
File "/home/mike/top_block.py", line 99, in main
tb.start()
File
"/usr/local/lib/python2.7/dist-packages/gnuradio/gr/top_block.py", line
109, in start
top_block_start_unlocked(self._impl, max_noutput_items)
File
"/usr/local/lib/python2.7/dist-packages/gnuradio/gr/runtime_swig.py",
line 5681, in top_block_start_unlocked
return _runtime_swig.top_block_start_unlocked(r, max_noutput_items)
RuntimeError: udp_sink(1): insufficient connected output ports (54929632
needed, 0 connected)
Other times, the flowgraph will start and then crash after the private
constructor finishes and the error will be related to python2 and
corrupted size vs prev_size...
*** Error in `/usr/bin/python2': corrupted size vs. prev_size:
0x0000000003428620 ***
Comment out the variable, and then it runs every time. Geez, while
writing this email, I restored the variable to get the corrupted size
error. Pressed Play again and it ran. Sometimes 2/3 tries works, and
other times 2/3 tries gets the python error (I did not notice that
behavior until just now). Added another variable, and then it went up to
like 1 in 6 worked or 1 in 10 worked, and with another variable added,
it doesn't ever work again. So this looks like a memory
allocation/corruption problem that somehow randomly works.
When testing block changes, what is the correct process to ensure
gnuradio sees the newest installed code only? I've been doing:
(current dir is build)
sudo make uninstall && sudo ldconfig && cd .. && rm -rf build
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE="Debug" ../
make -j8 && sudo make install && sudo ldconfig
Is there something I'm missing that completely removes old stuff or
failing to import all the new stuff?
I do have another local repo where I have many more changes and the
payload_size is working (don't know what changed when it suddenly
started working), but further adding variables like 'eof' result in the
error in question. The simplest changes I can make and reproduce this
can be seen here:
https://github.com/ghostop14/gr-grnet/compare/master...BCITMike:master.
I'm changing 4 files:
modified: grc/grnet_udp_sink.xml
modified: include/grnet/udp_sink.h
modified: lib/udp_sink_impl.cc
modified: lib/udp_sink_impl.h
So either I'm missing something fundamental or my dev setup is bad. If
someone can point out what I'm missing to make this work, I would
greatly appreciate it!
Thanks!
Mike
_______________________________________________
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
No comments:
Post a Comment