discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] csma_and_agc branch


From: Collins, Richard
Subject: Re: [Discuss-gnuradio] csma_and_agc branch
Date: Mon, 17 Jul 2017 20:27:36 -0400

Hello Bastian,

Thanks for the quick response! I don't expect that the old csma_and_agc mod is fresh in your mind, but appreciate any help.

strange errors. Any chances that you have another (old) installation somewhere on your system?

The long version:

No, I'm pretty sure that I don't. There were a couple empty "pmt" and/or "volk" folders in /usr/local/include, /usr/local/lib, as well as /usr/local/lib/python2.7/dist-packages/, which I've removed. The only output from running "ldconfig -vp | grep 'usr/local'" shows me libuhd.so entries, and when I instead grep the output for gnuradio (or 'gr-', or 'gr_'), nothing is reported. Also, I've checked there were no '*gnuradio*', 'gr-*', or 'gr_*' named files anywhere except for where I put my user's locally cloned git repositories.

I was also sure to clean out the gnuradio/build directory before attempting again.

I really don't know cmake, but poking around on my own a little, I found that within volk/lib/CMakeFiles.txt, the incorrectly addressed files (as discussed previously) are prepended with "${CMAKE_SOURCE_DIR}". Since that evaluates to the git cloned root gnuradio directory, when adding a '/volk' to those entries (e.g. ${CMAKE_SOURCE_DIR}/volk...) seems to pass further through running "cmake ../" again. I'm sure I didn't edit everything I needed to, because that didn't fix everything.

Next, I checked the version of Volk that HEAD points to (not csma_and_agc), and saw that in these volk/{lib,apps,...}/CMakeFiles.txt files, instead of saying CMAKE_SOURCE_DIR, they say PROJECT_SOURCE_DIR. So, I figured I'd give that one a try. I had a problem building the "next" branch due to something qt-related, and saw that other branches used the same Volk submodule git tag. Both master and maint seemed to build just fine (recursive checkout, same volk git tag).

The short version:

Running the following commands from within a freshly cloned (github/bastibl) gnuradio directory seemed to resolve my cmake problems:
$ git checkout -b master -t origin/master  # I probably shouldn't have added the -t
$ git submodule update --init              # using a compatible, seemingly correct volk git tag (4465f9b)
$ git checkout csma_and_agc                # using the correct branch for everything else
$ mkdir -p build && cd build && cmake -Wno-dev ../
$ make && sudo make install && sudo ldconfig


Thanks,
Richard

On Sat, Jul 15, 2017 at 9:17 AM, Bastian Bloessl <address@hidden> wrote:
Hi Richard,

strange errors. Any chances that you have another (old) installation somewhere on your system?

Best,
Bastian

> On 14. Jul 2017, at 22:36, Collins, Richard <address@hiddencom> wrote:
>
> Hello,
>
> I'm having trouble building Bastien Bloessl's csma_and_agc branch of GNURadio on Ubuntu 16.04 LTS. I've already run through his version of UHD (except for building the fpga bitstream and soft cpu firmware binaries -- it appears I'll need Windows XP or Suse 11, and a copy of Xilinx ISE 12.2 for that, eek).
>
> Here are the steps that I've taken, and the errors I received:
>
> $ git clone -b csma_and_agc --recursive https://github.com/bastibl/gnuradio.git
> $ cd gnuradio && mkdir build && cd build
> $ cmake -Wno-dev -DENABLE_GR_COMEDI=OFF ../
>
> The git clone command checks out the version of volk with the git tag:
> f28a8be8af974d1a55a176009cb0b7927bc22222
>
> Halfway through the build, I'll get a couple errors where it seems CMake (or python or something) is looking in gnuradio/ for files that exist in gnuradio/volk/... for example:
>
> /usr/bin/python2: can't open file '/home/ninja/workspace/src_bastibl/gnuradio/gen/volk_compile_utils.py': [Errno 2] No such file or directory
>
> Making a symlink to the gnuradio/volk/gen folder in gnuradio (i.e. cd ../ && ln -s volk/gen) seems to allow CMake to continue until it hits the next volk requirement and is looking again in the wrong directory.
>
> I also get the following error soon after (though it looks like outputs from two different sources are competing for stdout or stderr, given the out of order nature of some things being logged to the screen):
>
> -- CPU width is 64 bits, Overruled arch 32
> CMake Error at volk/lib/CMakeLists.txt:129 (list):
>   list sub-command REMOVE_ITEM requires list to be present.
> Call Stack (most recent call first):
>   volk/lib/CMakeLists.txt:253 (OVERRULE_ARCH)
>
> It finishes with:
>
> -- Configuring incomplete, errors occurred!
> See also "/home/ninja/workspace/src_bastibl/gnuradio/build/CMakeFiles/CMakeOutput.log".
> See also "/home/ninja/workspace/src_bastibl/gnuradio/build/CMakeFiles/CMakeError.log".
>
> Attached is the output from CMake that I'm referencing (cmake_user_displayed_output.txt), as well as the two CMake log files referenced at the end of the cmake run.
>
>
> Questions:
>
> Should I be using a different version of volk, or maybe change something in a CMakeLists.txt file?
> Does anyone happen to have a copy of the csma_and_agc binaries for the N210r4 (for use with the XCVR2450 daughtercard, as it appears the changes are only made for use with this radio card)?
>
>
> Thanks to anyone who can help,
>
> Richard
> <cmake_user_displayed_output.txt><CMakeError.log><CMakeOutput.log>

--
Dipl.-Inform. Bastian Bloessl
CONNECT Center
Trinity College Dublin

GitHub/Twitter: @bastibl
https://www.bastibl.net/



reply via email to

[Prev in Thread] Current Thread [Next in Thread]