Hi All,

I have just upgraded my laptop to the latest version of GNU Radio
Companion (ver 3.8.2.0 (Python 3.6.9)), and am now trying to use it to
monitor a block of spectrum with my USRP B210. Unfortunately the
flowgraph won’t run (even though it ran in my old GNU Radio setup),
and instead prints the following message to the console:

linux; GNU C++ version 7.3.0; Boost_106501; UHD_003.010.003.000-0-unknown

UHD Warning:

EnvironmentError: IOError: Could not find path for image: usrp_b200_fw.hex

Using images directory: <no images directory located>

Set the environment variable 'UHD_IMAGES_DIR' appropriately or follow
the below instructions to download the images package.

Please run:

"/usr/lib/x86_64-linux-gnu/uhd/utils/uhd_images_downloader.py"

Traceback (most recent call last):

File "/home/anyone/Documents/Brendan/GNU-Radio/top_block.py", line
244, in <module>

main()

File "/home/anyone/Documents/Brendan/GNU-Radio/top_block.py", line
220, in main

tb = top_block_cls()

File "/home/anyone/Documents/Brendan/GNU-Radio/top_block.py", line 87,
in init

channels=list(range(0,1)),

File "/usr/lib/python3/dist-packages/gnuradio/uhd/init.py", line
125, in constructor_interceptor

return old_constructor(*args)

File "/usr/lib/python3/dist-packages/gnuradio/uhd/uhd_swig.py", line
3259, in make

return _uhd_swig.usrp_source_make(device_addr, stream_args,
issue_stream_cmd_on_start)

RuntimeError: LookupError: KeyError: No devices found for ----->

Device Address:

serial: 318425D

The above message suggests GRC is calling version 003.010.003.000-0
of the UHD driver. This is weird, as last week I installed version
3.15.0.0 of the UHD driver on my laptop, after first uninstalling
the old driver (or so I thought…).

However, if I run uhd_usrp_probe or uhd_find_devices, I get a message
confirming that I am indeed running v3.15.0.0 of the UHD driver:

 linux; GNU C++ version 7.5.0; Boost_106501;
 *UHD_3.15.0.HEAD-0-gaea0e2de*

If I look in the folder “/usr/lib/x86_64-linux-gnu/”, I find the files
libuhd.so.003.010.003 and libuhd.so.3.15.0 are both present — but
I am pretty sure there should only be one of them present!

This “dual-install” problem seems to be fairly common among USRP/GNU
Radio users, but so far I haven’t found any actual solutions.

There is also a second error message in the above console output:
“EnvironmentError: IOError: Could not find path for image:
usrp_b200_fw.hex”. This is baffling, as I have run the script
“/usr/local/lib/uhd/utils/uhd_images_downloader.py“ three times, and
am confident that the FPGA images have downloaded successfully (for
the record, they are in /usr/local/share/uhd/images).

If anyone can tell me how to resolve these problems, I would be very
grateful!

Regards,

Brendan.

On 04/19/2021 03:15 AM, brendan.horsfield@vectalabs.com wrote:

Hi All,

I have just upgraded my laptop to the latest version of GNU Radio
Companion (ver 3.8.2.0 (Python 3.6.9)), and am now trying to use it to
monitor a block of spectrum with my USRP B210. Unfortunately the flowgraph
won’t run (even though it ran in my old GNU Radio setup), and instead
prints the following message to the console:

linux; GNU C++ version 7.3.0; Boost_106501; UHD_003.010.003.000-0-unknown

UHD Warning:

EnvironmentError: IOError: Could not find path for image: usrp_b200_fw.hex

Using images directory: <no images directory located>

Set the environment variable 'UHD_IMAGES_DIR' appropriately or follow the
below instructions to download the images package.

Please run:

"/usr/lib/x86_64-linux-gnu/uhd/utils/uhd_images_downloader.py"

Traceback (most recent call last):

File "/home/anyone/Documents/Brendan/GNU-Radio/top_block.py", line 244, in
<module>

main()

File "/home/anyone/Documents/Brendan/GNU-Radio/top_block.py", line 220, in
main

tb = top_block_cls()

File "/home/anyone/Documents/Brendan/GNU-Radio/top_block.py", line 87, in
init

channels=list(range(0,1)),

File "/usr/lib/python3/dist-packages/gnuradio/uhd/init.py", line 125,
in constructor_interceptor

return old_constructor(*args)

File "/usr/lib/python3/dist-packages/gnuradio/uhd/uhd_swig.py", line 3259,
in make

return _uhd_swig.usrp_source_make(device_addr, stream_args,
issue_stream_cmd_on_start)

RuntimeError: LookupError: KeyError: No devices found for ----->

Device Address:

serial: 318425D

The above message suggests GRC is calling version 003.010.003.000-0 of
the UHD driver. This is weird, as last week I installed version 3.15.0.0
of the UHD driver on my laptop, after first uninstalling the old driver (or
so I thought…).

However, if I run uhd_usrp_probe or uhd_find_devices, I get a message
confirming that I am indeed running v3.15.0.0 of the UHD driver:

-

linux; GNU C++ version 7.5.0; Boost_106501;
*UHD_3.15.0.HEAD-0-gaea0e2de*

If I look in the folder “/usr/lib/x86_64-linux-gnu/”, I find the files
libuhd.so.003.010.003 and libuhd.so.3.15.0 are both present — but I
am pretty sure there should only be one of them present!

This “dual-install” problem seems to be fairly common among USRP/GNU Radio
users, but so far I haven’t found any actual solutions.

There is also a second error message in the above console output: “EnvironmentError:
IOError: Could not find path for image: usrp_b200_fw.hex”. This is
baffling, as I have run the script
“/usr/local/lib/uhd/utils/uhd_images_downloader.py“ three times, and am
confident that the FPGA images have downloaded successfully (for the
record, they are in /usr/local/share/uhd/images).

If anyone can tell me how to resolve these problems, I would be very
grateful!

Regards,

Brendan.

That means that the version of Gnu Radio you used to produce whatever app
you have is linked against UHD 3.10.3, whereas all
your UHD utilities are linked against the newer version.

What happens when you run:

gnuradio-config-info -v

What is in your PYTHONPATH?  Is it perhaps pointing to older python code,
and you're picking up older (very older) python modules that
are themselves linked against both an older GR and older UHD?

USRP-users mailing list -- usrp-users@lists.ettus.com
To unsubscribe send an email to usrp-users-leave@lists.ettus.com

Importing uhd in Python 3:
/usr/local/lib/python3.6/dist-packages/uhd-1.1.0-py3.6-linux-x86_64.egg/uhd.cpython-36m-x86_64-linux-gnu.so
http://uhd.cpython-36m-x86_64-linux-gnu.so

(Note that this is an alternative Python UHD module written by
Christian Hahn, which I have been evaluating for the last few days.)

Trying to import uhd in Python 2 (which I never use) returns an error:

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/local/lib/python2.7/dist-packages/uhd/init.py", line
10, in <module>
File "/usr/local/lib/python2.7/dist-packages/uhd/types.py", line 10,
in <module>
# "iter" and "next" attributes instead.
ImportError: cannot import name libpyuhd

Brendan.

On 04/19/2021 06:21 PM, Brendan Horsfield wrote:

Importing uhd in Python 3:
/usr/local/lib/python3.6/dist-packages/uhd-1.1.0-py3.6-linux-x86_64.egg/
uhd.cpython-36m-x86_64-linux-gnu.so

(Note that this is an alternative Python UHD module written by Christian
Hahn, which I have been evaluating for the last few days.)

Trying to import uhd in Python 2 (which I never use) returns an error:

Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/local/lib/python2.7/dist-packages/uhd/init.py", line 10,
in <module>
File "/usr/local/lib/python2.7/dist-packages/uhd/types.py", line 10, in
<module>
# "iter" and "next" attributes instead.
ImportError: cannot import name libpyuhd

Brendan.

Given that gr-uhd wants a module called "uhd" to exist, and GR uses
gr-uhd, i'm not sure what will happen with the
name-space pollution that will result.

OK, I didn't realise that GNU Radio needed the Python UHD module to be
installed.  I deliberately excluded the Python API from my UHD driver
build in order to try out Christian's custom Python API.

I will remove the custom API today, and install a "stock" UHD driver
with the Ettus Python API.  Then I will give the GNU Radio
installation another try.

Thanks,
Brendan.