If I change directories and run the test script for the first test - the one which is returning test failure I get:
root@alfven:/opt/gnuradio/src/gr-oot/gr-fhss_utils/build/python# ./qa_cf_estimate_test.sh
DEPRECATED: Using filename with gr_unittest does no longer have any effect.
gr::log :INFO: message_emitter0 - Message emitter sent 1 messages
gr::log :INFO: message_emitter0 - ��:�:sent messages
free(): double free detected in tcache 2
Aborted
If I recall correctly, a double free error usually indicates an array is being used before it's allocated - which is probably the filename being passed to gr_unittest.
So it's probably a simple fix.
gnuradio-config-info -v 3.8.2.0
git rev-parse maint-3.8
42c8527069c14c1bd971f5c722cec116f6e4b694
-- Cinaed
On 12/3/20 8:32 AM, Kyle A Logue wrote:
git rev 42c8527069, gnuradio 3.8.2.0
$ make testRunning tests...Test project xyz/gnuradio21/src/gr-fhss_utils/buildStart 1: qa_cf_estimate1/3 Test #1: qa_cf_estimate ................... Passed 0.48 secStart 2: qa_fft_burst_tagger2/3 Test #2: qa_fft_burst_tagger .............. Passed 0.33 secStart 3: qa_tagged_burst_to_pdu3/3 Test #3: qa_tagged_burst_to_pdu ........... Passed 0.40 sec
100% tests passed, 0 tests failed out of 3
Total Test time (real) = 1.21 sec
From: Cinaed Simson <cinaed.simson@gmail.com>
Sent: Wednesday, December 2, 2020 18:43
To: Kyle A Logue <kyle.a.logue@aero.org>; discuss-gnuradio@gnu.org <discuss-gnuradio@gnu.org>
Subject: Re: how to auto detect peak frequency in QT FFT sinkActually, it's the opposite - maint-3.8 is the default and the master is 3.7.
Would you post your results of
make test
in the build directory of
gr-fhss_utils
for gnuradio-3.8?
-- Cinaed
On 12/2/20 5:20 PM, Kyle A Logue wrote:
FYI all gr-fhss_utils tests pass on a properly configured machine.
I highly recommend you use `pybombs` in the future.
Is it possible you have the `maint-3.7` branch of fhss_utils? I think the master branch is for 3.8.
Kyle
From: Cinaed Simson <cinaed.simson@gmail.com>
Sent: Wednesday, December 2, 2020 13:32
To: Kyle A Logue <kyle.a.logue@aero.org>; discuss-gnuradio@gnu.org <discuss-gnuradio@gnu.org>
Subject: Re: how to auto detect peak frequency in QT FFT sinkI was just curious - and the only reason I choose FFT peak was because you did.
I built gnuradio and the OOT modules from source. And it doesn't matter where you hang the gnuradio root.
And it's highly unlikely that I could compile, link to gnuradio, and run make test with missing dependencies.
On the machine I first tested it which was an old 64 bit Dell Optiplex desktop running Debian 10 (buster), and it logs the following error in dmesg:
[1233081.177466] traps: python3[22326] trap invalid opcode ip:7feef013fc5c sp:7ffed724b720 error:0 in libgnuradio-fhss_utils.so.42c85270[7feef0133000+1d000]
- where I believe error:0 implies division by zero.
Then I installed it on a new i5 laptop and it passes the first test - which I presume is the division by 0 error above - but fails the other 2 tests - and doesn't log any errors in dmesg.
Then I downloaded and installed gr-pdu_utils and gr-sandia_utils and both of those packages are riddled with make test errors.
I'm presuming these are work in progress - and I don't the time right now to explore those errors further.
-- Cinaed
On 12/1/20 2:17 PM, Kyle A Logue wrote:
No comments:
Post a Comment