target specific option mismatch with devel branch quiet-dsp

[StevenConyers]

Brian -- I am trying to build quiet modem. I misread directions and installed current liquid-dsp on my Ubuntu 18.04 system. It compiled and tested well, but cmake for quiet said I had to switch to devel branch. The location specified was obsolete but I believe that the proper version could be obtained via:
git clone https://github.com/quiet/quiet-dsp.git
Is this correct? I cloned it and ./configure seems happy but make fails with gcc complaining about:
dotprod_cccf.mmx.c function 'dotprod_cccf_execute_mmx': …/pmmintrin.h:
inlining failed in call to always_inline '_mm_addsub_ps: target specific option mismatch.
This can be a nasty one to debug. Any thoughts? Steve

[brian-armstrong]

Hi Steven,

I'm a bit confused about this issue being on the Android repo. To clarify, are you trying to build the libraries for Android? Some do come baked in already so that shouldn't be necessary, and if you want to rebake your own it's a somewhat involved process that may not work in the new SDK.

[StevenConyers]

Sorry for inadvertantly posting issue in wrong place. Not android, ubuntu 18.04 linux 64bit Intel i7. Although I have cloned many times from github repos, I have never been a "member" until prompted to join in order to post questions to you. Got sort of wrapped around the axle in the posting. I resolved the issue of not being able to build quiet-dsp by adding -msse4.1 to CONFIG_CFLAGS in the make file and then hacking out the libliquid version test in CMakeLists.txt in ~/quiet. Successfully built and installed /usr/local/lib/libquiet.a However after I added portaudio this morning, ~/quiet/.bootstrap fails in encoder.c. with an undefined reference to gmskframegen_set_header_len. If I need to post on that, I will try to post in correct place.

…

On Thu, Nov 21, 2019, 5:11 PM Brian Armstrong ***@***.***> wrote: Hi Steven, I'm a bit confused about this issue being on the Android repo. To clarify, are you trying to build the libraries for Android? Some do come baked in already so that shouldn't be necessary, and if you want to rebake your own it's a somewhat involved process that may not work in the new SDK. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#31?email_source=notifications&email_token=AN24ZBVJHUGULG4VQ3424LTQU4WV5A5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEE4GDEI#issuecomment-557343121>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN24ZBRAVDHB4IWIWZ7D5LTQU4WV5ANCNFSM4JQITGYA> .

[brian-armstrong]

Hi Steven,

The reason that method fails is because you removed the check that tries to ensure you have the right version of liquid installed.

Most likely, you will want to make sure you install the devel branch of the liquid repo from my fork and make sure it's suitable make install'd. If the safety check passes in CMake then you should be good to go.

Brian

[StevenConyers]

Thanks Brian. Even after switching to your variant of liquid-dsp (which you call quiet-dsp) I still had to remove the safety checks. I will try again with a clean system. My goal is to run in a browser with everything built from source and I am prepared to be persistent and tenacious. Your design is terrific. Steve San Ramon CA

…

On Sat, Nov 23, 2019, 1:46 AM Brian Armstrong ***@***.***> wrote: Hi Steven, The reason that method fails is because you removed the check that tries to ensure you have the right version of liquid installed. Most likely, you will want to make sure you install the devel branch of the liquid repo from my fork and make sure it's suitable make install'd. If the safety check passes in CMake then you should be good to go. Brian — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#31?email_source=notifications&email_token=AN24ZBVNBZVIQEG7BLIV6JTQVD3YJA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEE7RRSQ#issuecomment-557783242>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN24ZBXR3MTWBRMX3WRBBRDQVD3YJANCNFSM4JQITGYA> .

[StevenConyers]

Brian -- Starting over on a clean system (raspberry pi 4 instead of a dell i7 with ubuntu 18.04) I got the same complaint. See below. Does the safety check itself need to be updated? Do these email dialogs get into the github issue discussion? If it is more appropriate and you have the patient, you could reply to me directly at stevenjconyers@gmail.com. pi@raspberrypi:/media/pi/Elements/qmodem/quiet $ ./bootstrap -- The C compiler identification is GNU 8.3.0 -- The CXX compiler identification is GNU 8.3.0 -- Check for working C compiler: /usr/bin/cc -- Check for working C compiler: /usr/bin/cc -- works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: /usr/bin/c++ -- Check for working CXX compiler: /usr/bin/c++ -- works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - done -- Detecting CXX compile features -- Detecting CXX compile features - done -- Looking for pthread.h -- Looking for pthread.h - found -- Looking for pthread_create -- Looking for pthread_create - not found -- Looking for pthread_create in pthreads -- Looking for pthread_create in pthreads - not found -- Looking for pthread_create in pthread -- Looking for pthread_create in pthread - found -- Found Threads: TRUE -- Found Liquid: /usr/local/lib/libliquid.so -- Looking for flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block in liquid -- Looking for flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block in liquid - not found CMake Error at CMakeLists.txt:65 (message): libquiet requires the devel branch of libliquid but found a different branch installed fetch the devel branch using git clone https://github.com/quiet/liquid-dsp.git -b devel --single-branch and install it before continuing

…

-- Configuring incomplete, errors occurred! See also "/media/pi/Elements/qmodem/quiet/build/CMakeFiles/CMakeOutput.log". See also "/media/pi/Elements/qmodem/quiet/build/CMakeFiles/CMakeError.log". make: *** No targets specified and no makefile found. Stop. On Sat, Nov 23, 2019 at 10:05 AM Steve Conyers <stevenjconyers@gmail.com> wrote:

Thanks Brian. Even after switching to your variant of liquid-dsp (which you call quiet-dsp) I still had to remove the safety checks. I will try again with a clean system. My goal is to run in a browser with everything built from source and I am prepared to be persistent and tenacious. Your design is terrific. Steve San Ramon CA On Sat, Nov 23, 2019, 1:46 AM Brian Armstrong ***@***.***> wrote: > Hi Steven, > > The reason that method fails is because you removed the check that tries > to ensure you have the right version of liquid installed. > > Most likely, you will want to make sure you install the devel branch of > the liquid repo from my fork and make sure it's suitable make install'd. If > the safety check passes in CMake then you should be good to go. > > Brian > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub > <#31?email_source=notifications&email_token=AN24ZBVNBZVIQEG7BLIV6JTQVD3YJA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEE7RRSQ#issuecomment-557783242>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AN24ZBXR3MTWBRMX3WRBBRDQVD3YJANCNFSM4JQITGYA> > . >

[brian-armstrong]

Can you run git status from the liquid dsp directory and tell me what it prints?

[StevenConyers]

I just left work so I won't be able to do that from my i7 system until Monday. However I was able to get the work system to make completely as follows: instead of using the quiet-dsp that the quiet page points to, i.e., your fork, I switched to the version of liquid-dsp pointed to by CMakeLists.txt in the ~/quiet directory. Still had to remove the version test, but it built completely including the stuff that only builds when audio port is installed. Started reviewing quiet.js. So the web version uses pre-built (via emscripten) wasm module? Do you have a project that invokes emscripten to build the various components -- liquid, jansson, audio port, etc? I know emscripten somewhat and am learning cheerp, but if y'all have already created the projects, that would be super. I will do a git status on my home system tomorrow morning and send it to you before the Raider game starts. Steve

…

On Sat, Nov 23, 2019, 7:53 PM Brian Armstrong ***@***.***> wrote: Can you run git status from the liquid dsp directory and tell me what it prints? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#31?email_source=notifications&email_token=AN24ZBUBGBHYN2HK5GKFO23QVH3FNA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFADESA#issuecomment-557855304>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN24ZBXXGGMQX6SDZJVE3CTQVH3FNANCNFSM4JQITGYA> .

[StevenConyers]

This is from the repo I cloned in accordance with https://github.com/quiet , i.e., the fork: pi@raspberrypi:/media/pi/Elements/qmodem/quiet-dsp $ git status On branch master Your branch is up to date with 'origin/master'. nothing to commit, working tree clean pi@raspberrypi:/media/pi/Elements/qmodem/quiet-dsp $ Switching to the repo suggested by the error message from quiet/CMakeLists.txt, namely, git clone https://github.com/quiet/liquid-dsp.git -b devel --single-branch I get the following: pi@raspberrypi:/media/pi/Elements/qmodem/liquid-dsp $ sudo make install installing... mkdir -p /usr/local/lib mkdir -p /usr/local/include/liquid install -m 644 -p libliquid.a libliquid.so /usr/local/lib install -m 644 -p include/liquid.h /usr/local/include/liquid

…

--------------------------------------------------------- liquid-dsp was successfully installed. On some machines (e.g. Linux) you should rebind your libraries by running 'ldconfig' to make the shared object available. You might also need to modify your LD_LIBRARY_PATH environment variable to include the directory /usr/local Please report bugs to support@liquidsdr.org --------------------------------------------------------- pi@raspberrypi:/media/pi/Elements/qmodem/liquid-dsp $ git status On branch devel Your branch is up to date with 'origin/devel'. nothing to commit, working tree clean pi@raspberrypi:/media/pi/Elements/qmodem/liquid-dsp $ Using this version of liquid-dsp as opposed to quiet-dsp, then I build ok (with portaudio) from quiet PROVIDED I do the hack to CMakeLists.txt in quiet. In other words, same behavior on home machine as my work machines. Steve

On Sat, Nov 23, 2019 at 7:53 PM Brian Armstrong ***@***.***> wrote: Can you run git status from the liquid dsp directory and tell me what it prints? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#31?email_source=notifications&email_token=AN24ZBUBGBHYN2HK5GKFO23QVH3FNA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFADESA#issuecomment-557855304>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN24ZBXXGGMQX6SDZJVE3CTQVH3FNANCNFSM4JQITGYA> .

[brian-armstrong]

In the quiet directory, if you ran bootstrap you should have a build directory. In that directory there should be a CMakeFiles directory and inside that a CMakeOutput.log. Can you find the bit that says

Determining if the function flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block exists in the liquid passed with the following output:

and paste what comes after?

I believe in that directory there may also be a CMakeError.log that contains more specific info. If you could paste both or just the CMakeError.log contents, I think I might be able to tell what's happening.

[StevenConyers]

Because quiet appeared to be building ok with the liquid version check hacked out on my raspberry pi, I proceeded to attempt the encoder example and got a couldn't find liquid.h when I did a gcc encoder.c -o encoder -lquiet -ljansson -lportaudio -lliquid -lfec so I noticed that quiet.h was not in /usr/local/include thus I concluded that perhaps quiet was not fully installed after all. I blasted out quiet and recloned it and did another bootstrap without the CMakeLists.txt hack and the build and install seemed OK. Now encoder.c builds with the above gcc command but I get a segmentation fault when I execute. I will try to deal with that I hope next time I work with my home system. For now, I will try again with my work system (more important case for me) in Sunnyvale Monday morning and send you the captures you requested. Thanks for you patient support. Regards, Steve

…

On Sun, Nov 24, 2019 at 12:53 PM Brian Armstrong ***@***.***> wrote: In the quiet directory, if you ran bootstrap you should have a build directory. In that directory there should be a CMakeFiles directory and inside that a CMakeOutput.log. Can you find the bit that says Determining if the function flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block exists in the liquid passed with the following output: and paste what comes after? I believe in that directory there may also be a CMakeError.log that contains more specific info. If you could paste both or just the CMakeError.log contents, I think I might be able to tell what's happening. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#31?email_source=notifications&email_token=AN24ZBRGFYW5DFGUSHIOAV3QVLSTLA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFAUTAY#issuecomment-557926787>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN24ZBXMH4EZODGGOKYVSG3QVLSTLANCNFSM4JQITGYA> .

[StevenConyers]

Hi Brian. user1@Lap4-Precision-5540:~/quiet$ ./bootstrap CMake Error at CMakeLists.txt:65 (message): libquiet requires the devel branch of libliquid but found a different branch installed fetch the devel branch using git clone https://github.com/quiet/liquid-dsp.git -b devel --single-branch and install it before continuing

…

-- Configuring incomplete, errors occurred! See also "/home/user1/quiet/build/CMakeFiles/CMakeOutput.log". See also "/home/user1/quiet/build/CMakeFiles/CMakeError.log". CMake Error at CMakeLists.txt:65 (message): libquiet requires the devel branch of libliquid but found a different branch installed fetch the devel branch using git clone https://github.com/quiet/liquid-dsp.git -b devel --single-branch and install it before continuing

-- Configuring incomplete, errors occurred! See also "/home/user1/quiet/build/CMakeFiles/CMakeOutput.log". See also "/home/user1/quiet/build/CMakeFiles/CMakeError.log". Makefile:959: recipe for target 'cmake_check_build_system' failed make: *** [cmake_check_build_system] Error 1 user1@Lap4-Precision-5540:~/quiet$ cd build However red version is installed. But if I totally blast out quiet (in /usr/local/lib, /usr/local/include, and ~/quiet), and then reclone it from your repo and do a ./configure, it builds OK, I can install it, and I can press on to examples such as clang encoder.c -o encoder -lquiet -ljansson -portaudio -lliquid -lfec sudo ./encoder Segmentation fault (ugh, guess I need to run it under debugger to see what's up). BTW, does installing clang via sudo apt install clang cause any problems with an Emscripten install? Been doing non-quiet Emscripten builds and running stuff in browsers in anticipation of quiet going. Steve

On Sun, Nov 24, 2019 at 12:53 PM Brian Armstrong ***@***.***> wrote: In the quiet directory, if you ran bootstrap you should have a build directory. In that directory there should be a CMakeFiles directory and inside that a CMakeOutput.log. Can you find the bit that says Determining if the function flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block exists in the liquid passed with the following output: and paste what comes after? I believe in that directory there may also be a CMakeError.log that contains more specific info. If you could paste both or just the CMakeError.log contents, I think I might be able to tell what's happening. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#31?email_source=notifications&email_token=AN24ZBRGFYW5DFGUSHIOAV3QVLSTLA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFAUTAY#issuecomment-557926787>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AN24ZBXMH4EZODGGOKYVSG3QVLSTLANCNFSM4JQITGYA> .

[StevenConyers]

Hi Brian. The errors that I was getting initially when trying to run your encoder.c example were due to portaudio not finding any audio devices. I installed the Advanced Linux Sound Architecture (ALSA) package and now I can build and run all of the pure portaudio examples with good sound outpt. I then changed the gcc options for encoder to be gcc -g encoder.c -o encoder -lquiet -ljansson -lportaudio -lliquid -lfec -lasound -pthread and started stepping through the code to isolate the problem(s) that I still have (Segmentation fault). Problem happens after the call to quiet_portaudio_encoder_send(). Stepping into that function I am able to execute most the code in quiet/src/portaudio_encoder.c and portaudio/src/common/pa_process.c and portaudio/hostapi/alsa/pa_linux_alsa.c. I can set breakpoints in the callback and step through it yet I never get the sound output and gdb gets lost after the cb. If I let it rip in the main thread, I get the Segmentation fault. Any suggestions? Regards, Steve On Sun, Nov 24, 2019 at 10:24 PM Steve Conyers <stevenjconyers@gmail.com> wrote:

…

Because quiet appeared to be building ok with the liquid version check hacked out on my raspberry pi, I proceeded to attempt the encoder example and got a couldn't find liquid.h when I did a gcc encoder.c -o encoder -lquiet -ljansson -lportaudio -lliquid -lfec so I noticed that quiet.h was not in /usr/local/include thus I concluded that perhaps quiet was not fully installed after all. I blasted out quiet and recloned it and did another bootstrap without the CMakeLists.txt hack and the build and install seemed OK. Now encoder.c builds with the above gcc command but I get a segmentation fault when I execute. I will try to deal with that I hope next time I work with my home system. For now, I will try again with my work system (more important case for me) in Sunnyvale Monday morning and send you the captures you requested. Thanks for you patient support. Regards, Steve On Sun, Nov 24, 2019 at 12:53 PM Brian Armstrong ***@***.***> wrote: > In the quiet directory, if you ran bootstrap you should have a build > directory. In that directory there should be a CMakeFiles directory and > inside that a CMakeOutput.log. Can you find the bit that says > > Determining if the function flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block exists in the liquid passed with the following output: > > and paste what comes after? > > I believe in that directory there may also be a CMakeError.log that > contains more specific info. If you could paste both or just the > CMakeError.log contents, I think I might be able to tell what's > happening. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#31?email_source=notifications&email_token=AN24ZBRGFYW5DFGUSHIOAV3QVLSTLA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFAUTAY#issuecomment-557926787>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AN24ZBXMH4EZODGGOKYVSG3QVLSTLANCNFSM4JQITGYA> > . >

[StevenConyers]

Good morning Brian. All is good now. Instead of using the (probably outdated) source code on the Documentation>Examples page, I switched to the files quiet/programs/encode_soundcard.c and decode_soundcard.c and now I am making some nasty hissing/squawking sounds (when "audible") and recovering fairly large files on the decode side without any errors. Getting a few annoying complaints from ALSA but those will be OBE when I switch to Emscripten and browsers. Regards, Steve On Wed, Nov 27, 2019 at 4:46 PM Steve Conyers <stevenjconyers@gmail.com> wrote:

…

Hi Brian. The errors that I was getting initially when trying to run your encoder.c example were due to portaudio not finding any audio devices. I installed the Advanced Linux Sound Architecture (ALSA) package and now I can build and run all of the pure portaudio examples with good sound outpt. I then changed the gcc options for encoder to be gcc -g encoder.c -o encoder -lquiet -ljansson -lportaudio -lliquid -lfec -lasound -pthread and started stepping through the code to isolate the problem(s) that I still have (Segmentation fault). Problem happens after the call to quiet_portaudio_encoder_send(). Stepping into that function I am able to execute most the code in quiet/src/portaudio_encoder.c and portaudio/src/common/pa_process.c and portaudio/hostapi/alsa/pa_linux_alsa.c. I can set breakpoints in the callback and step through it yet I never get the sound output and gdb gets lost after the cb. If I let it rip in the main thread, I get the Segmentation fault. Any suggestions? Regards, Steve On Sun, Nov 24, 2019 at 10:24 PM Steve Conyers ***@***.***> wrote: > Because quiet appeared to be building ok with the liquid version check > hacked out on my raspberry pi, I proceeded to attempt the encoder example > and got a couldn't find liquid.h when I did a > gcc encoder.c -o encoder -lquiet -ljansson -lportaudio -lliquid -lfec > so I noticed that quiet.h was not in /usr/local/include thus I concluded > that perhaps quiet was not fully installed after all. I blasted out quiet > and recloned it and did another bootstrap without the CMakeLists.txt hack > and the build and install seemed OK. Now encoder.c builds with the above > gcc command but I get a segmentation fault when I execute. I will try to > deal with that I hope next time I work with my home system. For now, I > will try again with my work system (more important case for me) in > Sunnyvale Monday morning and send you the captures you requested. Thanks > for you patient support. > Regards, Steve > > On Sun, Nov 24, 2019 at 12:53 PM Brian Armstrong < > ***@***.***> wrote: > >> In the quiet directory, if you ran bootstrap you should have a build >> directory. In that directory there should be a CMakeFiles directory and >> inside that a CMakeOutput.log. Can you find the bit that says >> >> Determining if the function flexframesync_set_header_len,flexframegen_set_header_len,ofdmflexframesync_set_header_len,ofdmflexframegen_set_header_len,resamp_rrrf_execute_output_block,liquid_pack_array_block exists in the liquid passed with the following output: >> >> and paste what comes after? >> >> I believe in that directory there may also be a CMakeError.log that >> contains more specific info. If you could paste both or just the >> CMakeError.log contents, I think I might be able to tell what's >> happening. >> >> — >> You are receiving this because you commented. >> Reply to this email directly, view it on GitHub >> <#31?email_source=notifications&email_token=AN24ZBRGFYW5DFGUSHIOAV3QVLSTLA5CNFSM4JQITGYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFAUTAY#issuecomment-557926787>, >> or unsubscribe >> <https://github.com/notifications/unsubscribe-auth/AN24ZBXMH4EZODGGOKYVSG3QVLSTLANCNFSM4JQITGYA> >> . >> >