Fail to build on aarch64 Linux #18353

hfiguiere · 2023-07-02T21:02:13Z

Issue type

Other type of issue

Bug description

MuseScore 4.1.0-beta fails to build on aarch64 Linux.

/usr/bin/ccache /usr/bin/c++ -DDECODE_ON_THE_FLY -DFLAC__NO_DLL -DHAVE_CONFIG_H -DHAVE_MPGLIB -DHAVE_STDINT_H -DHAW_LOGGER_QT_SUPPORT -DHAW_PROFILER_ENABLED -DMUE_BUILD_ACCESSIBILITY_MODULE -DMUE_BUILD_APPSHELL_MODULE -DMUE_BUILD_AUDIO_MODULE -DMUE_BUILD_AUTOBOT_MODULE -DMUE_BUILD_BRAILLE_MODULE -DMUE_BUILD_CLOUD_MODULE -DMUE_BUILD_CONVERTER_MODULE -DMUE_BUILD_DIAGNOSTICS_MODULE -DMUE_BUILD_IMAGESEXPORT_MODULE -DMUE_BUILD_IMPORTEXPORT_MODULE -DMUE_BUILD_INSPECTOR_MODULE -DMUE_BUILD_INSTRUMENTSSCENE_MODULE -DMUE_BUILD_LANGUAGES_MODULE -DMUE_BUILD_LEARN_MODULE -DMUE_BUILD_MIDI_MODULE -DMUE_BUILD_MPE_MODULE -DMUE_BUILD_MULTIINSTANCES_MODULE -DMUE_BUILD_MUSESAMPLER_MODULE -DMUE_BUILD_NETWORK_MODULE -DMUE_BUILD_NOTATION_MODULE -DMUE_BUILD_PALETTE_MODULE -DMUE_BUILD_PLAYBACK_MODULE -DMUE_BUILD_PLUGINS_MODULE -DMUE_BUILD_PROJECT_MODULE -DMUE_BUILD_SHORTCUTS_MODULE -DMUE_BUILD_UI_MODULE -DMUE_BUILD_UPDATE_MODULE -DMUE_BUILD_WORKSPACE_MODULE -DMUE_ENABLE_AUDIO_EXPORT -DMUSESCORE_BUILD_NUMBER=\"\" -DMUSESCORE_INSTALL_NAME=\"mscore-4.1/\" -DMUSESCORE_INSTALL_PREFIX=\"/app\" -DMUSESCORE_INSTALL_SUFFIX=\"\" -DMUSESCORE_REVISION=\"\" -DMUSESCORE_UNSTABLE -DMUSESCORE_VERSION=\"4.1.0\" -DMUSESCORE_VERSION_LABEL=\"dev\" -DNDEBUG -DNO_GLIB -DNO_THREADS -DOUTSIDE_SPEEX -DPROJECT_ROOT_DIR=\"/run/build/musescore\" -DQT_CONCURRENT_LIB -DQT_CORE_LIB -DQT_DBUS_LIB -DQT_GUI_LIB -DQT_NETWORKAUTH_LIB -DQT_NETWORK_LIB -DQT_NO_DEBUG -DQT_OPENGL_LIB -DQT_PRINTSUPPORT_LIB -DQT_QMLMODELS_LIB -DQT_QML_LIB -DQT_QUICKCONTROLS2_LIB -DQT_QUICKTEMPLATES2_LIB -DQT_QUICKWIDGETS_LIB -DQT_QUICK_LIB -DQT_SUPPORT -DQT_SVG_LIB -DQT_WIDGETS_LIB -DQT_XMLPATTERNS_LIB -DQT_XML_LIB -DRANDOM_PREFIX=opusenc_prefix -DSCRIPT_INTERFACE -DSTDC_HEADERS -DTAKEHIRO_IEEE754_HACK -DUSE_FAST_LOG -Daudio_QML_IMPORT=\"/run/build/musescore/src/framework/audio/qml\" -Dflac_QML_IMPORT=\"\" -Dfluidsynth_QML_IMPORT=\"\" -Dglobal_EXPORTS=1 -Dglobal_QML_IMPORT=\"\" -Dieee754_float32_t=float -Dlame_QML_IMPORT=\"\" -Dopusenc_QML_IMPORT=\"\" -I/run/build/musescore/_flatpak_build/src/framework/audio -I/run/build/musescore/src/framework/audio -I/run/build/musescore/_flatpak_build/src/framework/audio/audio_autogen/include -I/run/build/musescore/_flatpak_build -I/run/build/musescore -I/run/build/musescore/src -I/run/build/musescore/src/framework -I/run/build/musescore/src/framework/global -I/run/build/musescore/src/engraving -I/run/build/musescore/thirdparty/googletest/googletest/include -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/include -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/external -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/utils -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/midi -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/rvoice -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/sfloader -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/bindings -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/synth -I/run/build/musescore/thirdparty/fluidsynth/fluidsynth-2.1.4/src/drivers -I/run/build/musescore/_flatpak_build/src/framework/global -I/run/build/musescore/_flatpak_build/src/framework/audio/fluidsynth -I/run/build/musescore/_flatpak_build/src/framework/audio/lame -I/run/build/musescore/thirdparty/lame/include -I/run/build/musescore/thirdparty/lame/libmp3lame -I/run/build/musescore/thirdparty/lame/mpglib -I/run/build/musescore/thirdparty/lame -I/run/build/musescore/_flatpak_build/src/framework/audio/opusenc -I/run/build/musescore/thirdparty/opusenc/libopusenc-0.2.1/include -I/run/build/musescore/thirdparty/opusenc/libopusenc-0.2.1/src -I/run/build/musescore/thirdparty/opusenc/libopusenc-0.2.1 -I/run/build/musescore/thirdparty/opus/include -I/run/build/musescore/_flatpak_build/src/framework/audio/flac -I/run/build/musescore/thirdparty/flac/flac-1.3.4 -I/run/build/musescore/thirdparty/flac/flac-1.3.4/include -I/run/build/musescore/thirdparty/flac/flac-1.3.4/src -I/run/build/muse
core/thirdparty/flac/flac-1.3.4/src/libFLAC/include -I/run/build/musescore/thirdparty/flac/flac-1.3.4/src/libFLAC++/include -isystem /usr/include/QtCore -isystem /usr/lib/mkspecs/linux-g++ -isystem /usr/include/QtGui -isystem /usr/include/QtNetwork -isystem /usr/include/QtNetworkAuth -isystem /usr/include/QtQml -isystem /usr/include/QtQuick -isystem /usr/include/QtQmlModels -isystem /usr/include/QtQuickControls2 -isystem /usr/include/QtQuickTemplates2 -isystem /usr/include/QtQuickWidgets -isystem /usr/include/QtWidgets -isystem /usr/include/QtXml -isystem /usr/include/QtXmlPatterns -isystem /usr/include/QtSvg -isystem /usr/include/QtPrintSupport -isystem /usr/include/QtOpenGL -isystem /usr/include/QtConcurrent -isystem /usr/include/QtDBus -O2 -g -pipe -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions -fstack-protector-strong -grecord-gcc-switches -fasynchronous-unwind-tables -fstack-clash-protection -fno-omit-frame-pointer -O2 -g -DNDEBUG -std=gnu++17 -Wall -Wextra -fPIC -Winvalid-pch -include /run/build/musescore/_flatpak_build/src/framework/global/CMakeFiles/global.dir/cmake_pch.hxx -MD -MT src/framework/audio/CMakeFiles/audio.dir/Unity/unity_3_cxx.cxx.o -MF src/framework/audio/CMakeFiles/audio.dir/Unity/unity_3_cxx.cxx.o.d -o src/framework/audio/CMakeFiles/audio.dir/Unity/unity_3_cxx.cxx.o -c /run/build/musescore/_flatpak_build/src/framework/audio/CMakeFiles/audio.dir/Unity/unity_3_cxx.cxx
In file included from /run/build/musescore/src/framework/audio/internal/fx/reverb/simdtypes.h:40,
                 from /run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:37,
                 from /run/build/musescore/_flatpak_build/src/framework/audio/CMakeFiles/audio.dir/Unity/unity_3_cxx.cxx:11:
/run/build/musescore/src/framework/audio/internal/fx/reverb/simdtypes_neon.h: In constructor ‘mu::audio::fx::simd::float_x4::float_x4(float, float, float, float)’:
/run/build/musescore/src/framework/audio/internal/fx/reverb/simdtypes_neon.h:69:30: error: cannot convert ‘<brace-enclosed initializer list>’ to ‘float32x4_t’ in assignment
   69 |         s = { v0, v1, v2, v3 };
      |                              ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp: In member function ‘void mu::audio::fx::ReverbProcessor::calculateTailParams()’:
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:414:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  414 |         d->damping_cf1_x4[x].a1[y] = cf1.a1;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:415:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  415 |         d->damping_cf1_x4[x].a2[y] = cf1.a2;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:416:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  416 |         d->damping_cf1_x4[x].b0[y] = cf1.b0;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:417:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  417 |         d->damping_cf1_x4[x].b1[y] = cf1.b1;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:418:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  418 |         d->damping_cf1_x4[x].b2[y] = cf1.b2;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:420:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  420 |         d->damping_cf2_x4[x].a1[y] = cf2.a1;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:421:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  421 |         d->damping_cf2_x4[x].a2[y] = cf2.a2;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:422:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  422 |         d->damping_cf2_x4[x].b0[y] = cf2.b0;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:423:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  423 |         d->damping_cf2_x4[x].b1[y] = cf2.b1;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:424:32: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  424 |         d->damping_cf2_x4[x].b2[y] = cf2.b2;
      |                                ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:426:33: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  426 |         d->ag_filter_x4[x].cf.b0[y] = ag_cf.b0;
      |                                 ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:427:33: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  427 |         d->ag_filter_x4[x].cf.b1[y] = ag_cf.b1;
      |                                 ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:428:33: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  428 |         d->ag_filter_x4[x].cf.a1[y] = ag_cf.a1;
      |                                 ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp: In member function ‘void mu::audio::fx::ReverbProcessor::_processLines(float**, int32_t)’:
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:732:54: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  732 |                 mat_in[i] = delay_out_ptr[i][cnt] = s[0];
      |                                                      ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:733:62: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  733 |                 mat_in[i + 1] = delay_out_ptr[i + 1][cnt] = s[1];
      |                                                              ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:734:62: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  734 |                 mat_in[i + 2] = delay_out_ptr[i + 2][cnt] = s[2];
      |                                                              ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:735:62: error: no match for ‘operator[]’ (operand types are ‘mu::audio::fx::simd::float_x4’ and ‘int’)
  735 |                 mat_in[i + 3] = delay_out_ptr[i + 3][cnt] = s[3];
      |                                                              ^
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp: In member function ‘void mu::audio::fx::ReverbProcessor::Processor::setupParameter(int, const std::string&, std::pair<float, float>, float)’:
/run/build/musescore/src/framework/audio/internal/fx/reverb/reverbprocessor.cpp:816:6: note: parameter passing for argument of type ‘std::pair<float, float>’ when C++17 is enabled changed to match C++14 in GCC 10.1
  816 | void ReverbProcessor::Processor::setupParameter(int index, const std::string& name,
      |      ^~~~~~~~~~~~~~~

Steps to reproduce

Build Linux aarch64 with gcc 12.2.0

Screenshots/Screen recordings

No response

MuseScore Version

4.1.0-beta

Regression

Yes, this used to work in a previous version of MuseScore 4.x

Operating system

Linux aarch64

Additional context

-- Using Compiler GCC 12.2.0

It all in src/framework/audio/internal/fx/reverb/simdtypes_neon.h

https://github.com/musescore/MuseScore/blob/master/src/framework/audio/internal/fx/reverb/simdtypes_neon.h#L73

You need to add GNUC to the #if so that the helper is built with gcc.

https://github.com/musescore/MuseScore/blob/master/src/framework/audio/internal/fx/reverb/simdtypes_neon.h#L61C1-L61C60

the initialization with an initializer list is not allowed in gcc 12.2.0, using simd intrinsics shall be used.

(I do have a working patch)

The text was updated successfully, but these errors were encountered:

bkunda · 2023-07-03T13:11:52Z

@cbjeukendrup @shoogle when you have the opportunity could you please see if you can assist here?

cbjeukendrup · 2023-07-05T21:53:52Z

(I do have a working patch)

@hfiguiere If you could submit this patch as a pull request, that would be great! (Especially since we don't really have arm64 Linux machines ready to test)

hfiguiere · 2023-07-05T21:55:51Z

I can't because if I do you'll ask me to sign a CLA.

cbjeukendrup · 2023-07-05T22:51:21Z

Okay, that's a pity. (The CLA is not such a big deal in my view, but I respect your choice of course.)
I'll need to read up on NEON intrinsics then... 😄

For the [] operators: do you know if GCC allows to return s[n] as a float& reference, or does it need the same treatment as Clang?

And for the { … } initialiser: is the solution to init each component separately, like s[0] = v0; s[1] = v1; …?

theofficialgman · 2023-07-10T13:47:35Z

@hfiguiere just drop the .patch file here so we understand what you are talking about. I think you intended to link multiple lines of code https://github.com/musescore/MuseScore/blob/master/src/framework/audio/internal/fx/reverb/simdtypes_neon.h#L61C1-L61C60 but this link is broken and only shows one

for musescore#18353

cheesy94 · 2023-11-09T01:22:17Z

Hi all, yesterday I was trying to build Musescore from the master branch in a Debian based system (Raspberry Pi OS Bullseye for RPi4, aarch64) following the instructions in the Wiki:

I installed the dependencies and Qt version 5.15.2 through apt (I suggest adding "qml-module-qt-labs-platform" in the "To run MuseScore" part as I needed it later).
Then, I used the build.cmake script (didn't need the locating Qt step).

I got the same error messages posted here so I went ahead with the solution given, replacing line 73 in simdtypes_neon.h

#if __clang__
with
#if __clang__ or __GNUC__

The "operator[]" errors disappeared, but the ‘float32x4_t’ assignment error was still there. Then I found this PR https://github.com/audacity/audacity/pull/5503/files/4f165d7afcbb7cb2bb9769b02b460d8a2994afc9 in the audacity repository, which fixes the ‘float32x4_t’ assignment error.

After a succesful build, I got Musescore to run, being able to create a new score. Nevertheless I couldn't open an existing project because the dialog wouldn't appear, so I installed the "qml-module-qt-labs-platform" package mentioned before (saw something in the debug info about it). Now, despite many warnings and errors in the debug info, I can use it properly.

hfiguiere · 2023-11-09T01:29:55Z

The flatpak is built using that patch:
https://github.com/flathub/org.musescore.MuseScore/blob/master/patches/musescore_simd_aarch64.patch

Jojo-Schmitz · 2023-11-09T11:01:22Z

It'd be quite usefull if package maintainers would send their patches upstream

hfiguiere · 2023-11-09T11:16:37Z

It'd be quite usefull if package maintainers would send their patches upstream

#18353 (comment)

Jojo-Schmitz · 2023-11-09T11:23:53Z

Not a valid reason IMHO

for musescore#18353

theofficialgman · 2023-11-14T21:18:24Z

IMHO due to the simplicity and only one way of solving the problem the original creator of the patch does not hold any copyright to it and anyone could reasonably PR it to the repo. I will just add to to my open PR as its own commit once I get it in a workable state.

for musescore#18353

for #18353

muse-bot added needs review The issue needs review to set priority, fix version or change status etc. regression MS4 Regression on a prior release labels Jul 2, 2023

muse-bot assigned bkunda and DmitryArefiev Jul 2, 2023

cbjeukendrup assigned cbjeukendrup and unassigned bkunda and DmitryArefiev Jul 5, 2023

cbjeukendrup added needs info More information is required before action can be taken and removed needs review The issue needs review to set priority, fix version or change status etc. labels Jul 10, 2023

cbjeukendrup mentioned this issue Jul 10, 2023

incorrect neon extentions attempt to get used on arm64 #18497

Closed

theofficialgman mentioned this issue Jul 13, 2023

Updates to Fix Linux ARM32/ARM64 CI #18575

Merged

8 tasks

cbjeukendrup added the dev Related to developer experience (compiling, code base, CI), rather than end user experience label Jul 15, 2023

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Sep 6, 2023

use helper for all GNU C compilers

333b7cc

for musescore#18353

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Sep 6, 2023

use helper for all GNU C compilers

cc67263

for musescore#18353

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Nov 14, 2023

use helper for all GNU C compilers

cbc9438

for musescore#18353

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Nov 15, 2023

minor patch for NEON helpers on ARM

fc65d47

for musescore#18353

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Nov 17, 2023

minor patch for NEON helpers on ARM

011939d

for musescore#18353

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Nov 18, 2023

minor patch for NEON helpers on ARM

b1e74b8

for musescore#18353

theofficialgman added a commit to theofficialgman/MuseScore that referenced this issue Nov 18, 2023

minor patch for NEON helpers on ARM

2743b0d

for musescore#18353

igorkorsukov closed this as completed in #18575 Nov 23, 2023

igorkorsukov pushed a commit that referenced this issue Nov 23, 2023

minor patch for NEON helpers on ARM

77bc6ec

for #18353

simonhollingshead mentioned this issue Mar 3, 2024

musescore: Remove 'broken' for aarch64-linux. NixOS/nixpkgs#292930

Merged

13 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fail to build on aarch64 Linux #18353

Fail to build on aarch64 Linux #18353

hfiguiere commented Jul 2, 2023 •

edited

Loading

bkunda commented Jul 3, 2023

cbjeukendrup commented Jul 5, 2023

hfiguiere commented Jul 5, 2023

cbjeukendrup commented Jul 5, 2023

theofficialgman commented Jul 10, 2023

cheesy94 commented Nov 9, 2023 •

edited

Loading

hfiguiere commented Nov 9, 2023

Jojo-Schmitz commented Nov 9, 2023

hfiguiere commented Nov 9, 2023

Jojo-Schmitz commented Nov 9, 2023

theofficialgman commented Nov 14, 2023

Fail to build on aarch64 Linux #18353

Fail to build on aarch64 Linux #18353

Comments

hfiguiere commented Jul 2, 2023 • edited Loading

Issue type

Bug description

Steps to reproduce

Screenshots/Screen recordings

MuseScore Version

Regression

Operating system

Additional context

bkunda commented Jul 3, 2023

cbjeukendrup commented Jul 5, 2023

hfiguiere commented Jul 5, 2023

cbjeukendrup commented Jul 5, 2023

theofficialgman commented Jul 10, 2023

cheesy94 commented Nov 9, 2023 • edited Loading

hfiguiere commented Nov 9, 2023

Jojo-Schmitz commented Nov 9, 2023

hfiguiere commented Nov 9, 2023

Jojo-Schmitz commented Nov 9, 2023

theofficialgman commented Nov 14, 2023

hfiguiere commented Jul 2, 2023 •

edited

Loading

cheesy94 commented Nov 9, 2023 •

edited

Loading