libfluidsynth.so for Android crashes when Unload->Load soundfont few times

[syntaro]

FluidSynth version

2.4.6 and some version

(From github release section)

Describe the bug

Reload another font -> reload another font -> Reload another font -> Crashes.

I tested some openmp and some fluidsynth version

(From github release section)

reload succcess count matrix here.

fluidsynth-2.4.0-android24 ,omp27.2.12479018->only 1 count succecss, omp28.2.13676358->only 1, omp29.0.13113456->only 1
fluidsynth-2.3.7-android24 not tested
fluidsynth-2.4.3-android24 not tested
fluidsynth-2.4.4-android24 not tested
fluidsynth-2.4.5-android24,0time ok, 1time ok, 1time ok
fluidsynth-2.4.6-android24,1time ok,3time ok,1time ok

Expected behavior

Unload font and reload font with Android app success.

and few silence time (1 sec) ok (For safety calculation)

Steps to reproduce

call it twce or more.

while (true) {
    int count = fluid_synth_sfcount(handle->synth);
    if (count == 0) {
        break;
    }
    fluid_sfont_t *font = fluid_synth_get_sfont(handle->synth, count - 1);

    if (font == nullptr) {
        break;
    }
    fluid_synth_remove_sfont(handle->synth, font);
}

    handle->soundfont_id = fluid_synth_sfload(handle->synth, file.mb_str(), 0);

If this step have mistake, It is nice for me. I wish It is not bug.
But Just in Case , Writing this Issue. Thank you.

Additoinal information following.

Additional context

When I call this twice or more, I often crashes.

And, Can I write my App URL?
https://play.google.com/store/apps/details?id=org.star_advance.mixandcc&hl=ja
This app have a problem.

jint JNICALL jfluid_open(JNIEnv *env, jobject obj, jstring fontFile, jboolean lowlatency) {
    int id = findEmptyId();

    if (obj != nullptr && obj != _javaObj) {
        _javaObj = obj;
    }
    if (id < 0) {
        return id;
    }

    fluid_handle_t *handle = getFluid(id);

    if (handle == nullptr) {
        return -1;
    }
    if (handle->audio != nullptr) {
        delete_fluid_audio_driver(handle->audio);
        handle->audio = nullptr;
        std::this_thread::sleep_for(std::chrono::milliseconds(400));
    }

    if (handle->settings == nullptr) {
        handle->settings = new_fluid_settings();

        fluid_settings_setstr(handle->settings, "synth.audio-driver", "oboe");
        fluid_settings_setstr(handle->settings, "audio.oboe.sample-rate-conversion-quality", "Fastest");
        if (lowlatency) {
            fluid_settings_setstr(handle->settings, "audio.oboe.sharing-mode", "Exclusive");
            fluid_settings_setint(handle->settings, "audio.periods", 4);
            fluid_settings_setint(handle->settings, "audio.period-size", 64);
            fluid_settings_setstr(handle->settings, "audio.oboe.performance-mode", "LowLatency");
        } else {
            fluid_settings_setstr(handle->settings, "audio.oboe.sharing-mode", "Shared");
            fluid_settings_setint(handle->settings, "audio.periods", 16);
            fluid_settings_setint(handle->settings, "audio.period-size", 64);
            fluid_settings_setstr(handle->settings, "audio.oboe.performance-mode", "None");
        }
        fluid_settings_setint(handle->settings, "synth.audio-channels", 2);
        fluid_settings_setint(handle->settings, "synth.cpu-cores", 4);
        fluid_settings_setint(handle->settings, "synth.sample-rate", 48000);
    }

    if (handle->synth == nullptr) {
        handle->synth = new_fluid_synth(handle->settings);
    }

    if (handle->synth == nullptr) {
        //jfluid_close(env, obj, id);
        return -1;
    }

    while (true) {
        int count = fluid_synth_sfcount(handle->synth);
        if (count == 0) {
            break;
        }
        fluid_sfont_t *font = fluid_synth_get_sfont(handle->synth, count - 1);

        if (font == nullptr) {
            break;
        }
        fluid_synth_remove_sfont(handle->synth, font);
    }
    if (handle->soundfont_id >= 0) {
        fluid_synth_sfunload(handle->synth, handle->soundfont_id, 0);
        handle->soundfont_id = -1;
    }
    MBString file(env, fontFile);
    if (fluid_is_soundfont(file.mb_str())) {
        handle->soundfont_id = fluid_synth_sfload(handle->synth, file.mb_str(), 0);
    } else {
        return -1;
    }

    if (handle->soundfont_id == FLUID_FAILED) {
        //jfluid_close(env, obj, id);
        return -1;
    }


    if (handle->audio == nullptr) {
        handle->audio = new_fluid_audio_driver(handle->settings, handle->synth);
        initializeMod(handle);
    }

    return id;
}

Error log is it

2025-07-09 14:26:18.970  9871-9879  dvance.mixandcc         org.star_advance.mixandcc            I  NativeAlloc concurrent mark compact GC freed 39MB AllocSpace bytes, 51(2116KB) LOS objects, 75% free, 11MB/46MB, paused 784us,5.609ms total 430.261ms
2025-07-09 14:26:19.271  9871-9952  libc                    org.star_advance.mixandcc            A  Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x4e8 in tid 9952 (Thread-6), pid 9871 (dvance.mixandcc)
2025-07-09 14:26:19.456  9871-9871  Ads                     org.star_advance.mixandcc            E  JS: Uncaught TypeError: Cannot read properties of undefined (reading 'setRushSimulatedLocalEvents') (https://s0.2mdn.net/sadbundle/9194850363899531535/index.html?ev=01_253:401)
2025-07-09 14:26:19.457  9871-9871  chromium                org.star_advance.mixandcc            I  [INFO:CONSOLE:401] "Uncaught TypeError: Cannot read properties of undefined (reading 'setRushSimulatedLocalEvents')", source: https://s0.2mdn.net/sadbundle/9194850363899531535/index.html?ev=01_253 (401)
2025-07-09 14:26:19.514  9871-9871  DynamiteModule          org.star_advance.mixandcc            I  Considering local module com.google.android.gms.ads.dynamite:0 and remote module com.google.android.gms.ads.dynamite:250505301
2025-07-09 14:26:19.516  9871-9871  DynamiteModule          org.star_advance.mixandcc            I  Selected remote version of com.google.android.gms.ads.dynamite, version >= 250505301
2025-07-09 14:26:19.944 10127-10127 DEBUG                   pid-10127                            A  Cmdline: org.star_advance.mixandcc
2025-07-09 14:26:19.944 10127-10127 DEBUG                   pid-10127                            A  pid: 9871, tid: 9952, name: Thread-6  >>> org.star_advance.mixandcc <<<
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #00 pc 000000000004c310  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #01 pc 000000000009346c  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #02 pc 0000000000091ed4  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #03 pc 000000000008c2c0  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #04 pc 000000000008f634  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #05 pc 000000000009bddc  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #06 pc 0000000000097368  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #07 pc 0000000000095928  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #08 pc 000000000005e188  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (__kmpc_barrier+320) (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.945 10127-10127 DEBUG                   pid-10127                            A        #09 pc 00000000000e644c  /data/app/~~VxbfId4L9kN_lQiNbZdu8w==/org.star_advance.mixandcc-sOXoNEtTqnGDTz0CInfiFw==/split_config.arm64_v8a.apk!libomp.so (__kmp_invoke_microtask+156) (BuildId: 108e35d3224dd18eadfe72c6b2c33f1400058b87)
2025-07-09 14:26:19.974  9871-9918  dvance.mixandcc         org.star_advance.mixandcc            E  No package ID 6a found for ID 0x6a0b000f.
2025-07-09 14:26:19.975  9871-9921  FA                      org.star_advance.mixandcc            E  Missing google_app_id. Firebase Analytics disabled. See https://goo.gl/NAOOOI
2025-07-09 14:26:19.977  9871-10005 Ads                     org.star_advance.mixandcc            I  HTTP timeout: 60000 milliseconds.
2025-07-09 14:26:20.033  9871-9952  debuggerd               org.star_advance.mixandcc            I  uid: 10384, dumpable: 1
---------------------------- PROCESS ENDED (9871) for package org.star_advance.mixandcc ----------------------------

and management handle (not working for now, need write it simple)

typedef struct {
    fluid_settings_t *settings;
    fluid_synth_t *synth;
    fluid_audio_driver_t *audio;

    int soundfont_id;
}  __attribute__ ((aligned(32)))  fluid_handle_t;

#define MAX_SYNTH 10
fluid_handle_t *_allHandle = nullptr;

int findEmptyId() {
    if (_allHandle == nullptr) {
        auto ptr = (unsigned long)malloc(sizeof(fluid_handle_t ) * MAX_SYNTH * 4); // can I get margin for overun
        _allHandle = (fluid_handle_t *) (void *)(ptr + 16);
        if (_allHandle == nullptr) {
            return -1;
        }
        memset(_allHandle, 0, sizeof(fluid_handle_t) * MAX_SYNTH);
    }
    return 0;
    /*
    for (int i = 0; i < MAX_SYNTH; ++i) {
        if (_allHandle[i].settings == nullptr) {
            return i;
        }
    }
    return -1;*/
}

fluid_handle_t *getFluid(int handle) {
    if (handle < 0 || handle >= MAX_SYNTH) {
        return nullptr;
    }
    fluid_handle_t *ptr = _allHandle + handle;
    return ptr;
}

[derselbst]

I have no means to reproduce this an Android. Could you pls. provide a full stack trace where the functions are named correctly? The stack trace you've provided only shows the topmost 10 frames and ends somewhere deep in openmp, making it impossible to tell what's going on.

[Sisyphus-wang]

I have no means to reproduce this an Android. Could you pls. provide a full stack trace where the functions are named correctly? The stack trace you've provided only shows the topmost 10 frames and ends somewhere deep in openmp, making it impossible to tell what's going on.

I encountered a similar situation while performing fuzz testing on Ubuntu, where the error 'ERROR: AddressSanitizer: BUS on unknown address' frequently appeared.

Unfortunately, I deleted the PoC sample from that time. I believe the issue was related to either OpenMP or address randomization. By disabling the OpenMP option during compilation and turning off ASLR on the system, the errors stopped occurring.

These are the two sanitizer logs I kept from that time：(I build fluidsynth with tags/v2.4.6)

Firts:

AddressSanitizer:DEADLYSIGNAL
  3815460  ERROR: AddressSanitizer: BUS on unknown address (pc 0x5e600c790a30 bp 0x7ffecf659a10 sp 0x7ffecf659188 T0)
  3815460  The signal is caused by a READ memory access.
  3815460  Hint: this fault was caused by a dereference of a high value address (see register values below).  Disassemble the provided pc to learn which register was used.
    #0 0x5e600c790a30 in __sanitizer::internal_strlen(char const*) (/src/fluidsynth+0xe8a30) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83)
    #1 0x5e600c7040c5 in printf_common(void*, char const*, __va_list_tag*) asan_interceptors.cpp.o
    #2 0x5e600c704a09 in __interceptor___vsnprintf_chk (/src/fluidsynth+0x5ca09) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83)
    #3 0x7b861595d436  (/lib/x86_64-linux-gnu/libomp.so.5+0x61436) 4681(BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #4 0x7b8615942705  (/lib/x86_64-linux-gnu/libomp.so.5+0x46705) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #5 0x7b86159340fb  (/lib/x86_64-linux-gnu/libomp.so.5+0x380fb) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #6 0x7b8615934044  (/lib/x86_64-linux-gnu/libomp.so.5+0x38044) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #7 0x7b8615926e75 in __kmpc_global_thread_num (/lib/x86_64-linux-gnu/libomp.so.5+0x2ae75) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #8 0x5e600c8ab93c in fluid_rvoice_mixer_render /src/fluidsynth-2.4.6/src/rvoice/fluid_rvoice_mixer.c
    #9 0x5e600c80be39 in fluid_synth_write_s16_channels /src/fluidsynth-2.4.6/src/synth/fluid_synth.c:4790:45
    #10 0x5e600c80b700 in fluid_synth_write_s16 /src/fluidsynth-2.4.6/src/synth/fluid_synth.c:4681:12
    #11 0x5e600c86688c in fluid_file_renderer_process_block /src/fluidsynth-2.4.6/src/bindings/fluid_filerenderer.c:450:5
    #12 0x5e600c7b9321 in fast_render_loop /src/fluidsynth-2.4.6/src/fluidsynth.c:301:12
    #13 0x5e600c7b9321 in main /src/fluidsynth-2.4.6/src/fluidsynth.c:1106:9
    #14 0x7b86153b3d8f  (/lib/x86_64-linux-gnu/libc.so.6+0x29d8f) (BuildId: d5197096f709801829b118af1b7cf6631efa2dcd)
    #15 0x7b86153b3e3f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x29e3f) (BuildId: d5197096f709801829b118af1b7cf6631efa2dcd)
    #16 0x5e600c6df7b4 in _start (/src/fluidsynth+0x377b4) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83)

Second:

FluidSynth runtime version 2.4.6
Copyright (C) 2000-2025 Peter Hanappe and others.
Distributed under the LGPL license.
SoundFont(R) is a registered trademark of Creative Technology Ltd.

fluidsynth: warning: Preset 'Contrabassoon': Discarding invalid global zone
fluidsynth: warning: Preset 'Contrabassoon': Some invalid generators were discarded
fluidsynth: warning: Instrument 'Contrabassoon      ': Discarding invalid global zone
AddressSanitizer:DEADLYSIGNAL
=================================================================
3922446  ERROR: AddressSanitizer: BUS on unknown address (pc 0x56a5c9d40a30 bp 0x7ffe336e1000 sp 0x7ffe336e0778 T0)
3922446  The signal is caused by a READ memory access.
3922446  Hint: this fault was caused by a dereference of a high value address (see register values below).  Disassemble the provided pc to learn which register was used.
    #0 0x56a5c9d40a30 in __sanitizer::internal_strlen(char const*) (/src/fluidsynth+0xe8a30) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83)
    #1 0x56a5c9cb40c5 in printf_common(void*, char const*, __va_list_tag*) asan_interceptors.cpp.o
    #2 0x56a5c9cb4a09 in __interceptor___vsnprintf_chk (/src/fluidsynth+0x5ca09) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83)
    #3 0x7f938e1e8436  (/lib/x86_64-linux-gnu/libomp.so.5+0x61436) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #4 0x7f938e1cd705  (/lib/x86_64-linux-gnu/libomp.so.5+0x46705) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #5 0x7f938e1bf0fb  (/lib/x86_64-linux-gnu/libomp.so.5+0x380fb) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #6 0x7f938e1bf044  (/lib/x86_64-linux-gnu/libomp.so.5+0x38044) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #7 0x7f938e1b235d in __kmpc_fork_call (/lib/x86_64-linux-gnu/libomp.so.5+0x2b35d) (BuildId: 942188deb6a36c4e4329baf4ba82177f3832e0e7)
    #8 0x56a5c9e2abd7 in fluid_defsfont_load_all_sampledata /src/fluidsynth-2.4.6/src/sfloader/fluid_defsfont.c:382:5
    #9 0x56a5c9e26aef in fluid_defsfont_load /src/fluidsynth-2.4.6/src/sfloader/fluid_defsfont.c:524:12
    #10 0x56a5c9e254c3 in fluid_defsfloader_load /src/fluidsynth-2.4.6/src/sfloader/fluid_defsfont.c:108:8
    #11 0x56a5c9dbeeae in fluid_synth_sfload /src/fluidsynth-2.4.6/src/synth/fluid_synth.c:5423:21
    #12 0x56a5c9d687f5 in main /src/fluidsynth-2.4.6/src/fluidsynth.c:948:16
    #13 0x7f938dc3ed8f  (/lib/x86_64-linux-gnu/libc.so.6+0x29d8f) (BuildId: d5197096f709801829b118af1b7cf6631efa2dcd)
    #14 0x7f938dc3ee3f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x29e3f) (BuildId: d5197096f709801829b118af1b7cf6631efa2dcd)
    #15 0x56a5c9c8f7b4 in _start (/src/fluidsynth+0x377b4) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83)

AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: BUS (/src/fluidsynth+0xe8a30) (BuildId: 04e8428ce97f10a9116481a2e9b4f3d951d2cc83) in __sanitizer::internal_strlen(char const*)
3922446  ABORTING

[syntaro]

• Thank you very much everyone,

Yes, I think It is library vs Special System Issue, Last 5 stack is by OpenMP.

But now, at first I need to success compile "FluidSynth.so" for local computer.
Because I don't have DebugSymbol enabled binary now.

I Can not report details.
I will try it and will report later. Excuse me...

I feel Android system seems having too many gimics.

I know this, Memory Mapping File require File Permission for every Users.
and 16 byte Alignment Required since os version 15.

Im just planning do not use Memory Mapping, and use only allocated Memory.

Android is Reasonable to use it, but It is major.

I know too many users who using desktop system + Any Tablet both, and collaborating.

And I have soundfont decoder by myself.
It can decode 10% of SoundFont Specs, and I gave up, It was hard for me.

and having sound skipping trouble on low spec processor system like Android.

We really need FluidSynth for Android. thank you

[syntaro]

Oh,,, I tried ndk-build, cmake, AndroidStudio, clang.exe, gcc, mingw etc.

but all build Failed to get fluidsynth.so.
succeed 1/6. Then deploy to my device got runtime error.

Any hints? Im not using Unix. Windows11 + AndroidStudio + Genuine NDK, only.
This time installed gcc (Installed since 15 years ago). But new Technology is very very hard.

Can I get any hints? or Can I get Binary which enabled DebugSymbols for Report.

[Sisyphus-wang]

Oh,,, I tried ndk-build, cmake, AndroidStudio, clang.exe, gcc, mingw etc.

but all build Failed to get fluidsynth.so. succeed 1/6. Then deploy to my device got runtime error.

Any hints? Im not using Unix. Windows11 + AndroidStudio + Genuine NDK, only. This time installed gcc (Installed since 15 years ago). But new Technology is very very hard.

Can I get any hints? or Can I get Binary which enabled DebugSymbols for Report.

Perhaps you can use the CI artifacts on Azure.
Download Link: https://dev.azure.com/tommbrt/d3638885-de4a-4ce7-afe7-f237ae461c07/_apis/build/builds/13030/artifacts?artifactName=fluidsynth-android24&api-version=7.1&%24format=zip

[syntaro]

My compilation not succeed today.
but I got little chance with Ubuntu on Windows11.
Maybe next day I will challenge this.

And, only one information on this post.
When I used file read hook and logging on.
This error happened after my_close return and about after 0.5 sec.

and, This is my 1st time for azure. will check later.
But this one have debug symbols? file size seems little and i don't see detail information.

I will make it next day. thank you about take care.

Added Today:

I don't know what difference, but I can see debug log today. Is this log file helps handle my issue? Thank you.
Sorry about late.

and It happen inside fluid_synth_sfload and after my_close.

aftersfclose.txt

Best regards.

[syntaro]

In last post, I misread another error.
Im still having error log that not contains debug symbol.

I am planning to try multi-platform-compile at 1st time in my life for reporting.
I don't think [__kmpc_barrier+320] and [__kmp_invoke_microtask+152] is enough.
Maybe next weak..

Best regards.

libc_err.txt

[derselbst]

Here is a debug build: https://dev.azure.com/tommbrt/tommbrt/_build/results?buildId=13050&view=artifacts&pathAsName=false&type=publishedArtifacts

Here is a debug build which additionally enables ubsan (I not sure if this will work on Android). Unfortunately, it also implicitly disables openMP, since having both doesn't seem to be supported: https://dev.azure.com/tommbrt/tommbrt/_build/results?buildId=13051&view=artifacts&pathAsName=false&type=publishedArtifacts

And here is a release build which has openMP disabled: https://dev.azure.com/tommbrt/tommbrt/_build/results?buildId=13052&view=artifacts&pathAsName=false&type=publishedArtifacts

Though technically, only testing the last one is sufficient to know whether this is a bug in openMP or not, pls. test all of them.

[syntaro]

I am very happy.
Thank you so much, derselbst

And, I tested 3 binaries. Please check this, When you have time.

fluidsynth-android24-noopenmp

--No Error Log

It worked perfect. It can reload over 10 times without any error.

fluidsynth-android24-debug

LogCat shows this log..

fluidsynth-android24-debug.txt

--fluidsynth-android24-withubsan

fluidsynth-android24-withubsan.txt

--fluidsynth-android24-16kalign

LogCat shows this log.. Do I Need response another Thread?
fluidsynth-android24-16kalign.txt

Thank you.

[derselbst]

Ok, thanks. The missing libclang_rt.asan-aarch64-android.so can be obtained from the NDK r27. You could give it a try to see whether there are potentially other problems hidden somewhere.

However, since openMP doesn't seem to be running stable on Android, I'll disable it for the next release.

[syntaro]

Thank you very much.

Well,,, I uploaded Android Code for 'Use libfluidsynth.so with Java Android'.
https://github.com/syntaro/sfexample

If you need this for short test, Can you leverage it.
or Let me know if you mind if I publish it.

PS

I attached log file and cmakefile for ubsan.

errorlog_and_cmake.txt

I am looking forward to get next release of FluidSynth soon for my device and online friends device.