# Crash in mixerSuspend **URL:** \ **Category:** FMOD Engine\ **Tags:** android\ **Created:** [August 21, 2026, 1:34pm UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471 "2026-08-21T13:34:10Z")\ **Posts on this page:** 11\ **Page:** 1
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [August 21, 2026, 1:34pm UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/1 "2026-08-21T13:34:10Z")
We encountered an Android ARM64 crash inside the FMOD mixer thread while calling System::mixerSuspend(). The complete FMOD logging-library output is available and will be attached. Environment: - FMOD Engine: 2.03.14 - FMOD library: libfmodL.so, ARM64 - Device: Samsung SM-A715F - OS: Android 12 / SDK 31 - Application build: 26.5.0.42912 - FMOD output: AAudio - Output configuration: default output, 48 kHz, stereo - Local libfmodL.so build ID: 7f6812d87ea9d1a9ea3369c789434094 Crash summary: Thread: FMOD mixer thre Signal: 11 (Segmentation fault) Fault address: 0x0000000000000000 0 libc.so sem\_post 1 libfmodL.so 0x140f90 2 libfmodL.so 0x140f90 3 libfmodL.so 0x1381a8 4 libfmodL.so 0x1409d8 5 libc.so 6 libc.so The crash happened during: DeviceManagerPollEvents → Pause → FMODAudioManager → FMOD\_System\_MixerSuspend Relevant sequence from the log: 1. FMOD successfully initialized with AAudio: 20:56:06.615 FMOD output after initialization: type=AAudio (18), drivers=4, selected driver=0 ‘(Default Output)’, rate=48000 Hz, speaker mode=3, channels=2 1. The application completed two earlier mixer suspend/resume cycles successfully. Each cycle logged: FMOD mixer suspend requested SystemI::mixerSuspend: Suspending output. Thread::callback: FMOD mixer thread finished. FMOD mixer suspend completed FMOD mixer resume requested Thread::initThread: Init FMOD mixer thread. FMOD mixer resume completed 1. Microphone capture was subsequently opened: 20:56:54.272 MicrophoneController: Starting audio capture at 44100 Hz 20:56:54.548 AudioManager: Opened audio capture device with sample rate 44100, frames 2048, channels 1. 1. Approximately one minute later, while the application was active and not paused, FMOD failed to stop an AAudio stream: 20:57:53.727 fmod\_output\_aaudio.cpp:201 AAudio\_StopStreamBlocking Failed to stop stream after 1000ms. State = 13. Result = -885. 20:57:53.728 fmod\_output\_aaudio.cpp:1010 AAudio\_Stop Cannot stop stream. Result = 0xFFFFFC8B. At this point the diagnostic context reported: Paused:false StackInfo:/Update/GameAudioManager We understand state 13 to be AAUDIO\_STREAM\_STATE\_DISCONNECTED and result -885 to be AAUDIO\_ERROR\_TIMEOUT. 1. FMOD then automatically reconstructed the output: 20:57:54.202 AAudio\_CreateOutputStream: Could not set driver at index [0]. Defaulting to driver with id: 3. 20:57:54.202 AAudio\_Init: Sample rate: HW = 48000, SW = 24000 Block size: HW = 192, SW = 512. 20:57:54.207 Thread::initThread: Init FMOD mixer thread. Semaphore: Yes, Looping: Yes. 20:57:54.209 Thread::callback: FMOD mixer thread finished. Before this reinitialization, the hardware block size was 96 rather than 192. 1. About five seconds later, the application was paused because the user started an Android in-app purchase: 20:57:58.934 StartBuying(com.outfit7.xxx.soc\_bundle) 20:57:58.954 Extern call: Pause The application first closed its microphone capture device and then suspended FMOD: 20:57:58.968 MicrophoneController: Stopping audio capture 20:57:58.968 AudioManager: Closing audio capture device 20:57:59.043 FMOD pause callback: initialized=1, mixer suspended=0 20:57:59.044 FMOD mixer suspend requested: suspended before=0 20:57:59.044 SystemI::mixerSuspend: Suspending output. 20:57:59.046 Thread::callback: FMOD mixer thread finished. 20:57:59.055 CRASH: FMOD mixer thread, SIGSEGV, fault address 0x0 The FMOD mixer suspend completed message was never reached. We inspected the bundled ARM64 binary around the crash offsets. The sequence corresponding to the stack is: 0x1381a0 Load a semaphore pointer from the FMOD thread object 0x1381a4 Call FMOD’s semaphore-post wrapper 0x140f8c Call libc sem\_post 0x140f90 Return address reported in the crash 0x1381a8 Caller return address reported in the crash 0x1409d8 FMOD OS thread trampoline after the thread callback This appears to indicate that the mixer thread logged that it had finished and then attempted to post a null or already-cleared internal completion semaphore. Integration details: - mixerSuspend() and mixerResume() are called from the same main/lifecycle thread. - Duplicate suspend/resume requests are guarded. - FMOD Studio updates are skipped while the mixer is suspended. - During the fatal transition, an asynchronous UI sound load completed before the pause handler called mixerSuspend(). - We do not see another project-initiated FMOD API call after the final mixerSuspend() begins. - Earlier suspend/resume cycles in the same session completed successfully. - The crash only occurred after the AAudio stop timeout and automatic output/mixer-thread reconstruction. ---
**Author:** ![jeff\_fmod](https://yyz2.discourse-cdn.com/flex036/user_avatar/qa.fmod.com/jeff_fmod/32/1766_2.png) [@jeff\_fmod](https://qa.fmod.com/u/jeff_fmod)\ **Post date:** [August 26, 2026, 12:00am UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/2 "2026-08-26T00:00:44Z")
Thank you for the detailed write up. I can reproduce the `Failed to stop stream` error by disconnecting headphones during a suspend, but I haven’t reproduced a segfault yet. If you have the full log available, can you please upload it to your [FMOD Profile](https://fmod.com/profile#uploads)? ---
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [August 26, 2026, 6:33am UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/3 "2026-08-26T06:33:33Z")
its up ---
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [August 31, 2026, 11:42am UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/4 "2026-08-31T11:42:22Z")
any info on this? ---
**Author:** ![jeff\_fmod](https://yyz2.discourse-cdn.com/flex036/user_avatar/qa.fmod.com/jeff_fmod/32/1766_2.png) [@jeff\_fmod](https://qa.fmod.com/u/jeff_fmod)\ **Post date:** [September 1, 2026, 1:03am UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/5 "2026-09-01T01:03:22Z")
Sorry for the delay, I’ve got a hypothetical reproduction of this issue- the problem is with a recent change that was supposed to solve a similar deadlock, but now means the mixer thread won’t close if an error is encountered during `mixerSuspend`, resulting in the mixer thread calling into a freed semaphore. I’ll write this up for the dev team to fix in a future release, thank you for reporting this issue. The only workaround for the time being would be to pause/unpause the master bus during background/foreground transitions instead of suspending/resuming the mixer. i.e ```cpp FMOD::Studio::Bus* masterBus; system->getBus("bus:/", &masterBus); FMOD::ChannelGroup* masterChannelGroup; masterBus->getChannelGroup(&masterChannelGroup); //Suspend masterChannelGroup->setPaused(true); //Reume masterChannelGroup->setPaused(false); ``` ---
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [September 1, 2026, 12:43pm UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/6 "2026-09-01T12:43:18Z")
ok, we’ll see how critical it is when we go live. any timeline on the fix? ---
**Author:** ![jeff\_fmod](https://yyz2.discourse-cdn.com/flex036/user_avatar/qa.fmod.com/jeff_fmod/32/1766_2.png) [@jeff\_fmod](https://qa.fmod.com/u/jeff_fmod)\ **Post date:** [September 3, 2026, 11:17pm UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/7 "2026-09-03T23:17:51Z")
We are close to the end of our current release window, so it has been scheduled for the release after that, likely in a month’s time. I can however get you a non-QA approved build once the fix becomes available. ---
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [September 4, 2026, 1:37pm UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/8 "2026-09-04T13:37:54Z")
Yeah, that would be nice 🙂 Thanks ---
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [September 4, 2026, 1:42pm UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/9 "2026-09-04T13:42:15Z")
If I read correctly, current android impl of mixerSuspend/Resume shuts down audio device? ---
**Author:** ![jeff\_fmod](https://yyz2.discourse-cdn.com/flex036/user_avatar/qa.fmod.com/jeff_fmod/32/1766_2.png) [@jeff\_fmod](https://qa.fmod.com/u/jeff_fmod)\ **Post date:** [September 7, 2026, 4:59am UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/10 "2026-09-07T04:59:07Z")
Yes as of [2.03.06](https://fmod.com/docs/2.03/api/welcome-revision-history.html#20306) `mixerSuspend` shuts down the audio device on Android. This was to prevent an audio glitch: > - Core API - Android - Fixed small amount of old audio being played after calling **[System::mixerResume](https://fmod.com/docs/2.03/api/core-api-system.html#system_mixerresume)** when using OpenSL or AAudio output types. ---
**Author:** ![marko.ravnjak](https://avatars.discourse-cdn.com/v4/letter/m/c67d28/32.png) [@marko.ravnjak](https://qa.fmod.com/u/marko.ravnjak)\ **Post date:** [September 7, 2026, 11:11am UTC](https://qa.fmod.com/t/crash-in-mixersuspend/24471/11 "2026-09-07T11:11:34Z")
We are also having some issues on android with native os video recorders. On some samsung device we see long freezes when video recording is started, while on some pixel devices, recorder does not “collect” audio - the video is silent during playback. Need to try with mainbus pause/resume…