SIGSEGV from `ma_device_uninit`

[gen2brain]

SIGSEGV from `ma_device_uninit`

[gen2brain]

I cannot reproduce this, It works for me with both ALSA and PulseAudio. It would maybe help if you compile with the ma_debug tag, and if you can send a gdb backtrace. And finally of course, if you can build and test the C example, both with the latest miniaudio version and with the version that is used in this repo.

[santiagos01]

Hmm, interesting... Sure, let me try to attach more info first here.

---

Compilation with `ma_debug`

: Failed to initialize WASAPI backend.
: Failed to initialize DirectSound backend.
: Failed to initialize WinMM backend.
: Failed to initialize Core Audio backend.
: Failed to initialize sndio backend.
: Failed to initialize audio(4) backend.
: Failed to initialize OSS backend.
: Attempting to initialize PulseAudio backend...
[miniaudio] Endian:  LE
[miniaudio] SSE2:    YES
[miniaudio] AVX2:    NO
[miniaudio] AVX512F: NO
[miniaudio] NEON:    NO
Recording...
[PulseAudio] Capture attr: maxlength=5760, tlength=1920, prebuf=-1, minreq=-1, fragsize=1920; periodSizeInFrames=480
[PulseAudio] Capture actual attr: maxlength=5760, tlength=1920, prebuf=-1, minreq=-1, fragsize=1920; periodSizeInFrames=480
[PulseAudio] Playback attr: maxlength=5292, tlength=1764, prebuf=-1, minreq=-1, fragsize=1764; periodSizeInFrames=441
[PulseAudio] Playback actual attr: maxlength=5292, tlength=1324, prebuf=888, minreq=440, fragsize=1764; internalPeriodSizeInFrames=0
INFO: [PulseAudio]
INFO:   USB Camera Digital Stereo (IEC958) (Capture)
INFO:     Format:      16-bit Signed Integer -> 16-bit Signed Integer
INFO:     Channels:    2 -> 1
INFO:     Sample Rate: 48000 -> 44100
INFO:     Buffer Size: 480*3 (1440)
INFO:     Conversion:
INFO:       Pre Format Conversion:    NO
INFO:       Post Format Conversion:   NO
INFO:       Channel Routing:          YES
INFO:       Resampling:               YES
INFO:       Passthrough:              NO
INFO:   Built-in Audio Analog Stereo (Playback)
INFO:     Format:      16-bit Signed Integer -> 16-bit Signed Integer
INFO:     Channels:    1 -> 2
INFO:     Sample Rate: 44100 -> 44100
INFO:     Buffer Size: 441*3 (1323)
INFO:     Conversion:
INFO:       Pre Format Conversion:    NO
INFO:       Post Format Conversion:   NO
INFO:       Channel Routing:          YES
INFO:       Resampling:               NO
INFO:       Passthrough:              NO
Press Enter to stop recording...

fatal error: unexpected signal during runtime execution
[signal SIGSEGV: segmentation violation code=0x1 addr=0x7f079bfff910 pc=0x7f08034dac99]

runtime stack:
runtime.throw({0x507017?, 0x1b7bba0?})
        /usr/local/go/src/runtime/panic.go:1047 +0x5d fp=0x7ffc9b16ed28 sp=0x7ffc9b16ecf8 pc=0x43d59d
runtime.sigpanic()
        /usr/local/go/src/runtime/signal_unix.go:819 +0x369 fp=0x7ffc9b16ed78 sp=0x7ffc9b16ed28 pc=0x451709

goroutine 1 [syscall]:
runtime.cgocall(0x4a0500, 0xc00007bc08)
        /usr/local/go/src/runtime/cgocall.go:158 +0x5c fp=0xc00007bbe0 sp=0xc00007bba8 pc=0x40e03c
github.com/gen2brain/malgo._Cfunc_ma_device_uninit(0x1b7bb40)
        _cgo_gotypes.go:1595 +0x45 fp=0xc00007bc08 sp=0xc00007bbe0 pc=0x49bdc5
github.com/gen2brain/malgo.(*Device).Uninit.func1(0x4ef0a0?)

gdb backtrace

[Thread 0x7fff8fffe640 (LWP 208040) exited]

Thread 1 "capture" received signal SIGSEGV, Segmentation fault.
0x00007ffff7d2ac99 in ___pthread_detach (th=140735609300544) at ./nptl/pthread_detach.c:29
29      ./nptl/pthread_detach.c: No such file or directory.
(gdb) bt
#0  0x00007ffff7d2ac99 in ___pthread_detach (th=140735609300544) at ./nptl/pthread_detach.c:29
#1  0x00000000004c103b in ma_thread_wait__posix (pThread=0x60acb8) at [REDACTED]/malgo/miniaudio.h:10482
#2  ma_thread_wait (pThread=0x60acb8) at [REDACTED]/malgo/miniaudio.h:10659
#3  ma_device_uninit (pDevice=0x60ab40) at [REDACTED]/malgo/miniaudio.h:33497
#4  ma_device_uninit (pDevice=0x60ab40) at [REDACTED]/malgo/miniaudio.h:33480
#5  0x0000000000468fe4 in runtime.asmcgocall () at /usr/local/go/src/runtime/asm_amd64.s:844
#6  0x000000c000030500 in ?? ()
#7  0x00007fffffffd9b8 in ?? ()
#8  0x0000000000447ef8 in runtime.exitsyscallfast.func1 () at /usr/local/go/src/runtime/proc.go:3869
#9  0x0000000000467169 in runtime.systemstack () at /usr/local/go/src/runtime/asm_amd64.s:492
#10 0x000000000046b8c5 in runtime.newproc (fn=0x1) at <autogenerated>:1
#11 0x00000000005b8ec0 in runtime[scavenger] ()
#12 0x0000000000000001 in ?? ()
#13 0x0000000000467065 in runtime.mstart () at /usr/local/go/src/runtime/asm_amd64.s:390
#14 0x0000000000466fef in runtime.rt0_go () at /usr/local/go/src/runtime/asm_amd64.s:354
#15 0x00007fff00000001 in ?? ()
#16 0x00007fffffffdb38 in ?? ()
#17 0x0000000000001000 in ?? ()
#18 0x0000000000000001 in ?? ()
#19 0x00007fffffffdb38 in ?? ()
#20 0x00007ffff7cbed90 in __libc_start_call_main (main=0x466ea0 <main>, main@entry=0x0, argc=1, argc@entry=494544, argv=0x7fffffffdb38, argv@entry=0x0) at ../sysdeps/nptl/libc_start_call_main.h:58
#21 0x00007ffff7cbee40 in __libc_start_main_impl (main=0x0, argc=494544, argv=0x0, init=<optimized out>, fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffdb28)
    at ../csu/libc-start.c:392
#22 0x000000000040b575 in _start ()

---

Early hypothesis based on the gdb backtrace:

Noticed that the ma_thread_wait__posix in the version being used in this repo (v0.10.35) is as follows:

static void ma_thread_wait__posix(ma_thread* pThread)
{
    pthread_join(*pThread, NULL);
    pthread_detach(*pThread);
}

Meanwhile, in the current latest version (v0.11.9), it is slightly different (i.e., no detach call, which caused SIGSEGV for me):

static void ma_thread_wait__posix(ma_thread* pThread)
{
    pthread_join((pthread_t)*pThread, NULL);
}

Sorry I haven't had the chance to build and test with the latest miniaudio version at the moment since I think it requires quite some adjustments (I encountered some compile errors when I simply tried to update the miniaudio.h). Maybe I can try to dig into that later.

[gen2brain]

Thanks, it looks like it is related to some bug fixed in newer versions. There were some changes in callbacks in newer versions so not sure when will I have time to update. To confirm that this is the problem, you can pick the capture example from https://github.com/mackron/miniaudio/tree/199d6a7875b4288af6a7b615367c8fdc2019b03c. It is not exactly the same as in Go, but they both call device Uninit().

[santiagos01]

Sorry for the late response -- had been occupied with other things in the past two weeks.

And yes, I managed to confirm that it was indeed the problem. Just created a PR to bump the miniaudio.h version.
It seems to have resolved the issue in my local. Could you help to take a look? Thanks!