Can't get sound to work (LWJGL 3.2.1 build 12, Win10 64-bit)

[princec]

Try as I might... no sound. LWJGL2.9 works, of course.

My init code:

String defaultDeviceName = alcGetString(0, ALC_DEFAULT_DEVICE_SPECIFIER);
			LOGGER.finer(() -> "Default OpenAL device: " + defaultDeviceName);
			long device = alcOpenDevice(defaultDeviceName);
			if (device == 0L) {
				throw new IllegalStateException("Can't open device " + defaultDeviceName);
			}
			int[] attributes = {0};
			long context = alcCreateContext(device, attributes);
			if (!alcMakeContextCurrent(context)) {
				throw new IllegalStateException("Context cannot be made current");
			}
			alcCapabilities = createCapabilities(device);
			alCapabilities = createCapabilities(alcCapabilities);

Prints "Default OpenAL device: OpenAL Soft", as expected. No errors seem to be raised with alGetError (and why should they - the rest of my code is unchanged from 2.9)

Cas

[spasi]

Hey Cas,

A few things to try:

1. Pass NULL to alcOpenDevice:

alcOpenDevice((ByteBuffer)null)

2. Print the list of the available devices with:

List<String> devices = ALUtil.getStringList(NULL, ALC_ALL_DEVICES_SPECIFIER)

and use one of those names with alcOpenDevice.

3. Enable OpenAL Soft's logging by setting the ALSOFT_LOGLEVEL=3 environment variable. More info here. Anything interesting in the output?

4. Use the OpenAL Soft shared library that comes with LWJGL 2. You can do that easily with -Dorg.lwjgl.openal.libname=<path_to_OpenAL64.dll>. Does that resolve the issue?

[princec]

Firstly the AL debug output goes like this:

AL lib: (II) alc_initconfig: Initializing library v1.19.1-0da960c HEAD
AL lib: (II) alc_initconfig: Supported backends: wasapi, dsound, winmm, null, wave
AL lib: (II) ReadALConfig: Loading config C:\Users\foo\AppData\Roaming\alsoft.ini...
AL lib: (II) GetProcBinary: Got path: C:\Program Files\Java\zulu11.2.3-jdk11.0.1-win_x64\bin
AL lib: (II) ReadALConfig: Loading config C:\Program Files\Java\zulu11.2.3-jdk11.0.1-win_x64\bin\alsoft.ini...
AL lib: (II) GetConfigValue: Key disable-cpu-exts not found
AL lib: (II) FillCPUCaps: Detected max CPUID function: 0xd (ext. 0x80000008)
AL lib: (II) FillCPUCaps: Vendor ID: ""
AL lib: (II) FillCPUCaps: Name: "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz"
AL lib: (II) FillCPUCaps: Extensions: +SSE +SSE2 +SSE3 +SSE4.1
AL lib: (II) GetConfigValue: Key rt-prio not found
AL lib: (II) GetConfigValue: Key resampler not found
AL lib: (II) GetConfigValue: Key trap-al-error not found
AL lib: (II) GetConfigValue: Key trap-alc-error not found
AL lib: (II) GetConfigValue: Key reverb/boost not found
AL lib: (II) GetConfigValue: Key drivers not found
AL lib: (II) ALCwasapiProxy_messageHandler: Starting message thread
AL lib: (II) ALCwasapiProxy_messageHandler: Message thread initialization complete
AL lib: (II) ALCwasapiProxy_messageHandler: Starting message loop
AL lib: (II) alc_initconfig: Initialized backend "wasapi"
AL lib: (II) alc_initconfig: Added "wasapi" for playback
AL lib: (II) alc_initconfig: Added "wasapi" for capture
AL lib: (II) GetConfigValue: Key excludefx not found
AL lib: (II) GetConfigValue: Key default-reverb not found
AL lib: (II) GetConfigValue: Key channels not found
AL lib: (II) GetConfigValue: Key sample-type not found
AL lib: (II) GetConfigValue: Key frequency not found
AL lib: (II) GetConfigValue: Key periods not found
AL lib: (II) GetConfigValue: Key period_size not found
AL lib: (II) GetConfigValue: Key sources not found
AL lib: (II) GetConfigValue: Key slots not found
AL lib: (II) GetConfigValue: Key sends not found
AL lib: (II) ALCwasapiProxy_messageHandler: Got message "Open Device" (0x0400, lparam=00000176856784F8, wparam=00000076531FF120)
AL lib: (II) GetConfigValue: Key ambi-format not found
AL lib: (II) alcOpenDevice: Created device 000001768BB05A50, "OpenAL Soft on Speakers (High Definition Audio Device)"
AL lib: (II) GetConfigValue: Key hrtf not found
AL lib: (II) UpdateDeviceParams: Pre-reset: Stereo, Float, 44100hz, 1024 update size x3
AL lib: (II) ALCwasapiProxy_messageHandler: Got message "Reset Device" (0x0401, lparam=00000176856784F8, wparam=00000076531FF020)
AL lib: (II) UpdateDeviceParams: Post-reset: Stereo, Float, 48000hz, 960 update size x3
AL lib: (II) GetConfigValue: Key stereo-mode not found
AL lib: (II) aluInitRenderer: HRTF disabled
AL lib: (II) GetConfigValue: Key cf_level not found
AL lib: (II) aluInitRenderer: BS2B disabled
AL lib: (II) GetConfigValue: Key stereo-encoding not found
AL lib: (II) aluInitRenderer: UHJ disabled
AL lib: (II) UpdateDeviceParams: Channel config, Dry: 2, FOA: 0, Real: 0
AL lib: (II) UpdateDeviceParams: Allocating 2 channels, 16384 bytes
AL lib: (II) UpdateDeviceParams: Max sources: 256 (255 + 1), effect slots: 64, sends: 2
AL lib: (II) GetConfigValue: Key dither not found
AL lib: (II) GetConfigValue: Key dither-depth not found
AL lib: (II) UpdateDeviceParams: Dithering disabled
AL lib: (II) GetConfigValue: Key output-limiter not found
AL lib: (II) UpdateDeviceParams: Output limiter enabled
AL lib: (II) UpdateDeviceParams: Fixed device latency: 1000000ns
AL lib: (II) ALCwasapiProxy_messageHandler: Got message "Start Device" (0x0402, lparam=00000176856784F8, wparam=00000076531FF020)
AL lib: (II) GetConfigValue: Key volume-adjust not found
AL lib: (II) alcCreateContext: Created context 00000176FF514110

Which to my mind looks like everything is good.

Cas

[princec]

Also hampering my testing, I'm getting a ton of native crashes. It crashes 50% of the time, sometimes with a Hotspot error log, sometimes with bizarro VM corruption like being unable to find classes, sometimes just aborting without any explanation. I don't think this is related to the sound but it's hard to isolate. Here's an example:

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x00007ffb7f52b9c3, pid=1292, tid=20772
#
# JRE version: OpenJDK Runtime Environment (11.0.1+13) (build 11.0.1+13-LTS)
# Java VM: OpenJDK 64-Bit Server VM (11.0.1+13-LTS, mixed mode, tiered, compressed oops, g1 gc, windows-amd64)
# Problematic frame:
# C  [CoreMessaging.dll+0xb9c3]
#
# No core dump will be written. Minidumps are not enabled by default on client versions of Windows
#
# An error report file with more information is saved as:
# C:\Projects\Battledroid\hs_err_pid1292.log
#
# If you would like to submit a bug report, please visit:
#   http://www.azulsystems.com/support/
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

Cas:)

[princec]

FWIW the hotspot log referred to there contains this:

Command Line: -Djava.util.logging.config.file=C:\Projects\Resources\logging.properties -ea --illegal-access=permit -Xss1m -Djava.library.path=C:\Projects\Common11\lib\windows64;C:\Projects\Steam11\lib\windows -XX:MaxDirectMemorySize=1024m -Xmx1024m -Xms256m -XX:MaxGCPauseMillis=1 -Dorg.lwjgl.util.NoChecks=false -Dspgl3.game.Launcher.resources=resources-steam.dat -Dspgl3.game.Game.gameResource=game.steam -Dnet.puppygames.steam.Steam.DEBUG=true -Djava.util.logging.config.file=C:\Projects\Battledroid\src-client/logging.properties -Dnet.puppygames.common.threading.TaskGraphBuilder.check=truex -Dfile.encoding=Cp1252 spgl3.game.Launcher

Host: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz, 8 cores, 15G,  Windows 10 , 64 bit Build 17134 (10.0.17134.556)
Time: Sat Feb 16 13:25:56 2019 GMT Standard Time elapsed time: 4 seconds (0d 0h 0m 4s)

---------------  T H R E A D  ---------------

Current thread (0x00000210f6fcd800):  JavaThread "main" [_thread_in_native, id=20772, stack(0x000000f809800000,0x000000f809900000)]

Stack: [0x000000f809800000,0x000000f809900000],  sp=0x000000f8098fe630,  free space=1017k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C  [CoreMessaging.dll+0xb9c3]

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j  org.lwjgl.system.JNI.invokeV(J)V+0
j  org.lwjgl.glfw.GLFW.glfwPollEvents()V+8
j  spgl3.game.Display.processMessages()V+6
j  spgl3.game.Display.update()V+3
j  bots.client.Splash$1.execute()V+54
j  spgl3.opengl.AbstractRenderer.build()Lspgl3/sprites/GeometryData;+117
j  spgl3.opengl.GLRenderer.render()V+1
j  bots.client.Splash.update(II)V+385
j  spgl3.game.Game.init(Ljava/util/Properties;Ljava/io/InputStream;)V+802
j  spgl3.game.Launcher.main([Ljava/lang/String;)V+247
v  ~StubRoutines::call_stub

siginfo: EXCEPTION_ACCESS_VIOLATION (0xc0000005), reading address 0x00007ffb00000050

Cas

[spasi]

- Does it always crash in glfwPollEvents()?
- Does it always crash in CoreMessaging.dll?
- Does it always crash relatively early in the game execution? I.e. while loading stuff in background threads or while JIT compilation is still happening.
- Do you launch the game from Eclipse? Does it crash when launching from the cli?

[princec]

- Not always; it crashes in all sorts of totally random ways, in the manner of something clobbering memory randomly. VM crashes happen a lot but quite often it's in glfwPollEvents at the time. Might just be coincidence.
- See above
- Yes. Even running single-threaded.
- Yes, from Eclipse. Not tried CLI - will try that next.
- Also, sound mysteriously just started working 10 minutes ago. Didn't change anything. Stability however is still awful.

Cas

[princec]

No difference from the CLI. Crashes similarly.

This minimal test code works fine, without crashing, which probably exonerates my display handling class that wraps GLFW:

package spgl3.game;

import java.io.File;
import javax.imageio.ImageIO;
import org.lwjgl.glfw.GLFWErrorCallback;
import spgl3.ui.Input;

class DisplayTest {

	static GLFWErrorCallback errorCallback;

	public static void main(String[] args) throws Exception {

		errorCallback = GLFWErrorCallback.createPrint(System.err);
		Display.setTitle("Display Test");
		Display.setIcon(ImageIO.read(new File("C:\\Projects\\Battledroid\\src-client\\bd_icon.png")));
		Display.create();
		Input.init();
		Mouse.setVisible(true);
		while (!Display.isCloseRequested()) {
			Display.update();
			Display.sync(60);

			Input.receive(new InputEventReceiver() {
				@Override
				public InputEventAction onMouseButtonEvent(MouseButtonEvent event) {
					return InputEventAction.CONSUME;
				}

				@Override
				public InputEventAction onMouseEnteredEvent(MouseEnteredEvent event) {
					return InputEventAction.CONSUME;
				}

				@Override
				public InputEventAction onMouseWheelEvent(MouseWheelEvent event) {
					return InputEventAction.CONSUME;
				}

				@Override
				public InputEventAction onMouseMovedEvent(MouseMovedEvent event) {
					return InputEventAction.CONSUME;
				}

				@Override
				public InputEventAction onKeyTypedEvent(KeyTypedEvent event) {
					System.out.print(event.getCharacter());
					return InputEventAction.CONSUME;
				}

				@Override
				public InputEventAction onKeyEvent(KeyEvent event) {
					return InputEventAction.CONSUME;
				}
			});
		}

		Input.cleanup();
		Display.cleanup();
	}

}

Cas

[spasi]

Sounds similar to http://forum.lwjgl.org/index.php?topic=6871.0, which is still without explanation. Could you please try some of the possible workarounds mentioned there? (e.g. running with -XX:-CriticalJNINatives and/or -Xcomp, running with parallel instead of G1GC)

[princec]

-Xcomp : same symptoms
-XX:-CriticalJNINatives : same symptoms

Also, sound mysteriously stopped working again

Cas

[princec]

Tried -Xint too, so that rules out the compiler and probably therefore the JRE itself.

Cas

[princec]

Same problem with LWJGL3.2.0.

Cas

[princec]

Ok, LWJGL 3.1.6 is fine, in as much as I've just run everything 10 times and not had a single crash whereas 3.2.x would crash 80% of the time.
Still no sound tho'

What changed between 3.1.6 and 3.2.0?

Cas

[princec]

LOL, ignore me about the sound, I'd turned down the bloody volume to take a call. Sound is now fine, LWJGL is now fine, seems you have a massive memory crapping problem in 3.2.x though and I can't even begin to imagine how to start diagnosing it.

Ca s:)

[spasi]

I've reviewed all changes that went into 3.2.0, nothing stands out. Some things that might help:

- Which LWJGL bindings do you use? I'm guessing the core + GLFW, OpenGL and OpenAL only so far?
- Is there an unusual number of string encoding/decoding happening around the time it crashes?
- Are there multiple OpenGL contexts in use when it crashes?
- Would it be possible to produce a trace of LWJGL methods that Battledroid calls?
- A way to test this locally would be very useful. Could we work out a way for me to try a cut-down Battledroid build that reproduces the crash?