Posted by
Sven Gothel on
Sep 01, 2015; 12:32am
URL: https://forum.jogamp.org/JOGL-crashes-under-MacOSX-with-Java-8u60-tp4035193p4035209.html
On 08/31/2015 07:24 PM, Frank-Michael Moser [via jogamp] wrote:
> After updating Java from 8u45 to 8u60 I noticed that my 3D app craches when
> clothing the 3D-view:
>
> SIGSEGV (0xb) at pc=0x00007fff878d00dd, pid=5601, tid=1299
>
> Stack: [0x00007fff5884e000,0x00007fff5904e000], sp=0x00007fff59049368,
> free space=8172k
> Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
> C [libobjc.A.dylib+0x10dd] objc_msgSend+0x1d
> j
> jogamp.nativewindow.jawt.macosx.MacOSXJAWTWindow.UnsetJAWTRootSurfaceLayer0(JJ)V+0
>
> j jogamp.nativewindow.jawt.macosx.MacOSXJAWTWindow.access$200(JJ)V+2
> j jogamp.nativewindow.jawt.macosx.MacOSXJAWTWindow$1.run()V+38
> j com.jogamp.common.util.RunnableTask.run()V+24
>
> This does not happen with Java 8u45. I could reproduce the crash with JOGL
> 2.2.0, 2.3.1, and also with the latest version from the master branch
> (
http://jogamp.org/deployment/archive/master/gluegen_886-joal_612-jogl_1431-jocl_1078)
>
I could not reproduce the crash here:
- OSX 10.10.4
- GL Version 4.1 (Core profile, arb, compat[ES2], FBO, hardware) - 4.1 INTEL-10.6.31 [GL 4.1.0, vendor 10.6.31 (INTEL-10.6.31)]
- Quirks [NoOffscreenBitmap, GL4NeedsGL3Request, NeedSharedObjectSync]
- GL_VENDOR Intel Inc.
- GL_RENDERER Intel Iris OpenGL Engine
- GL_VERSION 4.1 INTEL-10.6.31
- Java 1.8.0_60-b27
Tests:
- com.jogamp.opengl.test.junit.jogl.demos.es2.awt.TestGearsES2AWT
- com.jogamp.opengl.test.junit.jogl.demos.es2.newt.TestGearsES2NewtCanvasAWT
A crash in 'UnsetJAWTRootSurfaceLayer0', called at JAWTWindow invalidate
may hint to an already released CALayer or jawtSurfaceLayer (owned by JAWT)?
I have to reproduce the crash first, can you
- test whether the above 2 test cases crash?
- state your platform details, similar to what I have written here.
- If both tests work and the OSX version / GL renderer also
don't make a difference, please try to write a small test case
able to reproduce this issue.
Updating to OSX 10.10.5 .. and will retest
Side note:
I had to apply the latest patch for a vsync regression thought:
<
http://jogamp.org/git/?p=jogl.git;a=commit;h=aacc8afdca0b1376f91dcc68aa3ae3b39c7aba51>
~Sven