freeglut (foo): ERROR: Internal error <FBConfig with necessary capabilities not found> in function fgOpenWindow

[xbpeng]

freeglut (foo): ERROR: Internal error in function fgOpenWindow

from deepmimic.

[xbpeng]

cool, glad to hear that worked out!

[xbpeng]

Haven't seen that error before. You shouldn't be commenting out RLWorld. Maybe try commenting out the three lines here:

DeepMimic/DeepMimic.py

Line 269 in 20f524c

glutInitContextVersion(3, 2)

glutInitContextVersion(3, 2)
glutInitContextFlags(GLUT_FORWARD_COMPATIBLE)
glutInitContextProfile(GLUT_CORE_PROFILE)

and see if that makes any difference.

[cookiegg]

#10
I solved this problem,just comment the line:

glutInitContextVersion(3, 2)

[cookiegg]

I find the above error appear when running to

• line 275 of DeepMimic.py

glutCreateWindow(b‘DeepMimic’)

If I comment

from learning.rl_world import RLWorld

Above error disappear, and a new error occur:

forrest@cyborg:/data/Documents/DeepMimic$ python DeepMimic.py --arg_file args/run_humanoid3d_spinkick_args.txt
Renderer: TITAN X (Pascal)/PCIe/SSE2
OpenGL version supported 4.3.0 NVIDIA 384.111
Compiling shader: data/shaders/Mesh_VS.glsl
Compiling shader: data/shaders/VertColor_PS.glsl
Compiling shader: data/shaders/FullScreenQuad_VS.glsl
Compiling shader: data/shaders/DownSample_PS.glsl
Compiling shader: data/shaders/Mesh_VS.glsl
Compiling shader: data/shaders/DownSample_PS.glsl
[cyborg:29427] *** Process received signal ***
[cyborg:29427] Signal: Segmentation fault (11)
[cyborg:29427] Signal code: Address not mapped (1)
[cyborg:29427] Failing at address: (nil)
[cyborg:29427] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x10330) [0x7f6807529330]
[cyborg:29427] *** End of error message ***
Segmentation fault (core dumped)

The new error seems occur during running to

• line 293 of DeepMimic.py:

  env = DeepMimicEnv(args, enable_draw)

• line 16 of deepmimic_env.py:

  self._core.Init

• line 636 of DeepMimicCore.py:

  def Init(self):
    return _DeepMimicCore.cDeepMimicCore_Init(self)

[cookiegg]

I try commenting out the three lines you mentioned, the old error about freeglut appear again.

[renzhendian]

@cookiegg I have the same problem. Do you solve this problem?

[cookiegg]

@renzhendian I upgrade my OS from ubuntu1404 to 1604, the old error about freeglut disappeared.
I do not know how to fix the Segmentation fault (core dumped) error about shader.
However, at least the code about training policy can run normally.

python mpi_run.py --arg_file args/train_humanoid3d_spinkick_args.txt --num_workers 4

So, the problem is only related to the shader,I am still analyzing the cause of the error.

[MarshallChen]

I have the same problem on Mac OSX High Sierra. Well build this project on mac is much painful, I also modified shaders version from 130 to 120, don't think it's shaders problem at all. @cookiegg have you solved your Segmentation fault error?

Renderer: Intel Iris Pro OpenGL Engine
OpenGL version supported 2.1 INTEL-10.36.19
Compiling shader: data/shaders/Mesh_VS.glsl
Compiling shader: data/shaders/VertColor_PS.glsl
Compiling shader: data/shaders/FullScreenQuad_VS.glsl
Compiling shader: data/shaders/DownSample_PS.glsl
Compiling shader: data/shaders/Mesh_VS.glsl
Compiling shader: data/shaders/DownSample_PS.glsl
[marshall:65363] *** Process received signal ***
[marshall:65363] Signal: Segmentation fault: 11 (11)
[marshall:65363] Signal code: Address not mapped (1)
[marshall:65363] Failing at address: 0x0
[marshall:65363] [ 0] 0   libsystem_platform.dylib            0x00007fff6862af5a _sigtramp + 26
[marshall:65363] [ 1] 0   ???                                 0x0000000000000000 0x0 + 0
[marshall:65363] [ 2] 0   _DeepMimicCore.so                   0x000000011ae9635c _ZN9cMeshUtil14BuildPointMeshERNSt3__110unique_ptrI9cDrawMeshNS0_14default_deleteIS2_EEEE + 44
[marshall:65363] [ 3] 0   _DeepMimicCore.so                   0x000000011ae8dae0 _ZN9cDrawUtil11BuildMeshesEv + 16
[marshall:65363] [ 4] 0   _DeepMimicCore.so                   0x000000011ae8d7d3 _ZN9cDrawUtil12InitDrawUtilEv + 147
[marshall:65363] [ 5] 0   _DeepMimicCore.so                   0x000000011adb2378 _ZN14cDeepMimicCore4InitEv + 24
[marshall:65363] [ 6] 0   _DeepMimicCore.so                   0x000000011aef404b _ZL25_wrap_cDeepMimicCore_InitP7_objectS0_ + 91
[marshall:66630] [ 7] 0   Python                              0x000000010b229005 _PyCFunction_FastCallDict + 166
[marshall:66630] [ 8] 0   Python                              0x000000010b28f0e6 call_function + 491
[marshall:66630] [ 9] 0   Python                              0x000000010b287621 _PyEval_EvalFrameDefault + 1659
[marshall:66630] [10] 0   Python                              0x000000010b2901e9 _PyFunction_FastCall + 121
[marshall:66630] [11] 0   Python                              0x000000010b28f0bd call_function + 450
[marshall:66630] [12] 0   Python                              0x000000010b287621 _PyEval_EvalFrameDefault + 1659
[marshall:66630] [13] 0   Python                              0x000000010b28f866 _PyEval_EvalCodeWithName + 1747
[marshall:66630] [14] 0   Python                              0x000000010b290119 _PyFunction_FastCallDict + 449
[marshall:66630] [15] 0   Python                              0x000000010b1f0708 _PyObject_FastCallDict + 196
[marshall:66630] [16] 0   Python                              0x000000010b1f082b _PyObject_Call_Prepend + 156
[marshall:66630] [17] 0   Python                              0x000000010b1f0589 PyObject_Call + 101
[marshall:66630] [18] 0   Python                              0x000000010b23c4d8 slot_tp_init + 57
[marshall:66630] [19] 0   Python                              0x000000010b239439 type_call + 184
[marshall:66630] [20] 0   Python                              0x000000010b1f06d3 _PyObject_FastCallDict + 143
[marshall:66630] [21] 0   Python                              0x000000010b28f0b6 call_function + 443
[marshall:66630] [22] 0   Python                              0x000000010b287621 _PyEval_EvalFrameDefault + 1659
[marshall:66630] [23] 0   Python                              0x000000010b28f866 _PyEval_EvalCodeWithName + 1747
[marshall:66630] [24] 0   Python                              0x000000010b28ff49 fast_function + 218
[marshall:66630] [25] 0   Python                              0x000000010b28f0bd call_function + 450
[marshall:66630] [26] 0   Python                              0x000000010b2876b8 _PyEval_EvalFrameDefault + 1810
[marshall:66630] [27] 0   Python                              0x000000010b2901e9 _PyFunction_FastCall + 121
[marshall:66630] [28] 0   Python                              0x000000010b28f0bd call_function + 450
[marshall:66630] [29] 0   Python                              0x000000010b287621 _PyEval_EvalFrameDefault + 1659
[marshall:66630] *** End of error message ***
[1]    66630 segmentation fault  python3 DeepMimic.py --arg_file args/run_humanoid3d_spinkick_args.txt

[xbpeng]

It looks like you have OpenGL 2.1, we need 3.2. So that might be the cause of the errors.

[MarshallChen]

It seems not, shaders compiled without error. @cookiegg he's on OpenGL 4.3 also has this error.

forrest@cyborg:/data/Documents/DeepMimic$ python DeepMimic.py --arg_file args/run_humanoid3d_spinkick_args.txt
Renderer: TITAN X (Pascal)/PCIe/SSE2
OpenGL version supported 4.3.0 NVIDIA 384.111
Compiling shader: data/shaders/Mesh_VS.glsl
Compiling shader: data/shaders/VertColor_PS.glsl
Compiling shader: data/shaders/FullScreenQuad_VS.glsl
Compiling shader: data/shaders/DownSample_PS.glsl
Compiling shader: data/shaders/Mesh_VS.glsl
Compiling shader: data/shaders/DownSample_PS.glsl
[cyborg:29427] *** Process received signal ***
[cyborg:29427] Signal: Segmentation fault (11)
[cyborg:29427] Signal code: Address not mapped (1)
[cyborg:29427] Failing at address: (nil)
[cyborg:29427] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0x10330) [0x7f6807529330]
[cyborg:29427] *** End of error message ***
Segmentation fault (core dumped)

[MarshallChen]

So may be we should use OpenGL 3.2 explicitly?

[xbpeng]

do you know what is triggering that seg fault?

[MarshallChen]

_ZN9cMeshUtil14BuildPointMeshERNSt3__110unique_ptrI9cDrawMeshNS0_14default_deleteIS2_EEEE + 44

By the line above it's pointing out that maybe cDrawMesh() function didn't working properly on my mac. I tried to std::cout some random strings to console like it says. I think new cDrawMesh() call crashed.

void cMeshUtil::BuildPointMesh(std::unique_ptr<cDrawMesh>& out_mesh)
{
	...
        std::cout << "This will be console logged" << std::endl;
	
        out_mesh = std::unique_ptr<cDrawMesh>(new cDrawMesh());
        
        std::cout << "This will not be logged" << std::endl;
	
        BuildDrawMesh(vert_data, pos_len, norm_data, norm_len, coord_data, coord_len, idx_data, idx_len, out_mesh.get());
}

Maybe you are right it's still the OpenGL problem.

[Superlee506]

@cookiegg Hi, thanks for this related issue and I face the same problem.
I commented the glutInitContextVersion(3, 2) in Line 269, but the bug still exists. Do I need to do some other modifications?

[cookiegg]

@Superlee506
My env:
ubuntu:16.04
opengl: 4.6.0
NVIDIA Driver: 390.87

I guess the error related with the mismatch between Opengl with OS and Graph Driver. So maybe you should upgrade your OS AND Driver.