Developing Games on Intel Graphics
If you are gaming on graphics integrated in your Intel Processor, this is the place for you! Find answers to your questions or post your issues with PC games
489 Discussions

Blender - crash with OpenGL selection

Antony_Riakiotakis
3,085 Views

Hi,

We have an issue in blender that looks like it's caused by the Intel driver.

The report is at:

https://developer.blender.org/T43713

And the reporter provides a file to reproduce the issue.

Graphics card is HD4600

Any build from builder.blender.org might have the issue though we have been unable to reproduce it with our dev team.

Below is a picture that should show how to reproduce the issue after opening the provided in the report file.

Cube Surfer.blend_.png

0 Kudos
47 Replies
Aaron_C_
Beginner
845 Views
# Blender 2.74 (sub 0), Commit date: 2015-03-14 01:10, Hash 3824e0f1
bpy.ops.object.select_all(action='TOGGLE')  # Operator

# backtrace
31: PyInit_mathutils_noise_types - 0xA9ABA140
30: fftw_plan_many_dft - 0xA8E447B0
29: fftw_plan_many_dft - 0xA8E447B0
28: fftw_plan_many_dft - 0xA8E447B0
27: UnhandledExceptionFilter - 0x332A8DD4
26: swscanf_s - 0x361DD3DC
25: _C_specific_handler - 0x36166FE8
24: RtlLookupFunctionEntry - 0x361661D0
23: DbgPrint - 0x3616727C
22: KiUserExceptionDispatcher - 0x36154B4C
21: DrvSetLayerPaletteEntries - 0x8014D0B0
20: RegisterProcTableCallback - 0x800095C0
19: DrvSetLayerPaletteEntries - 0x8014D0B0
18: fftw_plan_many_dft - 0xA8E447B0
17: fftw_plan_many_dft - 0xA8E447B0
16: fftw_plan_many_dft - 0xA8E447B0
15: fftw_plan_many_dft - 0xA8E447B0
14: fftw_plan_many_dft - 0xA8E447B0
13: fftw_plan_many_dft - 0xA8E447B0
12: fftw_plan_many_dft - 0xA8E447B0
11: fftw_plan_many_dft - 0xA8E447B0
10: fftw_plan_many_dft - 0xA8E447B0
9: fftw_plan_many_dft - 0xA8E447B0
8: fftw_plan_many_dft - 0xA8E447B0
7: fftw_plan_many_dft - 0xA8E447B0
6: fftw_plan_many_dft - 0xA8E447B0
5: fftw_plan_many_dft - 0xA8E447B0
4: fftw_plan_many_dft - 0xA8E447B0
3: fftw_plan_many_dft - 0xA8E447B0
2: osl_trunc_vv - 0xAB2D47E0
1: BaseThreadInitThunk - 0x35651828
0: RtlUserThreadStart - 0x3618F9B0

 

0 Kudos
Aaron_C_
Beginner
845 Views

as for the dump

0 Kudos
Aaron_C_
Beginner
845 Views
0 Kudos
Antony_Riakiotakis
845 Views

This doesn't look like the driver. If this is the backtrace generated by blender, which was added a few days back btw, then I'm not really sure if it's accurate (or even present) on a release build, you might want to try a debug build - we can help provide one. But probably the dump can help make things clearer here.

0 Kudos
Aaron_C_
Beginner
845 Views

I used a fresh build from buildbot that i downloaded i believe yesterday git hash: 3824e0f1

0 Kudos
Bernard
Valued Contributor I
845 Views

Calls from line #3 to line #18 looks like a recursive function calls. Is that by design?

0 Kudos
Bernard
Valued Contributor I
845 Views

Aaron C. wrote:

# Blender 2.74 (sub 0), Commit date: 2015-03-14 01:10, Hash 3824e0f1
bpy.ops.object.select_all(action='TOGGLE')  # Operator

# backtrace
31: PyInit_mathutils_noise_types - 0xA9ABA140
30: fftw_plan_many_dft - 0xA8E447B0
29: fftw_plan_many_dft - 0xA8E447B0
28: fftw_plan_many_dft - 0xA8E447B0
27: UnhandledExceptionFilter - 0x332A8DD4
26: swscanf_s - 0x361DD3DC
25: _C_specific_handler - 0x36166FE8
24: RtlLookupFunctionEntry - 0x361661D0
23: DbgPrint - 0x3616727C
22: KiUserExceptionDispatcher - 0x36154B4C
21: DrvSetLayerPaletteEntries - 0x8014D0B0
20: RegisterProcTableCallback - 0x800095C0
19: DrvSetLayerPaletteEntries - 0x8014D0B0
18: fftw_plan_many_dft - 0xA8E447B0
17: fftw_plan_many_dft - 0xA8E447B0
16: fftw_plan_many_dft - 0xA8E447B0
15: fftw_plan_many_dft - 0xA8E447B0
14: fftw_plan_many_dft - 0xA8E447B0
13: fftw_plan_many_dft - 0xA8E447B0
12: fftw_plan_many_dft - 0xA8E447B0
11: fftw_plan_many_dft - 0xA8E447B0
10: fftw_plan_many_dft - 0xA8E447B0
9: fftw_plan_many_dft - 0xA8E447B0
8: fftw_plan_many_dft - 0xA8E447B0
7: fftw_plan_many_dft - 0xA8E447B0
6: fftw_plan_many_dft - 0xA8E447B0
5: fftw_plan_many_dft - 0xA8E447B0
4: fftw_plan_many_dft - 0xA8E447B0
3: fftw_plan_many_dft - 0xA8E447B0
2: osl_trunc_vv - 0xAB2D47E0
1: BaseThreadInitThunk - 0x35651828
0: RtlUserThreadStart - 0x3618F9B0

 

Does this call stack belong to failed process?

0 Kudos
Antony_Riakiotakis
845 Views

I'm getting the exact same thing by causing a crash in a different place. I'll fix the crash report handler but for now don't treat this as reliable information please.

0 Kudos
Antony_Riakiotakis
845 Views

Here is a debug build of blender that should display a correct backtrace, but it probably needs debug libraries of MSVC 2013 to work

 

https://drive.google.com/file/d/0B5CLsFk4D9wHYUZnWWtKVUdPYVk/view?usp=sharing

 

 

0 Kudos
Bernard
Valued Contributor I
845 Views

What is the purpose of fftw_plan_many_dft function? Judging by the function name it is somehow related to DFT transform.

0 Kudos
Antony_Riakiotakis
845 Views

It is from the FFTW library that blender uses for composited waves and the audio system. I don't know why its in the backtrace always, but as I explained I have good reason to believe it is not related to the real issue.

0 Kudos
Aaron_C_
Beginner
845 Views

Getting a hold of the .dlls like we talked about on irc would be great so that I could test unless you can find a setup that crashes.

0 Kudos
Bernard
Valued Contributor I
845 Views

Antony Riakiotakis wrote:

It is from the FFTW library that blender uses for composited waves and the audio system. I don't know why its in the backtrace always, but as I explained I have good reason to believe it is not related to the real issue.

The best option for getting correct call stack if possible is to run Blender under windbg with break in any exception settings enabled. Hopefully windbg will be able to catch the error.

0 Kudos
Bernard
Valued Contributor I
845 Views

It seems that fftw function was called by os_trunc_vv. Just guessing by looking at backtrace.

0 Kudos
Antony_Riakiotakis
845 Views

I don't have an Intel card here. The only way to get a correct stack is to make have Aaron run the debug build or have someone from intel reproduce the issue with the provided build and file.

0 Kudos
Bernard
Valued Contributor I
845 Views

Later I will look at crash file uploaded by Aaron.

0 Kudos
Aaron_C_
Beginner
845 Views

any news?

0 Kudos
Bernard
Valued Contributor I
845 Views

I was busy last few days, so I will take a look at that dump file tomorrow.

0 Kudos
Aaron_C_
Beginner
845 Views

sorry for keep "bugging" /no pun intended\ you but any news ?

0 Kudos
Bernard
Valued Contributor I
808 Views

Sorry for delay I was quite busy.

Will post my findings soon.

0 Kudos
Malcolm_M_
Beginner
808 Views

I've just installed blender 2.7.4 on a new machine, Windows 8.1, Radeon R7, OpenGL Version 6.14.10. It's using the bundled Python 

As soon as I try an OpenGL render, even of the default scene, Blender crashes out with exactly this stack trace (except for the hex addresses) in blender,crash.txt.

It reports EXCEPTION_ACCESS_VIOLATION on the console.

I've installed the latest graphics driver.

 

0 Kudos
Reply