2.66 - segmentation fault when selecting an object during the initial phase of rendering #34322

New Issue

Dani Epstein · 2013-02-19T13:29:36+01:00

Dani Epstein commented

2013-02-19 13:29:36 +01:00

%%%Open the file, hit F12, and then in one of the right-hand windows select a mesh. This results in a segmentation fault, at least on my computer.

Ubuntu 12.10 64-bit
NVIDIA Driver Version: 304.43
Graphics Crad: GeForce GTX 460 2GB

- Blender version with error, and version that worked ---

This bug occurs from at least r51699. It might in earlier versions post r51611 (see further), but I only have r51611 to check with.
The bug also occurs in build 54535 (from the Buildbot)
I tried it with a build 51611 and that error does not occur.%%%

%%%Open the file, hit F12, and then in one of the right-hand windows select a mesh. This results in a segmentation fault, at least on my computer. Ubuntu 12.10 64-bit NVIDIA Driver Version: 304.43 Graphics Crad: GeForce GTX 460 2GB - Blender version with error, and version that worked --- This bug occurs from at least r51699. It might in earlier versions post r51611 (see further), but I only have r51611 to check with. The bug also occurs in build 54535 (from the Buildbot) I tried it with a build 51611 and that error does not occur.%%%

Dani Epstein commented

2013-02-19 13:29:36 +01:00

Changed status to: 'Open'

Ton Roosendaal commented

2013-02-19 14:06:17 +01:00

%%%Editing 3d while render is always tricky and could be instable. For this case (select) things could just work though...
Brecht: crash is here:

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: 13 at address: 0x0000000000000000
[Switching to process 53428 thread 0xdf1b]
0x00000001015b7f2c in rna_MeshTextureFace_uv2_get (ptr=0x111df3118, values=0x111df3140) at rna_mesh.c:282
282 values- [x] = mtface->uv- [x][0];
(gdb) where

0 0x00000001015b7f2c in rna_MeshTextureFace_uv2_get (ptr=0x111df3118, values=0x111df3140) at rna_mesh.c:282
1 0x00000001015ccb63 in MeshTextureFace_uv2_get (ptr=0x111df3118, values=0x111df3140) at rna_mesh_gen.c:5234
2 0x00000001017cb71c in BL::MeshTextureFace::uv2 (this=0x111df3118) at RNA_blender_cpp.h:40876
3 0x00000001017c0128 in mikk_get_texture_coordinate (context=0x111df4580, uv=0x111df3188, face_num=41366, vert_num=1) at blender_mesh.cpp:89

On retry, it is here

[Switching to process 53434 thread 0xe70b]
0x00000001015ca7cb in MeshTessFace_vertices_raw_get (ptr=0x116e09d28, values=0x116e09d48) at rna_mesh_gen.c:4506
4506 values- [x] = (int)((&data->v1)[i]);
(gdb) where

0 0x00000001015ca7cb in MeshTessFace_vertices_raw_get (ptr=0x116e09d28, values=0x116e09d48) at rna_mesh_gen.c:4506
1 0x00000001017cb030 in BL::MeshTessFace::vertices_raw (this=0x116e09d28) at RNA_blender_cpp.h:40824

2 0x00000001017bff9d in mikk_get_position (context=0x116e0a580, P=0x116e09dc4, face_num=44989, vert_num=2) at blender_mesh.cpp:71

%%%

%%%Editing 3d while render is always tricky and could be instable. For this case (select) things could just work though... Brecht: crash is here: Program received signal EXC_BAD_ACCESS, Could not access memory. Reason: 13 at address: 0x0000000000000000 [Switching to process 53428 thread 0xdf1b] 0x00000001015b7f2c in rna_MeshTextureFace_uv2_get (ptr=0x111df3118, values=0x111df3140) at rna_mesh.c:282 282 values- [x] = mtface->uv- [x][0]; (gdb) where - 0 0x00000001015b7f2c in rna_MeshTextureFace_uv2_get (ptr=0x111df3118, values=0x111df3140) at rna_mesh.c:282 - 1 0x00000001015ccb63 in MeshTextureFace_uv2_get (ptr=0x111df3118, values=0x111df3140) at rna_mesh_gen.c:5234 - 2 0x00000001017cb71c in BL::MeshTextureFace::uv2 (this=0x111df3118) at RNA_blender_cpp.h:40876 - 3 0x00000001017c0128 in mikk_get_texture_coordinate (context=0x111df4580, uv=0x111df3188, face_num=41366, vert_num=1) at blender_mesh.cpp:89 On retry, it is here [Switching to process 53434 thread 0xe70b] 0x00000001015ca7cb in MeshTessFace_vertices_raw_get (ptr=0x116e09d28, values=0x116e09d48) at rna_mesh_gen.c:4506 4506 values- [x] = (int)((&data->v1)[i]); (gdb) where - 0 0x00000001015ca7cb in MeshTessFace_vertices_raw_get (ptr=0x116e09d28, values=0x116e09d48) at rna_mesh_gen.c:4506 - 1 0x00000001017cb030 in BL::MeshTessFace::vertices_raw (this=0x116e09d28) at RNA_blender_cpp.h:40824 # 2 0x00000001017bff9d in mikk_get_position (context=0x116e0a580, P=0x116e09dc4, face_num=44989, vert_num=2) at blender_mesh.cpp:71 %%%

Sergey Sharybin commented

2013-02-19 14:07:50 +01:00

%%%Developer note:
This is because of undo happens on select and this push will temporary set meshes' mface to NULL, making other threads fail if they needs tessfaces. In case of Cycles it's probably solvable by adding temporary meshes to separate Main. Not sure about BI render and baking, from first glance this areas seems to be robust for this issue, but i could be missing something.

Another (hackish) way could be to tag temp meshes as "skip for save"...%%%

%%%Developer note: This is because of undo happens on select and this push will temporary set meshes' mface to NULL, making other threads fail if they needs tessfaces. In case of Cycles it's probably solvable by adding temporary meshes to separate Main. Not sure about BI render and baking, from first glance this areas seems to be robust for this issue, but i could be missing something. Another (hackish) way could be to tag temp meshes as "skip for save"...%%%

Sergey Sharybin commented

2013-02-19 14:12:54 +01:00

%%%And assigning to Brecht back :)

P.S. This is actually an old issue, so probably could wait for after 2.66 (which we still hope will happen tomorrow), so maybe it makes more sense to fix the issue after release to avoid possible sideeffects caused by quick fix..%%%

%%%And assigning to Brecht back :) P.S. This is actually an old issue, so probably could wait for after 2.66 (which we still hope will happen tomorrow), so maybe it makes more sense to fix the issue after release to avoid possible sideeffects caused by quick fix..%%%

Dani Epstein commented

2013-02-19 14:16:23 +01:00

%%%FWIW, I don't think this is a serious issue, since from what I can see most people render with the "Full Screen" layout, making this much of a non-issue for them.%%%

Brecht Van Lommel commented

2013-04-12 17:29:03 +02:00

%%%Found a simpler fix for this case, just don't modify existing Mesh data structures for file save but rather a copy.

While this may crash mostly with Cycles, I think it's best to keep file save a read only operation because any threaded access can potentially crash even if they don't use temporary meshes, it's just less likely.%%%

%%%Found a simpler fix for this case, just don't modify existing Mesh data structures for file save but rather a copy. While this may crash mostly with Cycles, I think it's best to keep file save a read only operation because any threaded access can potentially crash even if they don't use temporary meshes, it's just less likely.%%%

Brecht Van Lommel commented

2013-04-12 17:29:04 +02:00

Changed status from 'Open' to: 'Resolved'

Brecht Van Lommel closed this issue

2013-04-12 17:29:04 +02:00

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

2.66 - segmentation fault when selecting an object during the initial phase of rendering #34322

2 0x00000001017bff9d in mikk_get_position (context=0x116e0a580, P=0x116e09dc4, face_num=44989, vert_num=2) at blender_mesh.cpp:71