Blender crashes and destroys my scene #82353

New Issue

Nis Wegmann · 2020-11-03T09:32:06+01:00

Nis Wegmann commented

2020-11-03 09:32:06 +01:00

System Information
Operating system: Mac OS Catalina
Graphics card: AMD Radeon R9 M290X 2 GB

Blender Version
Broken: 2.90.1
Worked: (newest version of Blender that worked as expected)

Short description of error

I've reported this bug before. While using Blender, Blender crashes, and when doing so it destroys my scene so it can't be opened again. When attempting to open the scene again, Blender crashes.

I use Blender everyday for commercial work. This bug happens about once per week.

I've attached a project that crashes immediately, if you attempt open it.

Maybe, if someone could debug Blender while opening the project, they could see what goes wrong.

Exact steps for others to reproduce the error
Based on the default startup or an attached .blend file (as simple as possible).juno.blend

**System Information** Operating system: Mac OS Catalina Graphics card: AMD Radeon R9 M290X 2 GB **Blender Version** Broken: 2.90.1 Worked: (newest version of Blender that worked as expected) **Short description of error** I've reported this bug before. While using Blender, Blender crashes, and when doing so it destroys my scene so it can't be opened again. When attempting to open the scene again, Blender crashes. I use Blender everyday for commercial work. This bug happens about once per week. I've attached a project that crashes immediately, if you attempt open it. Maybe, if someone could debug Blender while opening the project, they could see what goes wrong. **Exact steps for others to reproduce the error** Based on the default startup or an attached .blend file (as simple as possible).[juno.blend](https://archive.blender.org/developer/F9178899/juno.blend)

Nis Wegmann commented

2020-11-03 09:32:06 +01:00

Added subscriber: @nis

blender-admin commented

2020-11-03 09:32:06 +01:00

#80272 was marked as duplicate of this issue

blender-admin commented

2020-11-03 09:32:06 +01:00

#82504 was marked as duplicate of this issue

Nis Wegmann commented

2020-11-03 09:33:05 +01:00

I have the exact same problem on my other Macs. A brand new iMac and a MacBook Pro from 2017.

Nis Wegmann commented

2020-11-03 10:01:47 +01:00

This particular scene can be repaired by creating a new Blender project, appending all scenes from the juno.blend file to the new project, finding the "Keys - 18.5 mm" scene and deleting all the contents in the "Key - Black - X - 18.5mm" collection.

Robert Guetzkow commented

2020-11-03 10:05:04 +01:00

Added subscriber: @rjg

Robert Guetzkow commented

2020-11-03 10:05:04 +01:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Robert Guetzkow commented

2020-11-03 10:05:04 +01:00

Unfortunately, we cannot identify the cause of the corruption based on a broken file alone. If you can, please explain the steps you've made that resulted in Blender producing the broken file.

Nis Wegmann commented

2020-11-03 12:34:25 +01:00

The bug happens while using Blender in random situations. Sometimes while modelling in edit mode, sometimes while switching the active scene. I think it is related to collections somehow.

I've attached a project that I've created by appending all scenes from the juno.blend file that I uploaded earlier.

If you try and enable the collection (i.e. make it visible) called "Key - Black - X - 18.5mm" blender will crash immediately. Somehow, Blender has corrupted the object that is inside of the collection.juno_bug_report.blend

The bug happens while using Blender in random situations. Sometimes while modelling in edit mode, sometimes while switching the active scene. I think it is related to collections somehow. I've attached a project that I've created by appending all scenes from the juno.blend file that I uploaded earlier. If you try and enable the collection (i.e. make it visible) called "Key - Black - X - 18.5mm" blender will crash immediately. Somehow, Blender has corrupted the object that is inside of the collection.[juno_bug_report.blend](https://archive.blender.org/developer/F9180561/juno_bug_report.blend)

Evan Wilson commented

2020-11-03 15:52:58 +01:00

Added subscriber: @EAW

Evan Wilson commented

2020-11-03 15:52:58 +01:00

Triaging note: the objects that are corrupted are in disabled collections seems to be a running theme in these 2.83+ MacOS corruption reports.

Robert Guetzkow commented

2020-11-03 16:13:28 +01:00

Added subscriber: @ankitm

Robert Guetzkow commented

2020-11-03 16:13:28 +01:00

@EAW Yes, this is very much relevant. See #79665 and #79451. Ideally some Blender developer with a Mac and a lot of time (@ankitm ?), should work through the Donut tutorial, since this has triggered the corruption for several people. It currently provides the only instructions we have that might lead to the source of the problem.

Ankit Meel commented

2020-11-04 12:57:41 +01:00

Removed subscriber: @ankitm

Nis Wegmann commented

2020-11-05 09:56:04 +01:00

Same bug in another project.

Open the collection "Modular - Panel - DCO - Hi" in the "High-pass" collection and Blender will crash.

Why don't you run Blender in debugging mode and see where Blender crashes in the code, when opening the collection. Could be an idea.

It's a pretty serious bug that makes using Blender on macOS a nightmare; Blender crashes and destroys your scene/objects. Would be great if someone would look into this. Never experienced this bug in Blender 2.8.

juno_model.blend

Same bug in another project. Open the collection "Modular - Panel - DCO - Hi" in the "High-pass" collection and Blender will crash. Why don't you run Blender in debugging mode and see where Blender crashes in the code, when opening the collection. Could be an idea. It's a pretty serious bug that makes using Blender on macOS a nightmare; Blender crashes and destroys your scene/objects. Would be great if someone would look into this. Never experienced this bug in Blender 2.8. [juno_model.blend](https://archive.blender.org/developer/F9206267/juno_model.blend)

Robert Guetzkow commented

2020-11-05 10:40:13 +01:00

@nis Thank you for the file. The difficulty is not to reproduce the crash, that part is easy. It's finding what caused the corruption in the first place.

Nis Wegmann commented

2020-11-05 19:31:51 +01:00

Ah, okay, I see. I'll try and see if I can find a way reliable way to reproduce it. Maybe I can try that donut tutorial. I absolutely love Blender -- and that bug is killing me :)

Ah, okay, I see. I'll try and see if I can find a way reliable way to reproduce it. Maybe I can try that donut tutorial. I absolutely *love* Blender -- and that bug is killing me :)

Robert Guetzkow commented

2020-11-05 20:01:30 +01:00

@nis I'm really sorry about this. If you find a way to reproduce this, it would help a lot.

For reference, the stack trace of the crash after activating the Modular - Panel - DCO - Hi in the High-pass collection is:

mesh_calc_normals_poly_prepare_cb(void * userdata, const int pidx, const TaskParallelTLS * UNUSED_tls) Line 242	C
BLI_task_parallel_range(const int start, const int stop, void * userdata, void(*)(void *, const int, const TaskParallelTLS *) func, const TaskParallelSettings * settings) Line 141	C++
BKE_mesh_calc_normals_poly(MVert * mverts, float- [x] * r_vertnors, int numVerts, const MLoop * mloop, const MPoly * mpolys, int numLoops, int numPolys, float- [x] * r_polynors, const bool only_face_normals) Line 369	C
BKE_mesh_calc_normals_split_ex(Mesh * mesh, MLoopNorSpaceArray * r_lnors_spacearr) Line 1818	C
BKE_mesh_calc_normals_split(Mesh * mesh) Line 1847	C
mesh_calc_modifier_final_normals(const Mesh * mesh_input, const CustomData_MeshMasks * final_datamask, const bool sculpt_dyntopo, Mesh * mesh_final) Line 817	C
mesh_calc_modifiers(Depsgraph * depsgraph, Scene * scene, Object * ob, int useDeform, const bool need_mapping, const CustomData_MeshMasks * dataMask, const int index, const bool use_cache, const bool allow_shared_mesh, Mesh * * r_deform, Mesh * * r_final) Line 1344	C
mesh_build_data(Depsgraph * depsgraph, Scene * scene, Object * ob, const CustomData_MeshMasks * dataMask, const bool need_mapping) Line 1806	C
makeDerivedMesh(Depsgraph * depsgraph, Scene * scene, Object * ob, BMEditMesh * em, const CustomData_MeshMasks * dataMask) Line 1927	C
BKE_object_handle_data_update(Depsgraph * depsgraph, Scene * scene, Object * ob) Line 194	C
BKE_object_eval_uber_data(Depsgraph * depsgraph, Scene * scene, Object * ob) Line 385	C
 	[External Code]	
blender::deg::`anonymous namespace'::evaluate_node(const blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * operation_node) Line 116	C++
blender::deg::`anonymous namespace'::deg_task_run_func(TaskPool * pool, void * taskdata) Line 128	C++
Task::()::__l2::<Lambdafunktion>() Line 118	C++
tbb::interface7::internal::delegated_function<void <Lambdafunktion>(void) const ,void>::operator()() Line 94	C++
 	[External Code]	
tbb::interface7::internal::isolate_impl<void,void <Lambdafunktion>(void) const>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 161	C++
tbb::interface7::this_task_arena::isolate<void <Lambdafunktion>(void)>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 396	C++
Task::operator()() Line 122	C++
tbb_task_pool_run(TaskPool * pool, Task && task) Line 228	C++
BLI_task_pool_push(TaskPool * pool, void(*)(TaskPool *, void *) run, void * taskdata, bool free_taskdata, void(*)(TaskPool *, void *) freedata) Line 485	C++
blender::deg::`anonymous namespace'::schedule_node_to_pool(blender::deg::OperationNode * node, const int UNUSED_thread_id, TaskPool * pool) Line 75	C++
blender::deg::`anonymous namespace'::schedule_node<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, bool dec_parents, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 277	C++
blender::deg::`anonymous namespace'::schedule_children<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 294	C++
blender::deg::`anonymous namespace'::schedule_node<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, bool dec_parents, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 271	C++
blender::deg::`anonymous namespace'::schedule_children<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 294	C++
blender::deg::`anonymous namespace'::deg_task_run_func(TaskPool * pool, void * taskdata) Line 129	C++
Task::()::__l2::<Lambdafunktion>() Line 118	C++
tbb::interface7::internal::delegated_function<void <Lambdafunktion>(void) const ,void>::operator()() Line 94	C++
 	[External Code]	
tbb::interface7::internal::isolate_impl<void,void <Lambdafunktion>(void) const>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 161	C++
tbb::interface7::this_task_arena::isolate<void <Lambdafunktion>(void)>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 396	C++
Task::operator()() Line 122	C++
tbb_task_pool_run(TaskPool * pool, Task && task) Line 228	C++
tbb_task_pool_work_and_wait(TaskPool * pool) Line 240	C++
BLI_task_pool_work_and_wait(TaskPool * pool) Line 500	C++
blender::deg::deg_evaluate_on_refresh(blender::deg::Depsgraph * graph) Line 399	C++
deg_flush_updates_and_refresh(blender::deg::Depsgraph * deg_graph) Line 59	C++
DEG_evaluate_on_refresh(Depsgraph * graph) Line 74	C++
scene_graph_update_tagged(Depsgraph * depsgraph, Main * bmain, bool only_if_tagged) Line 1720	C
BKE_scene_graph_update_tagged(Depsgraph * depsgraph, Main * bmain) Line 1758	C
wm_event_do_depsgraph(bContext * C, bool is_after_open_file) Line 357	C
wm_event_do_refresh_wm_and_depsgraph(bContext * C) Line 384	C
wm_event_do_notifiers(bContext * C) Line 562	C
WM_main(bContext * C) Line 484	C
main(int argc, const unsigned char * * UNUSED_argv_c) Line 526	C
 	[External Code]

When you run the following Python script that validates the meshes in the collection, you will that it prints many errors to the console. They indicate that the mesh data modular_panel_dco_hi in object modular_panel_dco_hi is corrupted.

import bpy

for obj in bpy.data.collections["Modular - Panel - DCO - Hi"].objects:

if obj.data is not None:
print(f"obj: {obj.name}, data: {obj.data.name}")
obj.data.validate(verbose=True)

@nis I'm really sorry about this. If you find a way to reproduce this, it would help a lot. For reference, the stack trace of the crash after activating the *Modular - Panel - DCO - Hi* in the *High-pass* collection is: ```lines mesh_calc_normals_poly_prepare_cb(void * userdata, const int pidx, const TaskParallelTLS * UNUSED_tls) Line 242 C BLI_task_parallel_range(const int start, const int stop, void * userdata, void(*)(void *, const int, const TaskParallelTLS *) func, const TaskParallelSettings * settings) Line 141 C++ BKE_mesh_calc_normals_poly(MVert * mverts, float- [x] * r_vertnors, int numVerts, const MLoop * mloop, const MPoly * mpolys, int numLoops, int numPolys, float- [x] * r_polynors, const bool only_face_normals) Line 369 C BKE_mesh_calc_normals_split_ex(Mesh * mesh, MLoopNorSpaceArray * r_lnors_spacearr) Line 1818 C BKE_mesh_calc_normals_split(Mesh * mesh) Line 1847 C mesh_calc_modifier_final_normals(const Mesh * mesh_input, const CustomData_MeshMasks * final_datamask, const bool sculpt_dyntopo, Mesh * mesh_final) Line 817 C mesh_calc_modifiers(Depsgraph * depsgraph, Scene * scene, Object * ob, int useDeform, const bool need_mapping, const CustomData_MeshMasks * dataMask, const int index, const bool use_cache, const bool allow_shared_mesh, Mesh * * r_deform, Mesh * * r_final) Line 1344 C mesh_build_data(Depsgraph * depsgraph, Scene * scene, Object * ob, const CustomData_MeshMasks * dataMask, const bool need_mapping) Line 1806 C makeDerivedMesh(Depsgraph * depsgraph, Scene * scene, Object * ob, BMEditMesh * em, const CustomData_MeshMasks * dataMask) Line 1927 C BKE_object_handle_data_update(Depsgraph * depsgraph, Scene * scene, Object * ob) Line 194 C BKE_object_eval_uber_data(Depsgraph * depsgraph, Scene * scene, Object * ob) Line 385 C [External Code] blender::deg::`anonymous namespace'::evaluate_node(const blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * operation_node) Line 116 C++ blender::deg::`anonymous namespace'::deg_task_run_func(TaskPool * pool, void * taskdata) Line 128 C++ Task::()::__l2::<Lambdafunktion>() Line 118 C++ tbb::interface7::internal::delegated_function<void <Lambdafunktion>(void) const ,void>::operator()() Line 94 C++ [External Code] tbb::interface7::internal::isolate_impl<void,void <Lambdafunktion>(void) const>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 161 C++ tbb::interface7::this_task_arena::isolate<void <Lambdafunktion>(void)>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 396 C++ Task::operator()() Line 122 C++ tbb_task_pool_run(TaskPool * pool, Task && task) Line 228 C++ BLI_task_pool_push(TaskPool * pool, void(*)(TaskPool *, void *) run, void * taskdata, bool free_taskdata, void(*)(TaskPool *, void *) freedata) Line 485 C++ blender::deg::`anonymous namespace'::schedule_node_to_pool(blender::deg::OperationNode * node, const int UNUSED_thread_id, TaskPool * pool) Line 75 C++ blender::deg::`anonymous namespace'::schedule_node<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, bool dec_parents, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 277 C++ blender::deg::`anonymous namespace'::schedule_children<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 294 C++ blender::deg::`anonymous namespace'::schedule_node<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, bool dec_parents, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 271 C++ blender::deg::`anonymous namespace'::schedule_children<void __cdecl(blender::deg::OperationNode *,int,TaskPool *),TaskPool *>(blender::deg::`anonymous-namespace'::DepsgraphEvalState * state, blender::deg::OperationNode * node, void(*)(blender::deg::OperationNode *, int, TaskPool *) schedule_function, TaskPool * <schedule_function_args_0>) Line 294 C++ blender::deg::`anonymous namespace'::deg_task_run_func(TaskPool * pool, void * taskdata) Line 129 C++ Task::()::__l2::<Lambdafunktion>() Line 118 C++ tbb::interface7::internal::delegated_function<void <Lambdafunktion>(void) const ,void>::operator()() Line 94 C++ [External Code] tbb::interface7::internal::isolate_impl<void,void <Lambdafunktion>(void) const>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 161 C++ tbb::interface7::this_task_arena::isolate<void <Lambdafunktion>(void)>(const Task::()::__l2::void <Lambdafunktion>(void) & f) Line 396 C++ Task::operator()() Line 122 C++ tbb_task_pool_run(TaskPool * pool, Task && task) Line 228 C++ tbb_task_pool_work_and_wait(TaskPool * pool) Line 240 C++ BLI_task_pool_work_and_wait(TaskPool * pool) Line 500 C++ blender::deg::deg_evaluate_on_refresh(blender::deg::Depsgraph * graph) Line 399 C++ deg_flush_updates_and_refresh(blender::deg::Depsgraph * deg_graph) Line 59 C++ DEG_evaluate_on_refresh(Depsgraph * graph) Line 74 C++ scene_graph_update_tagged(Depsgraph * depsgraph, Main * bmain, bool only_if_tagged) Line 1720 C BKE_scene_graph_update_tagged(Depsgraph * depsgraph, Main * bmain) Line 1758 C wm_event_do_depsgraph(bContext * C, bool is_after_open_file) Line 357 C wm_event_do_refresh_wm_and_depsgraph(bContext * C) Line 384 C wm_event_do_notifiers(bContext * C) Line 562 C WM_main(bContext * C) Line 484 C main(int argc, const unsigned char * * UNUSED_argv_c) Line 526 C [External Code] ``` When you run the following Python script that validates the meshes in the collection, you will that it prints many errors to the console. They indicate that the mesh data `modular_panel_dco_hi` in object `modular_panel_dco_hi` is corrupted. ``` import bpy for obj in bpy.data.collections["Modular - Panel - DCO - Hi"].objects: ``` if obj.data is not None: print(f"obj: {obj.name}, data: {obj.data.name}") obj.data.validate(verbose=True) ```

Roger B commented

2020-11-05 21:03:06 +01:00

Added subscriber: @rboxman

Roger B commented

2020-11-05 21:03:06 +01:00

There's a chance the data was written to the file in a corrupted state already. Blender crashing may have nothing to do with it. Blender won't be touching the file during its crash. Even if the crash happened during auto-save, blender will first write to a temp file and only replace the original once completed.

Are you using any addons to help model that particular mesh by chance? They could be corrupting the mesh in such a way to only manifest on load (and perhaps even cause the crash in the first place during some later operation)

There's a chance the data was written to the file in a corrupted state already. Blender crashing may have nothing to do with it. Blender won't be touching the file during its crash. Even if the crash happened during auto-save, blender will first write to a temp file and only replace the original once completed. Are you using any addons to help model that particular mesh by chance? They could be corrupting the mesh in such a way to only manifest on load (and perhaps even cause the crash in the first place during some later operation)

Nis Wegmann commented

2020-11-06 10:34:33 +01:00

The only add-ons that I use are Quad Remesher and UV Pack Master. I'll try and disable those for a week and see if the bug still arises.

Yevgeny Makarov commented

2020-11-09 12:44:46 +01:00

Added subscriber: @jenkm

Yevgeny Makarov commented

2020-11-09 12:44:46 +01:00

@rjg - ~~There is a test script bl_mesh_validate.py, does it give a correct report?~~
~~I get a bunch of errors like "Loop XX is unused" when I run it.~~

This might be also related to #80203 (Crash when changing torus properties ).

So, it is likely that Blender initially creates a corrupted mesh.

@rjg - ~~There is a test script `bl_mesh_validate.py`, does it give a correct report?~~ ~~I get a bunch of errors like "Loop XX is unused" when I run it.~~ This might be also related to #80203 (Crash when changing torus properties ). So, it is likely that Blender initially creates a corrupted mesh.

Robert Guetzkow commented

2020-11-09 14:09:49 +01:00

Added subscriber: @GGK

Nis Wegmann commented

2020-11-10 14:07:46 +01:00

Today while working on rendering materials, Blender crashed many times again (approximately every 30 minutes). The project is super simple, only a sphere, a plane, and some lights. The last crash that happened destroyed the scene in a way that it can't even be recovered by creating a new scene and appending the scenes from the file that blender destroyed. Moreover, the autosave is also destroyed. Blender simply crashes when attempting to recover the autosave.

Point is, Blender crashes every once in a while on macOS, often destroying your scene completely, in a way that it can't be recovered, not even by using the auto save. It happens consistently on three different Macs that I use. My setup is simple and I've disabled all add-ons. None of my other programs crash (Substance painter / Designer + Affinity Photo / Designer / Publisher + MOI 3D).

I love Blender, but this bug almost renders Blender unusable on macOS. Constantly having to worry if Blender will crash your scene and destroy it is just a hell.

I really think one of the macOS developers should look into this.

Here's the scene in an openable version:

render_old.blend

Today while working on rendering materials, Blender crashed many times again (approximately every 30 minutes). The project is super simple, only a sphere, a plane, and some lights. The last crash that happened destroyed the scene in a way that it can't even be recovered by creating a new scene and appending the scenes from the file that blender destroyed. Moreover, the autosave is also destroyed. Blender simply crashes when attempting to recover the autosave. Point is, Blender crashes every once in a while on macOS, often destroying your scene completely, in a way that it can't be recovered, not even by using the auto save. It happens consistently on three different Macs that I use. My setup is simple and I've disabled all add-ons. None of my other programs crash (Substance painter / Designer + Affinity Photo / Designer / Publisher + MOI 3D). I love Blender, but this bug almost renders Blender unusable on macOS. Constantly having to worry if Blender will crash your scene and destroy it is just a hell. I really think one of the macOS developers should look into this. Here's the scene in an openable version: [render_old.blend](https://archive.blender.org/developer/F9252568/render_old.blend)

Ivo Wejgaard commented

2020-12-05 13:13:57 +01:00

Added subscriber: @info-360

Ivo Wejgaard commented

2020-12-05 13:13:57 +01:00

I'm sure the devs are aware of this, but I just want to make sure - these problems on macs, destroying objects/files and lot's of crashes, only started to happen with 2.83.
2.82 was and still is super stable. It's the last version which is usable for Mac users.

So the cause of these issues must be something which fundamentally changed with 2.83.
Wouldn't that narrow down possible culprits?
(I guess I massively underestimate the amount of code that changes with every new version - but I still kind of wonder?)

I'm sure the devs are aware of this, but I just want to make sure - these problems on macs, destroying objects/files and lot's of crashes, only started to happen with 2.83. 2.82 was and still is super stable. It's the last version which is usable for Mac users. So the cause of these issues must be something which fundamentally changed with 2.83. Wouldn't that narrow down possible culprits? (I guess I massively underestimate the amount of code that changes with every new version - but I still kind of wonder?)

Yevgeny Makarov commented

2020-12-05 13:24:08 +01:00

Most likely the reason is in the new Undo system, so you can try using "Undo Legacy" preference.

Richard Antalik commented

2020-12-08 13:35:37 +01:00

Added subscribers: @kylin.ye, @iss

Nis Wegmann commented

2020-12-08 13:53:47 +01:00

In #82353#1068822, @info-360 wrote:
I'm sure the devs are aware of this, but I just want to make sure - these problems on macs, destroying objects/files and lot's of crashes, only started to happen with 2.83.
2.82 was and still is super stable. It's the last version which is usable for Mac users.

So the cause of these issues must be something which fundamentally changed with 2.83.
Wouldn't that narrow down possible culprits?
(I guess I massively underestimate the amount of code that changes with every new version - but I still kind of wonder?)

Yes, true. 2.82 was rock solid -- in my experience -- on macOS. The crashes that destroys your files and autosave files started with 2.83.

> In #82353#1068822, @info-360 wrote: > I'm sure the devs are aware of this, but I just want to make sure - these problems on macs, destroying objects/files and lot's of crashes, only started to happen with 2.83. > 2.82 was and still is super stable. It's the last version which is usable for Mac users. > > So the cause of these issues must be something which fundamentally changed with 2.83. > Wouldn't that narrow down possible culprits? > (I guess I massively underestimate the amount of code that changes with every new version - but I still kind of wonder?) Yes, true. 2.82 was rock solid -- in my experience -- on macOS. The crashes that destroys your files and autosave files started with 2.83.

Nis Wegmann commented

2020-12-28 17:44:47 +01:00

I can confirm that enabling "undo legacy" in the "Experimental" preferences does not solve the problem. I've also disabled iCloud backup, which doesn't solve the problem either. Blender still crashes randomly and destroys the scene as a side-effect.

Ankit Meel commented

2021-01-05 20:31:44 +01:00

Added subscriber: @ankitm

Robert Guetzkow commented

2021-01-13 13:20:25 +01:00

@nis Although you say that this problem was still occurring the the Legacy Undo, could you please check if the current daily build of 2.92 still has the exact same problem. There have been fixes made to the dependency graph that or may not be related to your problem.

@nis Although you say that this problem was still occurring the the Legacy Undo, could you please check if the current [daily build of 2.92 ](https://builder.blender.org/download/) still has the exact same problem. There have been fixes made to the dependency graph that or may not be related to your problem.

John Meyer commented

2021-02-24 22:32:41 +01:00

Added subscriber: @cowzrblue

John Meyer commented

2021-02-24 22:32:41 +01:00

I can confirm, trying to enable the collection, blender will crash immediately. It looks like mesh_build_data causes the crash on my 2015 iMac 10.15.7. I'm using 2.92 and have tried both the undo legacy. No resolve at this time.

Ivo Wejgaard commented

2021-03-02 11:45:57 +01:00

I just tried the latest 2.93 Alpha version. Played around with Geometry Nodes. It took about 2 hours until I had a crash which corrupted the save file. So the same problem is still there.

To be honest, this is becoming a really huge issue now. Geometry Nodes are a total game changer. Not being able to use them will not only be super disappointing but it will also block Mac users from all particle & scatter related add-on's in the near future. Cause all of them will switch to Geometry Nodes.

I just tried the latest 2.93 Alpha version. Played around with Geometry Nodes. It took about 2 hours until I had a crash which corrupted the save file. So the same problem is still there. To be honest, this is becoming a really huge issue now. Geometry Nodes are a total game changer. Not being able to use them will not only be super disappointing but it will also block Mac users from all particle & scatter related add-on's in the near future. Cause all of them will switch to Geometry Nodes.

Ankit Meel commented

2021-03-29 13:54:35 +02:00

Please try to create a set of exact steps that lead to a corrupted mesh. Just having a final corrupted mesh is less helpful.

Ivo Wejgaard commented

2021-04-16 12:52:46 +02:00

In #82353#1137812, @ankitm wrote:
Please try to create a set of exact steps that lead to a corrupted mesh. Just having a final corrupted mesh is less helpful.

Unfortunately that is impossible in the cases I encountered. The problems never occurred while I was working on a mesh. They happened when I created one or several meshes and put them in a collection. Let's call it collection A. Then I deactivated collection A and went on working on other meshes in other collections B and C. Later, when I tried to activate collection A again, Blender would crash. Activating collection A will always cause Blender to crash from then on. Even if it's appended into another project.
In a big project with 20 or more collections (each of which has about 20 objects) several of the collections would have this issue, after some time. The longer I worked on a project, the more collections got corrupted. (But, I think it was only collections which were deactivated, if I remember correctly.)
And since these issues happen, while I'm working on something totally different and I just notice the problem once I go back to activate a collection created (and deactivated) earlier on, I never know when and why the actual problem happens.

Like other people I also encountered the issue of Blender crashing and completely corrupting the project file. If I tried to load it, it would instantly crash. And if I tried to append the objects and collections it would make the newly created project crash as well.
But the issues with activating deactivated collections happened more often to me.

For now I'm only working on 2.82. I simply do not have the time and the nerves to worry about these issues.....

> In #82353#1137812, @ankitm wrote: > Please try to create a set of exact steps that lead to a corrupted mesh. Just having a final corrupted mesh is less helpful. Unfortunately that is impossible in the cases I encountered. The problems never occurred while I was working on a mesh. They happened when I created one or several meshes and put them in a collection. Let's call it collection A. Then I deactivated collection A and went on working on other meshes in other collections B and C. Later, when I tried to activate collection A again, Blender would crash. Activating collection A will always cause Blender to crash from then on. Even if it's appended into another project. In a big project with 20 or more collections (each of which has about 20 objects) several of the collections would have this issue, after some time. The longer I worked on a project, the more collections got corrupted. (But, I think it was only collections which were deactivated, if I remember correctly.) And since these issues happen, while I'm working on something totally different and I just notice the problem once I go back to activate a collection created (and deactivated) earlier on, I never know when and why the actual problem happens. Like other people I also encountered the issue of Blender crashing and completely corrupting the project file. If I tried to load it, it would instantly crash. And if I tried to append the objects and collections it would make the newly created project crash as well. But the issues with activating deactivated collections happened more often to me. For now I'm only working on 2.82. I simply do not have the time and the nerves to worry about these issues.....

Philipp Oeser commented

2021-05-07 13:06:02 +02:00

Added subscriber: @lichtwerk

Philipp Oeser commented

2021-05-07 13:06:02 +02:00

This sounds like the donut case in #79665 (Blender crashes when enabling "Donut" Collection).
This really seems to be a thing, it is just too bad with all these cases we still have not been able to make a reproducable case...

This sounds like the donut case in #79665 (Blender crashes when enabling "Donut" Collection). This really seems to be a thing, it is just too bad with all these cases we still have not been able to make a reproducable case...

Philipp Oeser commented

2021-05-28 08:33:43 +02:00

The issue reported can't be reproduced so far, and has been moved to the *BF Blender: Unconfirmed* project.
If 2 other people aren't able to reproduce the bug (following reproduction steps) after 2 weeks, it will be closed.

We appreciate the effort that goes into making bug reports, but to be able to fix bugs we have to be able to redo them.
If there is anything you can provide to help others to reproduce the problem, be sure to include it.

``` The issue reported can't be reproduced so far, and has been moved to the *BF Blender: Unconfirmed* project. If 2 other people aren't able to reproduce the bug (following reproduction steps) after 2 weeks, it will be closed. We appreciate the effort that goes into making bug reports, but to be able to fix bugs we have to be able to redo them. If there is anything you can provide to help others to reproduce the problem, be sure to include it. ```

Ivo Wejgaard commented

2021-05-28 08:47:20 +02:00

I just wonder - is there a Dev who used a recent Blender version (2.83 or newer) on Mac for at least 2 hours? Creating a few objects and collections and playing around?
For me that was the 'easiest' way to reproduce the issue. ;-)

I'm pretty sure one of the Devs must have done that. Well, at least I hope....

I just wonder - is there a Dev who used a recent Blender version (2.83 or newer) on Mac for at least 2 hours? Creating a few objects and collections and playing around? For me that was the 'easiest' way to reproduce the issue. ;-) I'm pretty sure one of the Devs must have done that. Well, at least I hope....

Philipp Oeser commented

2021-07-02 12:27:17 +02:00

Changed status from 'Needs User Info' to: 'Archived'

Philipp Oeser closed this issue

2021-07-02 12:27:17 +02:00

Philipp Oeser commented

2021-07-02 12:27:17 +02:00

It seems this report is stuck.

And while there is not really doubt that this is indeed an issue, it is not really useful to have this report going further without clearer reproduction steps.
Will archive for now (but we can always reopen if further information can be provided).

It seems this report is stuck. And while there is not really doubt that this is indeed an issue, it is not really useful to have this report going further without clearer reproduction steps. Will archive for now (but we can always reopen if further information can be provided).

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

Blender crashes and destroys my scene #82353