Crash on startup with hackintosh and TBB #72015

Closed
opened 2019-11-28 16:25:29 +01:00 by Bob · 33 comments

System Information
Operating system: Catalina 10.15.1
Graphics card: RX 560

Blender Version
2.81

Crashing on launch. Here is a crash log, https://pastebin.com/WvU69fkn

**System Information** Operating system: Catalina 10.15.1 Graphics card: RX 560 **Blender Version** 2.81 Crashing on launch. Here is a crash log, https://pastebin.com/WvU69fkn
Author

Added subscriber: @iMacBob

Added subscriber: @iMacBob

#72165 was marked as duplicate of this issue

#72165 was marked as duplicate of this issue
Member

Added subscriber: @lichtwerk

Added subscriber: @lichtwerk
Member

Please attach the corresponding .blend file...

Please attach the corresponding .blend file...
Author

In #72015#820945, @lichtwerk wrote:
Please attach the corresponding .blend file...

The crash happens opening the application alone. There is no .blend file being loaded.

> In #72015#820945, @lichtwerk wrote: > Please attach the corresponding .blend file... The crash happens opening the application alone. There is no .blend file being loaded.

Added subscriber: @AnthonyEdlin

Added subscriber: @AnthonyEdlin

Understand we can't do much unless we can reproduce this issue.

According to crash log blender crashed on:
tbb::interface7::internal::task_arena_base::internal_max_concurrency(tbb::interface7::task_arena const*) + 50

tbb was only added recently to blender I think in quad remesher.

Some other things to try.

  1. Try a recent nightly build from here .
  2. Try starting blender with --factory-startup from the console.
  3. You don't say in your report if 2.80 crashes or not?
Understand we can't do much unless we can reproduce this issue. According to crash log blender crashed on: ```tbb::interface7::internal::task_arena_base::internal_max_concurrency(tbb::interface7::task_arena const*) + 50``` tbb was only added recently to blender I think in quad remesher. Some other things to try. 1. Try a recent nightly build from [here ](https://builder.blender.org/download/). 2. Try starting blender with --factory-startup from the console. 3. You don't say in your report if 2.80 crashes or not?
Philipp Oeser changed title from Crash on load to Crash on startup (TBB related) 2019-12-05 10:03:42 +01:00
Member

Added subscriber: @diba

Added subscriber: @diba
Author

I can confirm that the 2.82 alpha also crashes in relation to TBB. 2.80 runs as intended.

I can confirm that the 2.82 alpha also crashes in relation to TBB. 2.80 runs as intended.

Added subscribers: @WilliamReynish, @sebbas, @Sergey

Added subscribers: @WilliamReynish, @sebbas, @Sergey

Works fine on my MBP, both 2.81 and latest master builds from builder.blender.org.

@sebbas or @WilliamReynish should be able to test relatively easily as well.

Works fine on my MBP, both 2.81 and latest master builds from builder.blender.org. @sebbas or @WilliamReynish should be able to test relatively easily as well.

Both the latest master and 2.81 do not give me a startup crash either on macOS.

Both the latest master and 2.81 do not give me a startup crash either on macOS.

Added subscriber: @cruelplatypus67

Added subscriber: @cruelplatypus67

Same issue on latest alpha, any eta on this?

Same issue on latest alpha, any eta on this?
Member

Added subscriber: @ankitm

Added subscriber: @ankitm
Member

This comment was removed by @ankitm

*This comment was removed by @ankitm*
Member

This comment was removed by @ankitm

*This comment was removed by @ankitm*

Added subscriber: @Pick

Added subscriber: @Pick

This comment was removed by @Pick

*This comment was removed by @Pick*

Added subscriber: @brecht

Added subscriber: @brecht

@Pick, your crash is unrelated to this bug, please create a new report:
https://developer.blender.org/maniphest/task/edit/form/1/

@Pick, your crash is unrelated to this bug, please create a new report: https://developer.blender.org/maniphest/task/edit/form/1/

@ankitm, the links you provide are almost certainly unrelated to this specific crash. The terms you are searching for are generic crash signals that would happen with many types of crashes. System integrity protection seems unrelated since #72672 has the same crash log where it is enabled.

The most interesting thing here is that it crashes in TBB initialization where it detects the number of cores. We have been using TBB for previous releases as well, but started using it for sculpting as well now, which might affect for example static initialization order and cause problems related to that.

The issue might also be specific to certain processors, since it only crashes for some people. #72672 is a hackintosh system, and from the specs I'm guessing #72165 might be as well. The model information was left out from the crash log in this report, it would be interesting to know if that is a hackintosh as well. In #72672 there is an AMD Ryzen CPU but it is reported as an Intel i5 CPU by the system, that seems like a recipe for problems in this type of code.

We are using an older TBB version on macOS than Linux and Windows, 2018_U5 instead of 2019_U9. We should try upgrading and see if that happens to solve the problem.

@ankitm, the links you provide are almost certainly unrelated to this specific crash. The terms you are searching for are generic crash signals that would happen with many types of crashes. System integrity protection seems unrelated since #72672 has the same crash log where it is enabled. The most interesting thing here is that it crashes in TBB initialization where it detects the number of cores. We have been using TBB for previous releases as well, but started using it for sculpting as well now, which might affect for example static initialization order and cause problems related to that. The issue might also be specific to certain processors, since it only crashes for some people. #72672 is a hackintosh system, and from the specs I'm guessing #72165 might be as well. The model information was left out from the crash log in this report, it would be interesting to know if that is a hackintosh as well. In #72672 there is an AMD Ryzen CPU but it is reported as an Intel i5 CPU by the system, that seems like a recipe for problems in this type of code. We are using an older TBB version on macOS than Linux and Windows, 2018_U5 instead of 2019_U9. We should try upgrading and see if that happens to solve the problem.

Added subscriber: @girafic

Added subscriber: @girafic

@brecht I have the same issue, running a hackintosh system with AMD Ryzen CPU. Upgrading TBB to 2019_U9 or v2020.1 could solve the problem.

@brecht I have the same issue, running a hackintosh system with AMD Ryzen CPU. Upgrading TBB to 2019_U9 or v2020.1 could solve the problem.
Brecht Van Lommel changed title from Crash on startup (TBB related) to Crash on startup with hackintosh and TBB 2020-01-22 10:50:55 +01:00

It is working now 👍. Built the latest versions of 2.81a and 2.82 on the AMD hackintosh.

It is working now 👍. Built the latest versions of 2.81a and 2.82 on the AMD hackintosh.

Changed status from 'Needs Developer To Reproduce' to: 'Needs User Info'

Changed status from 'Needs Developer To Reproduce' to: 'Needs User Info'

The latest macOS 2.82 builds now also have TBB 2019_U9:
https://builder.blender.org/download/

Can others verify if the issue is fixed in those builds?

The latest macOS 2.82 builds now also have TBB 2019_U9: https://builder.blender.org/download/ Can others verify if the issue is fixed in those builds?

it is crashing for me with the builds from https://builder.blender.org/download/

it is crashing for me with the builds from https://builder.blender.org/download/

After 517870a I'm getting the same type of error on my Linux desktop with AMD Ryzen.

Thread 1 "blender" received signal SIGSEGV, Segmentation fault.
0x000055555cd5f39e in tbb::interface7::internal::task_arena_base::internal_max_concurrency(tbb::interface7::task_arena const*) ()
(gdb) bt
- 0  0x000055555cd5f39e in tbb::interface7::internal::task_arena_base::internal_max_concurrency(tbb::interface7::task_arena const*) ()
- 1  0x0000555557272b87 in mkldnn::impl::cpu::(anonymous namespace)::get_cache_size(int, bool) ()
- 2  0x0000555557272be6 in _GLOBAL__sub_I_jit_avx512_common_conv_winograd_kernel_f32.cpp ()
- 3  0x000055555cfbc47d in __libc_csu_init ()
- 4  0x00007ffff5fa2b28 in __libc_start_main (main=0x555557221730 <main>, argc=1, argv=0x7fffffffdb38, init=0x55555cfbc430 <__libc_csu_init>, fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffdb28) at ../csu/libc-start.c:266
- 5  0x00005555572d84fa in _start ()

This looks like a static initialization order problem, with TBB not being initialized before mkl uses it. The macOS backtraces do not contain mkl, so it's not clear if the problem is exactly the same.

After 517870a I'm getting the same type of error on my Linux desktop with AMD Ryzen. ``` Thread 1 "blender" received signal SIGSEGV, Segmentation fault. 0x000055555cd5f39e in tbb::interface7::internal::task_arena_base::internal_max_concurrency(tbb::interface7::task_arena const*) () (gdb) bt - 0 0x000055555cd5f39e in tbb::interface7::internal::task_arena_base::internal_max_concurrency(tbb::interface7::task_arena const*) () - 1 0x0000555557272b87 in mkldnn::impl::cpu::(anonymous namespace)::get_cache_size(int, bool) () - 2 0x0000555557272be6 in _GLOBAL__sub_I_jit_avx512_common_conv_winograd_kernel_f32.cpp () - 3 0x000055555cfbc47d in __libc_csu_init () - 4 0x00007ffff5fa2b28 in __libc_start_main (main=0x555557221730 <main>, argc=1, argv=0x7fffffffdb38, init=0x55555cfbc430 <__libc_csu_init>, fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffffffdb28) at ../csu/libc-start.c:266 - 5 0x00005555572d84fa in _start () ``` This looks like a static initialization order problem, with TBB not being initialized before mkl uses it. The macOS backtraces do not contain mkl, so it's not clear if the problem is exactly the same.

I've committed a workaround which worked for Brecht. Buildbots are still busy building, but give it a try when builds a ready!

P.S. This is a workaround because there is never an elegant solution for static order initialization fiasco. This is one of the reasons TBB documentations says use dynamic library. Oh well.

I've committed a workaround which worked for Brecht. Buildbots are still busy building, but give it a try when builds a ready! P.S. This is a workaround because there is never an elegant solution for static order initialization fiasco. This is one of the reasons TBB documentations says use dynamic library. Oh well.

@Sergey for me it is working now with the builds from https://builder.blender.org/download/ 👍

@Sergey for me it is working now with the builds from https://builder.blender.org/download/ 👍

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

Changed status from 'Needs User Info' to: 'Resolved'
Brecht Van Lommel self-assigned this 2020-01-24 18:16:09 +01:00

Thanks for testing.

Thanks for testing.
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
No Milestone
No project
No Assignees
11 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#72015
No description provided.