blender
/
blender
134
398
Fork 579
Code Issues 6.3k Pull Requests 755 Projects 19 Wiki Activity

Fix USD Export crash with complex scenes, due to race condition during depsgraph building. #113537

Merged
Bastien Montagne merged 1 commits from mont29/blender:tmp-usd-fix-deg-export-crash into blender-v4.0-release 2023-10-12 19:10:46 +02:00
Conversation 4 Commits 1 Files Changed 1 +18 -7
Bastien Montagne commented 2023-10-11 12:16:42 +02:00
Owner
Copy Link

Building the depsgraph from non-main thread is unsafe currently, as it
may trigger some deferred processing on the Main data itself (here, the
update/resync of viewlayer collections).

So this commit moves the building of the export depsgraph back into the
main thread for the USD exporter, while keeping the depsgraph evaluation
code (which is typically the heavy part with complex scenes) into the
threaded job worker code.

Issue was randomly (but fairly commonly) reproducible when trying to export
most of the Pets project production files (on linux debug build with ASAN).

Root issue has been reported and is being discussed in #112534, ideally a
better solution can be designed at depsgraph level in the future.

NOTE: Should likely also be back-ported to 3.6 and 3.3 LTS.

Building the depsgraph from non-main thread is unsafe currently, as it may trigger some deferred processing on the Main data itself (here, the update/resync of viewlayer collections). So this commit moves the building of the export depsgraph back into the main thread for the USD exporter, while keeping the depsgraph evaluation code (which is typically the heavy part with complex scenes) into the threaded job worker code. Issue was randomly (but fairly commonly) reproducible when trying to export most of the Pets project production files (on linux debug build with ASAN). Root issue has been reported and is being discussed in #112534, ideally a better solution can be designed at depsgraph level in the future. NOTE: Should likely also be back-ported to 3.6 and 3.3 LTS.
Bastien Montagne added the
Interest
USD
Module
Pipeline, Assets & IO
 labels 2023-10-11 12:16:42 +02:00
Bastien Montagne added 1 commit 2023-10-11 12:16:54 +02:00
737ac07dc9 Fix USD Export crash with complex scenes, due to race condition during depsgraph building. 
Building the depsgraph from non-main thread is unsafe currently, as it
may trigger some deferred processing on the Main data itself (here, the
update/resync of viewlayer collections).

So this commit moves the building of the export depsgraph back into the
main thread for the USD exporter, while keeping the depsgraph evaluation
code (which is typically the heavy part with complex scenes) into the
threaded job worker code.

Issue was randomly (but fairly commonly) reproducible when trying to export
most of the Pets project production files (on linux debug build with ASAN).

Root issue has been reported and is being discussed in #112534, ideally a
better solution can be designed at depsgraph level in the future.

NOTE: Should likely also be back-ported to 3.6 and 3.3 LTS.
Bastien Montagne requested review from Sybren A. Stüvel 2023-10-11 12:17:12 +02:00
Bastien Montagne requested review from Sergey Sharybin 2023-10-11 12:17:12 +02:00
Bastien Montagne requested review from Michael Kowalski 2023-10-11 12:17:12 +02:00
Bastien Montagne added this to the USD project 2023-10-11 12:17:21 +02:00
Sergey Sharybin commented 2023-10-11 12:31:28 +02:00
Owner
Copy Link

The first evaluation would need to do the Copy-on-Write step, which involves accessing to the original objects. As long as the interface is locked before the depsgraph is constructed it is fine (so that there is no risk of artist deleting object after the graph is build and worker thread is actually span).

What is unclear to me, is why access to the original objects is fine from the thread for evaluation, but is not for the building. Is there are still some operations which access the view layer are run from the main thread while the export job is running?

The first evaluation would need to do the Copy-on-Write step, which involves accessing to the original objects. As long as the interface is locked before the depsgraph is constructed it is fine (so that there is no risk of artist deleting object after the graph is build and worker thread is actually span). What is unclear to me, is why access to the original objects is fine from the thread for evaluation, but is not for the building. Is there are still some operations which access the view layer are run from the main thread while the export job is running?
Bastien Montagne commented 2023-10-11 13:03:37 +02:00
Author
Owner
Copy Link

Issue is with scenes' ViewLayers, and their LayerCollections hierarchies. These are now only tagged as dirty when collection (and their objects) hierarchies are modified, and lazily updated on-demand (calls to BKE_main_view_layers_synced_ensure and the like). This tagging and lazy update is not threadsafe at all, so it should only ever be called from main thread.

If that happens as part of depsgraph building (which it can), from some random thread, then main thread (or another thread) may also be calling the same update code (from it's own main loop event depsgraph update e.g), and we have concurrency issues.

Issue is with scenes' ViewLayers, and their LayerCollections hierarchies. These are now only tagged as dirty when collection (and their objects) hierarchies are modified, and lazily updated on-demand (calls to `BKE_main_view_layers_synced_ensure` and the like). This tagging and lazy update is not threadsafe at all, so it should only ever be called from main thread. If that happens as part of depsgraph building (which it can), from some random thread, then main thread (or another thread) may also be calling the same update code (from it's own main loop event depsgraph update e.g), and we have concurrency issues.
Sergey Sharybin commented 2023-10-11 16:19:45 +02:00
Owner
Copy Link

Sure, but the question is more about: how is it possible that main thread with locked UI does something with the dirty state of collections?

Sure, but the question is more about: how is it possible that main thread with locked UI does something with the dirty state of collections?
Sergey Sharybin approved these changes 2023-10-11 16:26:12 +02:00
Sergey Sharybin left a comment
Owner
Copy Link

Talked IRL.

So it is just the UI which is locked, and the notifiers etc are still handled, so it is possible that layer collection's dirty state is handled from multiple threads.

Since the UI is locked it is not possible for artists to remove objects, so the rest of the evaluation is fine to happen in the thread.

Talked IRL. So it is just the UI which is locked, and the notifiers etc are still handled, so it is possible that layer collection's dirty state is handled from multiple threads. Since the UI is locked it is not possible for artists to remove objects, so the rest of the evaluation is fine to happen in the thread.
Michael Kowalski commented 2023-10-11 17:31:34 +02:00
Member
Copy Link

I will do a quick local build and test for due diligence, but the change looks good to me! Thank you for the fix!

I will do a quick local build and test for due diligence, but the change looks good to me! Thank you for the fix!
Michael Kowalski approved these changes 2023-10-11 18:10:59 +02:00
Bastien Montagne merged commit 5cf3654d9d into blender-v4.0-release 2023-10-12 19:10:46 +02:00
Bastien Montagne deleted branch tmp-usd-fix-deg-export-crash 2023-10-12 19:10:48 +02:00
Bastien Montagne removed this from the USD project 2023-11-02 16:13:33 +01:00
Sign in to join this conversation.
Reviewers
No reviewers
Sybren A. Stüvel
Sergey Sharybin
Michael Kowalski
Labels
Clear labels   
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
This issue affects/is about backward or forward 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
Issues relating to Wayland windowing support.

  
Interest
Workbench

  
Interest: X11

Issues relating to Xorg/X11 support.

  
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
Issues relating to security: https://wiki.blender.org/wiki/Process/Vulnerability_Reports

  
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 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
Milestone
Clear milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
No Assignees
3 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#113537
No description provided.
Delete Branch "mont29/blender:tmp-usd-fix-deg-export-crash"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?