depsgraph.c bug / dag_add_material_nodetree_driver_relations #32017

New Issue

Chris Want · 2012-07-04T21:23:44+02:00

Chris Want commented

2012-07-04 21:23:44 +02:00

Duplicates: #32069

%%%Hi Jeroen & Joshua,

I'm getting crashes when loading a lot of my files while using revision
48609 because there is some bad recursion in depsgraph.c. In particular,
the code seems to bounce endlessly between two function
calls:

0 0x0000000001c57963 in dag_add_material_driver_relations (dag=0x4eba428,

  node=0x4e688f8, ma=0x6db9808)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389

1 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

  dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:367

2 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428,

  node=0x4e688f8, ma=0x6db9808)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389

3 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

  dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:367

.
.
.

558 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428,

  node=0x4e688f8, ma=0x6db9808)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389

559 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

  dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:367

560 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428,

  node=0x4e688f8, ma=0x6db9808)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389

561 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

  dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58)
  at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/d

.
.
.

(Same memory addresses keep getting passed each time.)

Cheers,
Chris
%%%

**Duplicates**: #32069 %%%Hi Jeroen & Joshua, I'm getting crashes when loading a lot of my files while using revision 48609 because there is some bad recursion in depsgraph.c. In particular, the code seems to bounce endlessly between two function calls: # 0 0x0000000001c57963 in dag_add_material_driver_relations (dag=0x4eba428, ``` node=0x4e688f8, ma=0x6db9808) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389 ``` # 1 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations ( ``` dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:367 ``` # 2 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428, ``` node=0x4e688f8, ma=0x6db9808) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389 ``` # 3 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations ( ``` dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:367 ``` . . . # 558 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428, ``` node=0x4e688f8, ma=0x6db9808) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389 ``` # 559 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations ( ``` dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:367 ``` # 560 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428, ``` node=0x4e688f8, ma=0x6db9808) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/depsgraph.c:389 ``` # 561 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations ( ``` dag=0x4eba428, node=0x4e688f8, ntree=0x6db9d58) at /home/cwant/blender/bf-blender/blender/source/blender/blenkernel/intern/d ``` . . . (Same memory addresses keep getting passed each time.) Cheers, Chris %%%

Chris Want commented

2012-07-04 21:23:44 +02:00

Changed status to: 'Open'

Chris Want commented

2012-07-04 21:34:48 +02:00

%%%Heres a blend. Seems the problem is when a Material node is the same material as the material itself (in the case of the blend, Material.001 has a node with Material.001).%%%

Jeroen Bakker commented

2012-07-05 08:04:15 +02:00

%%%Hi Joshua!

Doesn't seem to be a compositor issue, could you check into it.
Last monday I only added two null pointer checks as the id of a groupnode is not always fill with the correct tree.

Greets,
Jeroen%%%

%%%Hi Joshua! Doesn't seem to be a compositor issue, could you check into it. Last monday I only added two null pointer checks as the id of a groupnode is not always fill with the correct tree. Greets, Jeroen%%%

Chris Want commented

2012-07-05 19:08:43 +02:00

%%%Hi Joshua, Jeroen,

Attached is a patch that solves the crashes. It does so by keeping track of the material at the root of the node tree, and doesn't recurse further if it encounters it again within the nodetree.

You may need to review to see that this functionally does what you want it to.

Chris
%%%

%%%Hi Joshua, Jeroen, Attached is a patch that solves the crashes. It does so by keeping track of the material at the root of the node tree, and doesn't recurse further if it encounters it again within the nodetree. You may need to review to see that this functionally does what you want it to. Chris %%%

Dalai Felinto commented

2012-07-06 18:32:24 +02:00

%%%I have the same problem in OSX. posting only to keep track of the entry so when solved I can see if it fixes my problem as well.%%%

Jason Wilkins commented

2012-07-06 18:35:02 +02:00

%%%There is another place in object.c which does the same thing in certain situations: material_drivers_update
%%%

%%%There is another place in object.c which does the same thing in certain situations: material_drivers_update %%%

Chris Want commented

2012-07-06 18:57:20 +02:00

%%%Hi Jason,

I've committed my patch for this. Jason, please check that it addresses the problem that you are experiencing.
Joshua, please review commit 48687.

Cheers,
Chris
%%%

%%%Hi Jason, I've committed my patch for this. Jason, please check that it addresses the problem that you are experiencing. Joshua, please review commit 48687. Cheers, Chris %%%

Dalai Felinto commented

2012-07-06 19:06:57 +02:00

%%%the patch fixed the crash here, but also loading my "buggy" file in trunk seems to take a bit more time (~20%) than in swiss. I don't think it's the patch, but probably the original functionality. is it only me?%%%

Chris Want commented

2012-07-06 19:25:17 +02:00

%%%Hi Dalai,

I don't notice this issue here, but mind you my files only take about 2 seconds to load, so a 20% difference isn't very noticeable.

Chris

%%%

%%%Hi Dalai, I don't notice this issue here, but mind you my files only take about 2 seconds to load, so a 20% difference isn't very noticeable. Chris %%%

Chris Want commented

2012-07-06 19:40:18 +02:00

%%%I should add that my test was comparing trunk with revision 48514 (before the crash).
I did not test the speed difference between trunk and swiss cheese.

Chris%%%

%%%I should add that my test was comparing trunk with revision 48514 (before the crash). I did not test the speed difference between trunk and swiss cheese. Chris%%%

bassam kurdali commented

2012-07-12 00:11:38 +02:00

%%%Just piling on to confirm the bug - node material inside it's own nodetree crashing files.

this was likely created by accident a while ago.%%%

%%%Just piling on to confirm the bug - node material inside it's own nodetree crashing files. - this was likely created by accident a while ago.%%%

Sergey Sharybin commented

2012-07-12 09:13:48 +02:00

%%%There're still issues with this functionality -- infinite recursion still could happen here.

Originally reported to http://projects.blender.org/tracker/index.php?func=detail&aid=32069&group_id=9&atid=498 but think it's ok to redirect all the feedback from that report here.%%%

%%%There're still issues with this functionality -- infinite recursion still could happen here. Originally reported to http://projects.blender.org/tracker/index.php?func=detail&aid=32069&group_id=9&atid=498 but think it's ok to redirect all the feedback from that report here.%%%

Joshua Leung commented

2012-07-22 18:17:28 +02:00

%%%Just got around to looking at this (4am local time :-/ )...

Hopefully my latest commit finally solves these issues. I've gone with the LIB_DOIT flagging I've been contemplating for a while.%%%

%%%Just got around to looking at this (4am local time :-/ )... Hopefully my latest commit finally solves these issues. I've gone with the LIB_DOIT flagging I've been contemplating for a while.%%%

Joshua Leung commented

2012-07-22 18:17:28 +02:00

Changed status from 'Open' to: 'Resolved'

Joshua Leung closed this issue

2012-07-22 18:17:28 +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

depsgraph.c bug / dag_add_material_nodetree_driver_relations #32017

0 0x0000000001c57963 in dag_add_material_driver_relations (dag=0x4eba428,

1 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

2 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428,

3 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

558 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428,

559 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (

560 0x0000000001c57968 in dag_add_material_driver_relations (dag=0x4eba428,

561 0x0000000001c578a6 in dag_add_material_nodetree_driver_relations (