Crash after any alembic import undo in an empty scene #77754

New Issue

Yegor · 2020-06-11T12:31:08+02:00

Yegor commented

2020-06-11 12:31:08 +02:00

System Information
Operating system: Linux-3.10.0-1062.9.1.el7.x86_64-x86_64-with-centos-7.7.1908-Core 64 Bits
Graphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.44

Blender Version
Broken: version: 2.83.0, branch: master, commit date: 2020-06-03 14:38, hash: 211b6c29f7

Short description of error
The bug appears only on an empty scene after importing an alembic and undo.Crash after any alembic import, then immediately undo without any other steps

Exact steps for others to reproduce the error
#77754.abc

Open Blender and delete all the objects if there are any
import .abc
no other steps
undo --> crash

**System Information** Operating system: Linux-3.10.0-1062.9.1.el7.x86_64-x86_64-with-centos-7.7.1908-Core 64 Bits Graphics card: GeForce GTX 1070/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 440.44 **Blender Version** Broken: version: 2.83.0, branch: master, commit date: 2020-06-03 14:38, hash: `211b6c29f7` **Short description of error** The bug appears only on an empty scene after importing an alembic and undo.Crash after any alembic import, then immediately undo without any other steps **Exact steps for others to reproduce the error** [#77754.abc](https://archive.blender.org/developer/F8608021/T77754.abc) - Open Blender and delete all the objects if there are any - import .abc - no other steps - undo --> crash

Yegor commented

2020-06-11 12:31:08 +02:00

Added subscriber: @YegorSmirnov

Philipp Oeser commented

2020-06-11 12:37:18 +02:00

Added subscriber: @lichtwerk

Philipp Oeser commented

2020-06-11 12:37:18 +02:00

Changed status from 'Needs Triage' to: 'Confirmed'

Philipp Oeser commented

2020-06-11 12:37:18 +02:00

Can confirm.

Philipp Oeser changed title from ~~Crash after any alembic import undo in an empty scene~~ to Crash after any alembic import undo

2020-06-11 12:43:11 +02:00

Sybren A. Stüvel commented

2020-06-15 12:58:32 +02:00

Added subscriber: @dr.sybren

Sybren A. Stüvel commented

2020-06-15 12:58:32 +02:00

Changed status from 'Confirmed' to: 'Resolved'

Sybren A. Stüvel closed this issue

2020-06-15 12:58:32 +02:00

Sybren A. Stüvel self-assigned this 2020-06-15 12:58:32 +02:00

Sybren A. Stüvel commented

2020-06-15 12:58:32 +02:00

@lichtwerk reproduced this bug, but I can't reproduce it with the latest official release (2.83) or with the latest daily build (2.90 2797a63a66). so it appears the bug has been fixed already.

Please try the latest daily build: https://builder.blender.org/download/

If the problem persists, please let us know so we can re-open the report. Don't forget to mention the specific version you tested again.

@lichtwerk reproduced this bug, but I can't reproduce it with the latest official release (2.83) or with the latest daily build (2.90 2797a63a66). so it appears the bug has been fixed already. Please try the latest daily build: https://builder.blender.org/download/ If the problem persists, please let us know so we can re-open the report. Don't forget to mention the specific version you tested again.

Philipp Oeser commented

2020-06-15 13:34:59 +02:00

Changed status from 'Resolved' to: 'Confirmed'

Philipp Oeser reopened this issue

2020-06-15 13:34:59 +02:00

Philipp Oeser commented

2020-06-15 13:34:59 +02:00

I can still repro with todays buildbot and 2.83 release.

God only knows why I was able to reproduce this in a non-empty scene (and why I have updated the report description and title), sorry for the confusion!
This only happens in an empty scene (will update the report again)

I can still repro with todays buildbot and 2.83 release. God only knows why I was able to reproduce this in a non-empty scene (and why I have updated the report description and title), sorry for the confusion! **This only happens in an empty scene** (will update the report again)

Philipp Oeser changed title from ~~Crash after any alembic import undo~~ to Crash after any alembic import undo in an empty scene

2020-06-15 13:36:32 +02:00

Sybren A. Stüvel removed their assignment 2020-06-16 19:03:13 +02:00

Sybren A. Stüvel commented

2020-06-18 17:23:25 +02:00

I still can't reproduce it, even on an empty scene.

@lichtwerk what platform are you testing on? I have Kubuntu Linux 19.10 (kernel 5.3.0), NVidia GTX 1080 (driver version 450.36.06).

Also, which version of Blender did still work properly?

I still can't reproduce it, even on an empty scene. @lichtwerk what platform are you testing on? I have Kubuntu Linux 19.10 (kernel 5.3.0), NVidia GTX 1080 (driver version 450.36.06). Also, which version of Blender did still work properly?

Yegor commented

2020-06-18 17:42:10 +02:00

Just tested 2.90 from the buildbot and it still crashes on Mint and CenOS

Bastien Montagne commented

2020-06-22 12:19:08 +02:00

Added subscriber: @mont29

Bastien Montagne self-assigned this 2020-06-22 12:19:08 +02:00

Bastien Montagne commented

2020-06-22 12:19:08 +02:00

This is related to Undo I think. Alembic and Collada import operators are not flagged for undo (!!!), will fix this now, but this is not enough to fix the crash.

Bastien Montagne removed their assignment 2020-06-22 13:05:56 +02:00

Sybren A. Stüvel was assigned by Bastien Montagne

2020-06-22 13:05:56 +02:00

Bastien Montagne commented

2020-06-22 13:05:56 +02:00

Well in fact issue is with alembic importer, which starts a background job for actual import, so undo step stored by the operator management system is done before data is actually changed, and importer had to do its own undo push when it is done.

This quick patch solves it, but not sure it's the best way to handle it (it add extra useless undo step e.g.), will let @dr.sybren handle it from there.

P1481: (An Untitled Masterwork)

diff --git a/source/blender/io/alembic/intern/alembic_capi.cc b/source/blender/io/alembic/intern/alembic_capi.cc
index 0b4b184359b..a406b63df06 100644
--- a/source/blender/io/alembic/intern/alembic_capi.cc
+++ b/source/blender/io/alembic/intern/alembic_capi.cc
@@ -51,6 +51,8 @@
 #include "DEG_depsgraph.h"
 #include "DEG_depsgraph_build.h"
 
+#include "ED_undo.h"
+
 /* SpaceType struct has a member called 'new' which obviously conflicts with C++
  * so temporarily redefining the new keyword to make it compile. */
 #define new extern_new
@@ -429,6 +431,7 @@ enum {
 };
 
 struct ImportJobData {
+  bContext *C;
   Main *bmain;
   Scene *scene;
   ViewLayer *view_layer;
@@ -447,6 +450,7 @@ struct ImportJobData {
   char error_code;
   bool was_cancelled;
   bool import_ok;
+  bool is_background_job;
 };
 
 static void import_startjob(void *user_data, short *stop, short *do_update, float *progress)
@@ -630,6 +634,12 @@ static void import_endjob(void *user_data)
 
     DEG_id_tag_update(&data->scene->id, ID_RECALC_BASE_FLAGS);
     DEG_relations_tag_update(data->bmain);
+
+    if (data->is_background_job) {
+      /* We already returned from the import operator, so we need to store our own extra undo step.
+       */
+      ED_undo_push(data->C, "Alembic Import Finished");
+    }
   }
 
   for (iter = data->readers.begin(); iter != data->readers.end(); ++iter) {
@@ -675,6 +685,7 @@ bool ABC_import(bContext *C,
 {
   /* Using new here since MEM_* functions do not call constructor to properly initialize data. */
   ImportJobData *job = new ImportJobData();
+  job->C = C;
   job->bmain = CTX_data_main(C);
   job->scene = CTX_data_scene(C);
   job->view_layer = CTX_data_view_layer(C);
@@ -691,6 +702,7 @@ bool ABC_import(bContext *C,
   job->error_code = ABC_NO_ERROR;
   job->was_cancelled = false;
   job->archive = NULL;
+  job->is_background_job = as_background_job;
 
   G.is_break = false;

Well in fact issue is with alembic importer, which starts a background job for actual import, so undo step stored by the operator management system is done before data is actually changed, and importer had to do its own undo push when it is done. This quick patch solves it, but not sure it's the best way to handle it (it add extra useless undo step e.g.), will let @dr.sybren handle it from there. [P1481: (An Untitled Masterwork)](https://archive.blender.org/developer/P1481.txt) ``` diff --git a/source/blender/io/alembic/intern/alembic_capi.cc b/source/blender/io/alembic/intern/alembic_capi.cc index 0b4b184359b..a406b63df06 100644 --- a/source/blender/io/alembic/intern/alembic_capi.cc +++ b/source/blender/io/alembic/intern/alembic_capi.cc @@ -51,6 +51,8 @@ #include "DEG_depsgraph.h" #include "DEG_depsgraph_build.h" +#include "ED_undo.h" + /* SpaceType struct has a member called 'new' which obviously conflicts with C++ * so temporarily redefining the new keyword to make it compile. */ #define new extern_new @@ -429,6 +431,7 @@ enum { }; struct ImportJobData { + bContext *C; Main *bmain; Scene *scene; ViewLayer *view_layer; @@ -447,6 +450,7 @@ struct ImportJobData { char error_code; bool was_cancelled; bool import_ok; + bool is_background_job; }; static void import_startjob(void *user_data, short *stop, short *do_update, float *progress) @@ -630,6 +634,12 @@ static void import_endjob(void *user_data) DEG_id_tag_update(&data->scene->id, ID_RECALC_BASE_FLAGS); DEG_relations_tag_update(data->bmain); + + if (data->is_background_job) { + /* We already returned from the import operator, so we need to store our own extra undo step. + */ + ED_undo_push(data->C, "Alembic Import Finished"); + } } for (iter = data->readers.begin(); iter != data->readers.end(); ++iter) { @@ -675,6 +685,7 @@ bool ABC_import(bContext *C, { /* Using new here since MEM_* functions do not call constructor to properly initialize data. */ ImportJobData *job = new ImportJobData(); + job->C = C; job->bmain = CTX_data_main(C); job->scene = CTX_data_scene(C); job->view_layer = CTX_data_view_layer(C); @@ -691,6 +702,7 @@ bool ABC_import(bContext *C, job->error_code = ABC_NO_ERROR; job->was_cancelled = false; job->archive = NULL; + job->is_background_job = as_background_job; G.is_break = false; ```

blender-admin commented

2020-06-22 17:53:14 +02:00

This issue was referenced by a2c1f41d95

This issue was referenced by a2c1f41d95957a84e7a856a7b1dc889a0d98be5b

blender-admin commented

2020-06-22 17:53:14 +02:00

This issue was referenced by 25d2222c30

This issue was referenced by 25d2222c302acd7278b49f30a74adba6b9457012

Sybren A. Stüvel commented

2020-06-22 18:12:34 +02:00

Changed status from 'Confirmed' to: 'Resolved'

Sybren A. Stüvel closed this issue