Bug 166788 - Crash on completion of solver
Summary: Crash on completion of solver
Status: UNCONFIRMED
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: Calc (show other bugs)
Version:
(earliest affected)
25.2.3.2 release
Hardware: All All
: medium normal
Assignee: Not Assigned
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: Solver
  Show dependency tree / graph
 
Reported: 2025-05-29 22:03 UTC by Paul Welton
Modified: 2025-12-27 18:16 UTC (History)
2 users (show)

See Also:
Crash report or crash signature: ["std::_Tree<std::_Tmap_traits<rtl::OUString,std::unique_ptr<ScRangeData,std::default_delete<ScRangeData> >,std::less<rtl::OUString>,std::allocator<std::pair<rtl::OUString const ,std::unique_ptr<ScRangeData,std::default_delete<ScRangeData> > > >,0> >::_"]

Attachments
Screenshot of spreadsheet with libraOffice "About" showing (327.88 KB, image/png)
2025-05-29 22:03 UTC, Paul Welton 		Details
Spreadsheet for example of crash in Solver (447.75 KB, application/octet-stream)
2025-05-29 23:07 UTC, Paul Welton 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Paul Welton 2025-05-29 22:03:56 UTC 
Created attachment 201036 [details]
Screenshot of spreadsheet with libraOffice "About" showing

The solver had run successfully earlier, but I decided to run it again with cell M4 changed from 58% to 70%.

Tools->Solver

Parameters filled in automatically:

Target cell $AG$9
By changing cells $AF$3:$AF$4
Options:
    Engine: Non-linear Swarm, with parameter 0 for Differential Evolution.

Tried it again, after recovery and the crash did not happen. This has been normal behavior for a long time - crashes on 1 in 20 invocations. Recovery always successful.

Crash report:

https://crashreport.libreoffice.org/stats/crash_details/438c454b-1729-4a72-b6b5-613321fda44d
Comment 1 Rafael Lima 2025-05-29 22:49:30 UTC 
Hi, can you please share the file with the model you're running?

I need to test it locally to see what causes this crash.
Comment 2 Paul Welton 2025-05-29 23:07:46 UTC 
Created attachment 201037 [details]
Spreadsheet for example of crash in Solver
Comment 3 Rafael Lima 2025-05-30 12:14:07 UTC 
I ran the model multiple times with 70% efficiency and using the Differential Evolution engine, but could not reproduce the crash.

So you said that it happens occasionally, right?

It may be an edge case, since it uses random numbers.

Ideally this crash should have been caught after bug 162760 was fixed.
Comment 4 Jussi Suominen 2025-05-30 14:03:46 UTC 
I tried to reproduce the bug but was unable to reproduce it. However, solving took quite a long time (about a minute) and the application became unresponsive during the solving process (and I thought it crashed). Eventually the solving finished and the application became responsive again. I tried this multiple times and the behaviour was always the same.

Version: 25.2.3.2 (X86_64) / LibreOffice Community
Build ID: bbb074479178df812d175f709636b368952c2ce3
CPU threads: 16; OS: Windows 11 X86_64 (10.0 build 26100); UI render: Skia/Vulkan; VCL: win
Locale: fi-FI (fi_FI); UI: en-GB
Calc: CL threaded
Comment 5 Rafael Lima 2025-05-30 14:12:00 UTC 
(In reply to JussiS from comment #4)
> I tried to reproduce the bug but was unable to reproduce it. However,
> solving took quite a long time (about a minute) and the application became
> unresponsive during the solving process (and I thought it crashed).

The absence of a progress bar showing that the solver is actually running is a separate issue (maybe related to bug 162731).

As for the crash itself, let's wait and see if anyone can reproduce it.
Comment 6 Paul Welton 2025-05-30 23:48:51 UTC 
This crash is typical of the normal behavior I have had over the last six months. The solver takes 1 minute to run and crashes about 1 in 20 times. In all cases rerunning the solver with the same data has been successful, so it doesn't appear that the crashes are related to the spreadsheet content. And the bug is not reproducible, it occurs randomly.

As recovery is always successful, the bug is infrequent and saving the file is much quicker than running the solver so can always be done without penalty, the bug is not a problem in practice. I submitted the bug report because it generated a crash dump and asked me to do so.