Re: [Crawl-ref-discuss] [DCSS 0011734]: crash on game start after updating to latest (98c7a59) trunk

[Kristoffer B. <kri...@gm...>]

Yes, I tried multiple (>~5) times, immediate bounce back to dgamelaunch
every time, until Napkin added the "press any key" prompt.

On Tue, Oct 23, 2018 at 3:28 PM DCSS Bug Tracker <ma...@cr...>
wrote:

>
> A NOTE has been added by advil.
> ============================================================================
>
> https://crawl.develz.org/mantis/view.php?id=11734
> ============================================================================
>
> Reported By:                TAS2012
> Assigned To:
> ============================================================================
>
> Project:                    DCSS
> Product Branch:             0.23 development branch
> Issue ID:                   11734
> ----------------------------------------------------------------------------
>
> Category:                   Bug Report
> Reproducibility:            always
> Severity:                   crash
> Status:                     new
> ============================================================================
>
> Date Submitted:             2018-10-23 11:55 CEST
> Last Modified:              2018-10-23 15:28 CEST
> ============================================================================
>
> Summary:                    crash on game start after updating to latest
> (98c7a59) trunk
> ----------------------------------------------------------------------------
>
> </pre>Starting Dungeon Crawl Stone Soup 98c7a59...
>
> Crash caused by signal https://crawl.develz.org/mantis/view.php?id=6:
> Aborted
>
> We crashed! This is likely due to a bug in Crawl.
> Please submit a bug report at https://crawl.develz.org/mantis/ and
> include:
> - The crash report:
> /crawl-svn/morgue/TAS2012/crash-TAS2012-20181023-093600.txt
> - Your save file: /crawl-svn-98c7a59/saves/TAS2012.cs
> - A description of what you were doing when this crash occurred.
>
> /bin/crawl-svn: line 154: 32608 Aborted                 (core dumped)
> ${BINARY_NAME} $*
> --- any key to continue ---</pre>
>
> I had a game that had rested since 2018-10-19.
> When I tried to resume it today, I was asked if I wanted to update to
> latest trunk, answered yes, and then promptly got thrown out back to
> dgamelaunch.
> When I tried again to resume the game I momentarily could see "Starting
> [...] 98c7a59" and was then immediately bounced back.
>
> I contacted Napkin and thanks to the "any key to continue" being added to
> the script I could see what I have pasted above. I got copies of the
> crash-* and crash-recursive-* files.
>
> http://crawl.develz.org/morgues/trunk/TAS2012/crash-TAS2012-20181023-093600.txt
>
> http://crawl.develz.org/morgues/trunk/TAS2012/crash-recursive-TAS2012-20181023-093600.txt
>
> Napkin then moved my save to Sep 17 20:26 crawl-svn-440b721 and I tried to
> resume the game, without updating to latest trunk, and it seems to work
> fine.
>
> I then (sorry, maybe should have done this before, hope I didn't destroy
> the needed data) got a backup of the save file.
> https://crawl.develz.org/saves/dumps/TAS2012-440b721-181023-0943.tar.bz2
>
> I'll upload related files to mantis also.
> ============================================================================
>
>
> ----------------------------------------------------------------------------
>
>  (0032715) advil (developer) - 2018-10-23 15:28
>  https://crawl.develz.org/mantis/view.php?id=11734#c32715
> ----------------------------------------------------------------------------
>
> I can't get the uploaded save file to crash when updating to current trunk,
> so it might have gotten fixed somehow, or there was an extra random factor
> to the crash, or it's compiler specific (I'm testing on mac/clang, which is
> different than CDO).
>
> Roughly, this was crashing when it tried to update some artifact's property
> list to allow a slot for harm (added in 0.23-a0-330-g6e662b0). The
> immediate crash looks like it failed to resize the property vector in
> `artefact_fixup_props`; I can't figure out why this would fail. The
> recursive crash is just because `debug_item_scan` is run during the crash
> handler and the original crash happened in the middle of
> `tag_read_level_items`, so the items are half-loaded and practically
> guaranteed to be corrupted. (In this case it's crashing exactly because the
> fixup didn't complete, but all the unlinked items are because of this as
> well I'm pretty sure. In the working save, the items listed as unlinked in
> the crash are on the ground.) So basically the recursive crash isn't the
> interesting part, just a complication of when the original crash happened.
>
> Do you know if this crash was deterministic on CDO? I.e. did you try to
> load it multiple times after the update with the same result?
>
> Issue History
> Date Modified    Username       Field                    Change
>
> ============================================================================
>
> 2018-10-23 11:55 TAS2012        New Issue
>
> 2018-10-23 11:55 TAS2012        File Added:
> TAS2012-440b721-181023-0943.tar.bz2
>
> 2018-10-23 11:57 TAS2012        File Added:
> crash-TAS2012-20181023-093600.txt
>
> 2018-10-23 11:57 TAS2012        File Added:
> crash-recursive-TAS2012-20181023-093600.txt
> 2018-10-23 12:43 TAS2012        Issue Monitored: TAS2012
>
> 2018-10-23 15:28 advil          Note Added: 0032715
>
>
> ============================================================================
>
>