Segfault on Linux client since #9f129fe

[Matthew]

I get segmentation faults on start on both the updated client from last night (21-01-13 1909, #9f129fe) and today's nightly (21-01-13, #483db00).

But Freddy does not, even though we are both running recent Ubuntu (20.04.1 in my case).

Things ruled out

• It is not a corrupted download, as Freddy and I have matching SHA256 hashes
• It should not be trying to load a savefile corrupted by the unsuccessful integration of ceeac's I/O changes, as I have tried -load "", -load demo.sve (with that file copied to /simutrans/save), and -load net:bridgewater-brunel.me.uk.
• Renaming settings.xml does not help either.
• It should not be pakset incompatibility as I have run the Nightly Updater.

Backtrace

Freddy kindly supplied an optimized debug build, probably equivalent to last night's #9f129fe. The backtrace is:



The trace then continues going around in circles for several hundred lines.

Simu.log, the text file of the backtrace, and a core dump to use with that executable are here.

I think I am going to try downloading the pakset again, in case that has somehow got corrupted.

EDIT: It was a corrupted pakset download (specifically trains.pak which was updated yesterday). Please move this to 'solved'.