Motivation / Problem

ThreadSantizer was complaining on exiting the game. It is a bit silly, as it is a boolean value .. but making it a std::atomic is also a small effort.

Description

_exit_game is read by the draw-thread to know when to exit, but
most of the time written by the game-thread.

Most likely we could change all the loads and stores to memory_order_relaxed, as I doubt it really matters honestly. We are pretty resilient against _exit_game being flipped at the weirdest places.

The reason I didn't do it, is because there are roughly 30 places where we load/store _exit_game. That maybe highlights a completely different issue in our codebase :P

Limitations

• I am never all to sure about the performance impact of making something into a std::atomic. I would think it doesn't matter (at all) in this case, but I would love to be told wrong here :)

Checklist for review

Some things are not automated, and forgotten often. This list is a reminder for the reviewers.

• The bug fix is important enough to be backported? (label: 'backport requested')
• This PR affects the save game format? (label 'savegame upgrade')
• This PR affects the GS/AI API? (label 'needs review: Script API')
  • ai_changelog.hpp, gs_changelog.hpp need updating.
  • The compatibility wrappers (compat_*.nut) need updating.
• This PR affects the NewGRF API? (label 'needs review: NewGRF')
  • newgrf_debug_data.h may need updating.
  • PR must be added to API tracker