[Crash] on a alien spotted mission (Segmentation fault)

[Atrosha]

Crash during the mission in real time mode

gdb listing
Thread 1 "main" received signal SIGSEGV, Segmentation fault. OpenApoc::BattleHazard::update (this=0x60000000d, state=..., ticks=1) at /home/atrosha/OpenApoc/game/state/battle/battlehazard.cpp:490 490 if (ticksUntilVisible > 0) (gdb) info stack #0 OpenApoc::BattleHazard::update (this=0x60000000d, state=..., ticks=1) at /home/atrosha/OpenApoc/game/state/battle/battlehazard.cpp:490 #1 0x00000000007378f1 in OpenApoc::Battle::update (this=<optimized out>, state=..., ticks=1) at /home/atrosha/OpenApoc/game/state/battle/battle.cpp:1675 #2 0x00000000008aa757 in OpenApoc::GameState::update (this=0x5989410, ticks=1) at /home/atrosha/OpenApoc/game/state/gamestate.cpp:959 #3 0x000000000052bf2d in OpenApoc::BattleView::update (this=<optimized out>) at /home/atrosha/OpenApoc/game/ui/tileview/battleview.cpp:1443 #4 0x00000000004569b8 in OpenApoc::Framework::run (this=0xfd14d0, initialStage=...) at /home/atrosha/OpenApoc/framework/framework.cpp:584 #5 0x000000000043a1d4 in main (argc=<optimized out>, argv=<optimized out>) at /home/atrosha/OpenApoc/game/main.cpp:26

[JonnyH]

Hmm, looks like the BattleHazard::update() is being called on a junk object -0x60000000d doesn't look sane for a heap pointer....

[FranciscoDA]

I think this is caused by an invalidated iterator: BattleHazard::update may end up calling BattleHazard::expand, which in turn can call die() on a nearby BattleHazard. If the destroyed hazard happens to be the next one in the iteration, then the iterator used in Battle::update becomes invalid (see https://en.cppreference.com/w/cpp/container/set/erase)

This is the only explanation I can think of

[Atrosha]

#529 causes game to crash more often on that mission. But nobody react on it. ehhh...

Also, I was finished that mission in realtime mode without any errors later.

So strange.

[FranciscoDA]

#536 might fix this but it will be hard to test since this bug occurs once in a blue moon. We should close this issue once we merge the pull request and reopen if the bug persists.

[Atrosha]

Nice and fast response. Can I mention you in my next video?