Crash when entering or leaving cave

This is strange, but may be related to another bug report I saw. To be sure, you are talking about the regular caves (with the sinkhole), correct? It seems the steam thread is about the Hamlet caves, which are entirely different - they would be two different bugs.

I am pondering how to get more info on this crash, sadly your log seems to be missing the part before the actual crash.

1 hour ago, bizziboi said:

To be sure, you are talking about the regular caves (with the sinkhole), correct? It seems the steam thread is about the Hamlet caves, which are entirely different - they would be two different bugs.

Yes. It's a sinkhole entrance near where I started.

Two different crashing bugs for entering caves sounds suspicious, almost like it's a bug in shared code, like the code for jumping to a different map or init/activate a new map. However if that is the case, then why is the sinkhole cave working during the day and crashing at night?

1 hour ago, bizziboi said:

I am pondering how to get more info on this crash, sadly your log seems to be missing the part before the actual crash.

I was expecting a method callstack printout like ONI does when something goes wrong (so far only caused by my own mods during pre-release testing), but DS is consistently not providing anything like that when encountering this bug. I kind of suspected the log to be useless, but I was hoping it was better than nothing because it's all I can think of.

I agree with the shared code but the two issues trigger a completely different codepath to do what they do so they don't share a critical codepath that I can think of. For me a bigger question is why for some people (a very small group based on reports) it happens so often. 

As for there being no callstack as opposed to ONI - ONI is written in C# which by default acts very dfferent than C++ in critical situations. However, in the case of a hard crash steam should create a dump file in the game's executable directory (generally "C:\Program Files (x86)\Steam\steamapps\common\dont_starve\bin"\). See if there's any file named something like dontstarve_steam_302970_crash_2019_1_11T2_27_35C0.mdmp (timestamp would vary of course) with the date of the crash when it happens, it may help me track it down.

Thanks

I decided to get a fresh crash report because I wanted to know precisely what I did when the crash file was made. However for some mysterious reason the game didn't crash and trying repeatably, some nasty beast showed up and killed me.

However that might have been a lucky accident because starting a new game I find a new cave on day one and that one crashed. This means we know it isn't linked to a specific game and that it can't be fixed by starting a new game. Another reason why it was lucky is that I had a really lucky day 1 in terms of finding resources, but that's unrelated to the bug.

17 hours ago, bizziboi said:

For me a bigger question is why for some people (a very small group based on reports) it happens so often. 

I don't think I do anything special ingame. I hold a touch (it's at night), stand next to a sinkhole and press space. Using the steam version of DS in windows 7 doesn't sound abnormal either. What's left of what I can think of is the hardware.

There is a benchmark thread in the ONI forum where people download a savegame around cycle 2k and then report what FPS they get on which hardware. My computer stands out as I'm the only one using an Intel i7 4790k. I get 45 and 55 FPS on average (depending on which part of the base I look at). Most people report in the low 20s. Since this is a bigger difference than the single core cinebench score can explain, the current theory is that it's a mix of single core performance and memory I/O, particularly memory latency. I have low latency memory. While it isn't the fastest computer in the thread, it is certainly performing significantly differently than the majority.

The question is how would that crash DS? It's not overclocked and it doesn't crash other games. The only thing I can think of would be some critical race between two threads. I did encounter something like that back when dualcore was introduced. One thread would start another thread, but due to mutex locks not set correctly when the new thread started, being able to run the new and old threads in parallel would cause a crash. When running on a single core CPU, it didn't matter because the two threads wouldn't run at precisely the same time anyway and it would survive a bug in the locks during thread start.

We are past the single core CPU, but it really is the only thing I can think of where I trigger a bug the majority avoids. Some sort of critical race, which makes the game unstable if the CPU and/or memory is too fast.

dontstarve_steam_343524_crash_2019_6_16T17_19_37C0.mdmp

Edited June 16, 2019 by Nightinggale