Diakatana Works Fine Until Sub-Map Transition Near End of E1M1
Issue transferred from https://github.com/ValveSoftware/Proton/issues/5879.
@PorcelainMouse posted on 2022-06-01T08:36:29:
Game crashes on a specific sub-map transition. At least, that's were I think it is crashing. While working through the first map, I experienced five or so sub-map loadings/transitions, already, which worked fine, and those transitions still work. But the game crashes at a very specific point, when the player character walks over a specific point, which is very similar to the way the other map transitions were implemented; I'm assuming the crash occurs at such a point, although the loading screen doesn't appear before the crash.
While trying to provide requested logs, I ran a file integrity check, which turned up several corrupt files. The second file integrity check passed and I tried the game again. After this crash, file integrity check failed again and fixed itself. Seems like the crash is unsafe and corrupts files.
I am able to reproduce pretty reliably.
@Blisto91 commented on 2022-06-01T10:21:12:
Hi there.
Does it behave the same on proton 7.0-2 & experimental?
If yes does it change anything if you launch the game with
PROTON_USE_WINED3D=1 %command%
I tried 7.0-2 and got the same issue, but not experimental. I can do that. Would it be odd if that helped, though? I mean, there are a lot of protondb reports of it working on older versions, like 5.0. Anyway, I can try, just give me a little time.
I tried experimental, but that didn't work at all. Daikatana starts, but crashes immediately after loading a saved game or even a new game. After each crash, file integrity check detected multiple damaged files and fixed them or "reacquired" them or whatever.
I went back to v6.3 to test your other suggestion (PROTON_USE_WINED3D=1), but I couldn't make the problem recur with or without that set. Both ways allowed the game to start and the game appeared unaffected. Once a crash occurs, though, I'm guessing the damaged files cause the subsequent crashes that occur frequently, making it seem like there is a problem around the same place in the game. But, that is coincidence, I bet, and there's not point in troubleshooting after files get corrupted.
However, now that I know to check file integrity after the crashes, it isn't happening at the same place, like it was. I spent a while trying to reproduce, but I think that is why it isn't happening any more. I loading my last saved game I was trying to continue when I reported this problem, and ran around the same part of the map for a while, but nothing happened, even though I couldn't progress in the game because the crash was so frequent.
I found that Romero has released a native Linux binary, which is even updated from the Steam version. That was much more reliable, and I finished the game using that version. However, it also crashed at one point, causing damage to important files, and wouldn't even start up, afterward. I had to reinstall to fix it. So, there is clearly an bug in the program that causes files to be corrupted when the game crashes, but I have no idea why it crashes; it may not be one thing. These crashes are rare, and would remain so, if it were not for the other bug that corrupts files. I would think this symptom would be easily recognizable to any programmer who has written programs of this complexity, but that isn't me. (Maybe some non-standard, unsafe mmap-ing text or data? I don't remember enough C to be sure, but it's conceivable.)
proton 7.0-2x1 2022-06PROTON_USE_WINED3D=1`),x1 2022-06PROTON_USE_WINED3D=1x1 2022-06PROTON_USE_WINED3D=1 %command%x1 2022-06
Compatibility Report
System Information
LOG:
steam-242980.log
Symptoms
Game runs, but mouse movement feels very laggy