TestRunnerSRL
Extremely rare (but consistent) crash when entering Deku B1 (corrupt seed generation?)
The first report was on April 22nd 2022. She was doing a League race (iirc the League settings are rather basic so not a mixed pools bug or something), jumped from the top of Deku onto the web and crashed when she fell into B1. Yesterday I asked her if she remembered anything about the crash and she said that rerolling the seed fixed the issue. This was on PJ64
Initial report (vod is gone): https://discord.com/channels/274180765816848384/476723801032491008/967131300928565378
Yesterday we got a second report of this crash, and that user was able to consistently reproduce the crash when reloading the save file, and even when reinstalling the wad on Wii VC. What he did differently was he used qpa to break the web instead of jumping from the top. He was also adult, whereas the other user was probably child because I don't think League settings had dungeon ER back then. I'm not sure if this is relevant though. He regenerated the same seed later with the same cosmetics and the issue didn't occur again (nor did another issue where pots in Dampe's grave were invisible).
Initial report (with a clip and logs): https://discord.com/channels/274180765816848384/438698093354156032/1019724210739875912
The leading theory seems to be that something went wrong when generating or patching the seed. I find it odd that that could lead to the exact same crash in two different instances, but I understand very little about this. We do know that it's platform independent.
I'm not on the Discord so I can't see the messages, but this seems extremely similar to an issue I just encountered with a seed I rolled today. I actually made it decently in before I encountered the issue, but I consistently crash when trying to enter a number of areas, and it doesn't seem to matter if I'm child or adult, or on a fresh save:
- Deku Tree Compass Room
- Deku Tree Basement
- Dodongo's Cavern
- Spirit Temple
The first two are the easiest to rush and verify from a new save. Here's how to reproduce:
Version: 8.1.0 Release (Offline)
Settings String: BSA4BGBSWJAA3YCBAEAKLEASJCCACAG2EAAAASAAACAAAAJKSQU7ELWAAJAPAAA2AAA6KTMUAAJHAATBBJAAWAR
Seed: 5DSI9OBLMR
Edit: Unfortunately, after testing, it seems that like above re-generating the seed fixes the issue so beyond providing specifics of a case that failed once the settings/seed aren't as helpful as I'd hoped
Edit2: I've retained the original, broken output in the hopes it might be useful. I can't share it, for obvious reasons, but if someone more knowledgeable is able to point me in the direction of tools or a map that could be used for me to analyze the file I'd be happy to do so to hopefully figure out how, specifically, it broke which may be able to be traced back to code
Adding this because we've recently found a correlation between a lot of reports like this. The first instance I can find so far is Cubsrule21 on January 29, 2020 version 5.1.63.
After this report from Retropolis, we now believe this to be caused by a race condition in the rom compressor. #2389 is an attempt at fixing this.
