Crash When firmware.bin.bak Not Present; Writes to firmware.bin Cause White Screen about melonds HOT 13 CLOSED

This commit should fix it on Linux and most likely Windows too 1a00ffd

from melonds.

That was driving me crazy cause it started to work for whatever reason :| But I manage to reproduce it and I think I finally found the cause, and it's pretty weird oO

Basically if you don't have firmware.bin.bak in your system folder it crashes (yes, the .bak!), simply make a copy of firmware.bin and rename it into firmware.bin.bak and... it just works.

It doesnt crash with 6a12f61 because that version automatically creates the .bak file, but not the newest commits it seems.

edit: video of the bug and workaround: https://youtu.be/EhAyopwDNs8
crash log: crash-190826-032202.log

from melonds.

Thanks for looking into it! I was always deleting that file after running the older version of the core, so that's why I could never get the latest version to work. Now that there's a lead on the problem, hopefully it won't be hard to fix.

from melonds.

I can get the latest build to work with the .bak file in system. But, if I change something in the firmware menu (like language), back out to let the system reboot, then close and restart RetroArch, the next time I try to load a game it gets stuck on a white screen. It seems doing that modifies the firmware.bin file and I have to restore it from the .bak to get it working again.

from melonds.

Able to build it with debug symbols and give me the crash report so I find out where it crashes? Not able to do it myself right now.

from melonds.

When I tried building with DEBUG=1, the crash log didn't seem to contain anything more than it did with a normal build.

from melonds.

Not sure if I did this right: MelonDS.zip
180MB log file inside, I only ran it for a few seconds :p

I started to BIOS and changed language just as Awakened0 said above. Nothing else.

edit: it's just a regular log however, RA doesnt crash for me, its just stuck on a white screen so no crash log are created :/

from melonds.

I can confirm that this issue effects Linux aswell.

The firmware.bin.bak fix workaround fixes it.

from melonds.

<deleted. This has nothing to do with it>

from melonds.

It fails at line 147 SPI.cpp.

Variable 'f' is null when it tries to write so it crashes.

It appears to crash because in line 227 of libretro.cpp the file handle request ends with 'true'. This is the "mustexist" flag. If the file does not exist it returns 0 and does not open a new file.

from melonds.

Confirmed fixed on linux.

Unrelated. Is the OpenGL support still in development or should it work ok? I get a segfault when it is active. I will create a new bug if it should work.

from melonds.

Yup, the .bak issue is also fixed in W10!

from melonds.

closing this since people have reported it being fixed on both platforms.

also @7oxicshadow the OpenGL renderer should work fine for Linux, feel free to create an issue with information about your problem

from melonds.