dmitrysmagin/pcsx4all

Pcsx4all v2.4 with mips-to-mips recompiler by Ulrich Hecht https://github.com/uli/psx4all-dingoo/

pcsx4all's Issues

Save state failed: out of disk space

It is impossible to save state in gcw0 device. Ou of disk space error is raised up even if there's still lot of disk space.

opendingux:/media/data/local/home/.pcsx4all/sstates # df -h
Filesystem Size Used Available Use% Mounted on
/dev/mmcblk0p1 383.2M 134.7M 248.6M 35% /boot
/dev/loop0 64.8M 64.8M 0 100% /
udev 64.0K 0 64.0K 0% /dev
tmpfs 128.0M 0 128.0M 0% /dev/shm
tmpfs 128.0M 260.0K 127.7M 0% /var
tmpfs 4.0K 0 4.0K 0% /media
tmpfs 4.0K 0 4.0K 0% /mnt
/dev/loop0 4.0K 0 4.0K 0% /usr/share/ftphome/media
tmpfs 4.0K 0 4.0K 0% /usr/share/ftphome/media
/dev/loop1 640.0K 640.0K 0 100% /lib/modules
/dev/mmcblk0p2 14.3G 1.9G 12.2G 13% /media/data
/dev/mmcblk0p2 14.3G 1.9G 12.2G 13% /usr/share/ftphome/media/data

HLE Bios : Magic numbers, autobias, ReturnFromException...

So while investigating for the hang issues in Digimon Rumble Arena and Final Fantasy 8, i found this mysterious commented line of code

https://github.com/dmitrysmagin/pcsx4all/blob/master/src/psxbios.cpp#L1693

Commenting out this exact line allows Digimon Rumble Arena to go a bit further before it triggers the bios's exit function.
I asked Senquack about it and he has no idea, looking at nocash documentation does not give me any clues either.

This commit
4a18d9e#diff-58e9dc1093d659513f538cf2d0f50fc6
added a lot of stuff but the diffing did not work properly unfortunately...
Here's a diff between the initial import of PSX4ALL v2 and the commit followed after that touches psxbios.cpp
https://pastebin.com/crvN8wGU

Yeah there's a lot of stuff, including things not seen in other forks.
I later noticed that you removed autobias later
a7adf8b
followed by the magic numbers :
487a575

So hmm what was the purpose of autobias ? You said that it wasn't working properly but there's a few games requiring cycle accuracy for memory card functions afaik.
And what about the magic numbers ? How did you get them ? Reverse engineering ?

Pull Request

Please pull my changes.
Removes a lot of redundancy in the EmiPSX source.

See: https://github.com/xsacha/EmiPSX

Original issue reported on code.google.com by Xsacha on 31 Mar 2013 at 8:24

Digimon World HLE bios issue

Digimon World had two issues with the HLE implementation
The first one occurs when it uses the strcat by attaching ;1 to a string. Disabling the function lead to the game to boot, which led me to nocash's psx documentation and it turns out the function is not correctly implemented. It also needs psxRegs.CP0.n.Status |= 0x401; for it to work properly and go past the lock up, whatever this does. (related to IRQ ?)

The second one is input related : it would not detect button presses.
After much frustration trying to debug it, it turns out to be yet another missing check.
Doc says that Bx015 (OutdatedPadInitAndStart) will only work if type is 0x20000000 or 0x20000001.
If it's not, then it will refuse to do anything. Implementing this behaviour also fixed input in digimon.

You can see my fixes to those here
gameblabla/pcsx4all@79411d3

Unfortunately, while the game is playable, it still has some issues.
The other one is that sound effects will not playback (but other sounds will work ok). The other, more annoying issue is related to memory cards. If you start a game in Digimon World, save it, exit the emulator and try to continue the game from the game itself, it will fail to detect the saves in there.
And when you attempt to start a new game over the one you were using, it will refuse to start the game. I believe that the game will not create a new game and overwrite it over an existing one.
But it still shows as if no saves exists. Card_info function is used for checking saves (apparently).

So yeah, while i did allow the game to bootup and be playable, there are still other issues. I would be happy if anyone could fix these. (i will also let pcsx-rearmed about it but notaz isn't too active so i doubt)