shadps4-emu / shadps4-game-compatibility

Shadps4 game compatibility
179 stars 17 forks source link

CUSA13795 - Crash™ Team Racing Nitro-Fueled #41

Open StevenMiller123 opened 3 months ago

StevenMiller123 commented 3 months ago

Game Name

Crash™ Team Racing Nitro-Fueled

Game code

CUSA13795

Game version

v1.00

Used emulator's version (only released versions are acceptable)

v0.1.0

Current status

Nothing

Error

The game crashes without any clear cause in the log.

Description

The game crashes instantly.

Screenshots

No response

Log File

Crash Team Racing Nitro-Fueled Log.txt

StevenMiller123 commented 2 months ago

With emulator v0.2.0, the game now crashes with the following error: [Debug] <Critical> memory_management.cpp:sceKernelBatchMap2:287: Unreachable code! called: Unimplemented Operation = 4

No change to game status. Log: Crash Team Racing Nitro-Fueled log.txt

niko1point0 commented 1 month ago

@georgemoralis on the "main" branch, sceKernelBatchMap2 is now fixed. On PR #556 there are improvements that allow all the PM4 to be parsed.

On a debug experimental build, with PR #556, combined with hacks to skip all shader failures, and ignoring exception throws (very unsafe, for experimentation only), Nitro Fueled can go in-game. Here is Oxide Station. image

StevenMiller123 commented 3 weeks ago

On v0.3.0, CUSA13795 regressed, and now crashes almost instantly on the following error:

[Debug] <Critical> memory.cpp:operator():568: Assertion Failed!
Mapping cannot fit inside free region: size = 0x20000000

Game status remains at nothing. shad_log.txt

niko1point0 commented 3 weeks ago

That's not a regression, the official public builds never worked with NF. This game will not work in ShadPS4 until Geometry Shaders are implemented and recompiling. Also to fix the memory error, I just increased the size of the emu ram to bypass that

StevenMiller123 commented 3 weeks ago

I'm aware the game was always at nothing, I kept testing for a reason. These reports are specifically for released builds, and in this case, it's pretty clear that the "nothing" produced in this newer build is an earlier crash than the "nothing" in v0.2.0.