Major regression in jinpaish

[cuavas]

MAME version

0.283 (mame0283-109-g55c70b54e22)

System information

Windows 10, 64-bit, x86-64

INI configuration details

joystick                  1
joystick_deadzone         0.15
joystick_saturation       0.95
trackball_device          mouse
mouse_device              mouse
window                    1
maximize                  1

Emulated system/software

Jinpai Suoha - Show Hand (Chinese version 2004-09-22) (jinpaish)

Incorrect behaviour

The game boots to attract mode, then shows a demo game with an unrealistically high bet of 50,000. It takes minutes to total the winnings for four of a kind (四梅, total 63,806, which is incorrect due to integer overflow), then sits on the double up game for a while without actually choosing. After this, it freezes during the GMS logo animation.

If you reset it, it may lock up again at the initial GMS logo animation, or run the same flawed attract mode. You can get it to repeat the broken attract mode if you change the control panel type DIP switch when it locks up on the initial GMS logo animation, it may cause it to run the broken attract mode again.

It also won’t accept coins and can’t be played.

Expected behaviour

As of 5b7e4ad, the game actually ran attract mode, accepted coins and could be played. It’s regressed pretty badly.

Steps to reproduce

No response

Additional details

I suspect ba58772 caused the regression. I’m now questioning the value of it, given it still requires ROM patches, but makes it impossible to even play games to test other functionality.

[ajrhacker]

8fa0daf fixes the outright freezes, though the ridiculous bet/payout amounts and refusal to coin up are still present.

[ajrhacker]

jinpaish does use a few more unemulated commands, apparently arithmetic functions, which might be responsible for the other regressions.

The MACHINE_UNEMULATED_PROTECTION flag is set on almost all games in this driver for a good reason. There's just too many unknown ASIC-related bits being checked.