search

mgba-emu / mgba

mGBA Game Boy Advance Emulator
https://mgba.io/
Mozilla Public License 2.0
5.75k stars 795 forks source link

Pinball Deluxe/Fantasies completly broken #271

Open sczther opened 8 years ago

sczther commented 8 years ago

This game is well known for not working in emulators that are more accurate, it runs completely fine on old and hacky emulators such as no$gmb, but BGB, Gambatte and also mGBA have a problem with it. It runs on hardware just fine of course.

After selecting any of the tables, sound goes wonky, then garbage starts appearing and afterwards the screen switches colors around (something like b/w only, or white and dark grey, with a few things colored black) or into a solid wall of stripes and stops. The sound goes on for a while, but it's only screeching. After a while the screen is filled with tiled garbage. (Each of the tables behaves a bit differently, I tried to summarize)

endrift commented 8 years ago

IDK, when I tried it it just seems to freeze up randomly (but then unfreeze a second or two later).

E] Nevermind, tried it again and it's completely broken.

endrift commented 8 years ago

Made a thread on byuu's forums to discuss potential solutions to this issue: http://board.byuu.net/viewtopic.php?f=8&t=1307

alyosha-tas commented 7 years ago

Hi,

I got interested in this after reading your (very entertaining) glitch article. I tried to trace log out what was happening in gambette in BizHawk. Aside from what is already known about the delayed OAM IRQ blocking VBlank, I also noticed Gambette was doing some STAT IRQ blocking at an LY=LYC interrupt, blocking some OAM interrupts. It seems like it was blocking it for 2 scanlines though:

0048: C3 2C 13 JP #132Ch A:f8 LY:7e 132C: F5 PUSH AF A:f8 LY:7e 132D: F0 41 LDH A,(#FF41h) A:f8 LY:7e 132F: CB 57 BIT 2,A A:e6 LY:7e <--LY=LYC & OAM enabled . . longer codepath at this scanline takes it about 3/4 into LY=7f . another OAM IRQ assertion should happen somewhere in here . 136C: D9 RETI A:f8 LY:7f 44DB: F1 POP AF A:f8 LY:7f 44DC: 21 1A C0 LD HL,#C01Ah A:05 LY:7f 44DF: 19 ADD HL,DE A:05 LY:7f . . no interrupt until next scanline (~100 cycles) . not sure if IRQs get de-asserted ever? . 0048: C3 2C 13 JP #132Ch A:82 LY:7f 132C: F5 PUSH AF A:82 LY:7f 132D: F0 41 LDH A,(#FF41h) A:82 LY:7f 132F: CB 57 BIT 2,A A:e2 LY:80

(Interrupt looks like it triggers before LY actually equals 80 but maybe that is correct idk the internals.)

I'm not sure which interrupt has priority in this case LYC or OAM, but it seems strange that the IRQ assertion should be blocked for 2 scanlines here. Maybe this is already known but I didn't see it mentioned anywhere so just thought I'd point it out. Hope it helps somehow.

LIJI32 commented 7 years ago

Made a thread about this game on Reddit now that it works on SameBoy: https://www.reddit.com/r/EmuDev/comments/7206vh/sameboy_now_correctly_emulates_pinball_deluxe/