02881: thunderj: Thunderjaws sometimes resets during gameplay.

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
02881	Crash/Freeze	Critical (emulation)	Random	Jan 26, 2009, 07:58	Aug 9, 2018, 06:27

Tester	Machone	View Status	Public	Platform	MAME (Self-compiled)
Assigned To		Resolution	Open	OS
Status [?]	Confirmed			Driver	atari/thunderj.cpp
Version	0.129u2	Fixed in Version		Build	C2D
		Fixed in Git Commit		Github Pull Request #

Summary	02881: thunderj: Thunderjaws sometimes resets during gameplay.
Description	Sometimes during gameplay, Thunderjaws resets itself to its title screen (very much the same as if F3 had been pressed). I tried testing using the official 0.129 plain binary from mamedev.org as well as my self-compiled 0.129u2 and the problem appears in both versions.. Unfortunately I couldn't get a gameplay recording as the inp file I tried to record didn't seem to want to play back correctly. I must add that this isn't a new problem; it occurred pre-Aaron's 0.129 synchronisation updates to this particular driver too. If I manage to get an inp file that works I'll upload it at a later date unless anyone else can get there first.
Steps To Reproduce
Additional Information
Github Commit

Flags
Regression Version
Affected Sets / Systems	thunderj

Attached Files

No.03651 Fujix Administrator Jan 26, 2009, 10:21	I played for some times to the middle of the 3rd stage, but no repro. I set the report Acknowledged for now.
No.03652 Machone Tester Jan 26, 2009, 10:52	Ok. I should mention that I've also reproduced this behaviour on another computer which is quite different to mine hardware-wise. The only thing they have in common is they both have XP installed. If anyone could confirm this bug it would be useful (I have a feeling it may be more likely to happen after you play, lose your lives, then play again).
No.11884 system11 Senior Tester Jul 19, 2015, 15:11	It's possible the PCB does this - I just had mine crash while running the RAM test.
No.15316 Machone Tester Aug 7, 2018, 17:04	Just a long overdue update: the MAME source code for this game mentions that there's a routine running on the second 68000 CPU (at $1E56) that checks to see if interrupts are being processed at the correct time. Apparently, there is a hack in MAME to give the game the result it expects. If the game detects anything wrong, it starts to corrupt memory locations. Anyway, the hack isn't working - the game gets messed up (not able to pick up objects, high score corruption etc.) and eventually the watchdog kicks in and resets the game. You have to play a few times to experience this, it never seems to happen on the first game & sometimes you can play for a reasonable amount of time before it crashes. I tried making an inp file but it goes haywire on playback so is useless. I worked around the problem by patching the ROMs to jump over the routine at $1E56, and have yet to experience a crash; not an ideal situation but it seems to work fine.
No.15318 NekoEd Senior Tester Aug 9, 2018, 06:27	Given new information I'm going to confirm this. It would appear we have improperly emulated protection.