06440: Most/all sets in segag80r.cpp: game resets itself during play

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
06440	Gameplay	Critical (emulation)	Always	Nov 29, 2016, 20:28	Dec 13, 2016, 06:51

Tester	haynor666	View Status	Public	Platform	MAME (Official Binary)
Assigned To	smf	Resolution	Fixed	OS	Windows Vista/7/8 (64-bit)
Status [?]	Resolved			Driver	sega/segag80r.cpp
Version	0.179	Fixed in Version	0.181	Build	64-bit
		Fixed in Git Commit		Github Pull Request #

Summary	06440: Most/all sets in segag80r.cpp: game resets itself during play
Description	Quite offen game either resets itself during gameplay or generates trash on screen (sprites stay on screen).
Steps To Reproduce	Just start game and play couple minutes.
Additional Information	Specific issues: 005, monsterb - When playing a game, at odd times, the game suddenly resets astrob - If the demo runs for about 11 minutes, in testing, the screen goes black and emulation halts spaceod - Let emulation run for a bit over 5 minutes and the screen will stop as shown in screenshot sindbadm - At random times during play, it will either reset or freeze emulation while music plays
Github Commit

Flags
Regression Version	0.179
Affected Sets / Systems	Most/all sets in segag80r.cpp

Attached Files	spaceod.png (5,114 bytes) Nov 29, 2016, 20:59 Uploaded by Robbbert

No.13439 Robbbert Senior Tester Nov 29, 2016, 21:00	I left it in attract mode and after a while the attached picture occurred. The background continued to slide along, but the rest was frozen.
No.13441 B2K24 Senior Tester Nov 30, 2016, 01:59	Fairly easily to repo and takes anywhere from 1 - 2 minutes after coining up and starting up a fresh run.
No.13442 Tafoid Administrator Nov 30, 2016, 05:40 edited on: Dec 1, 2016, 20:22	Appears to start malfunctioning immediately after 0.178 was released. Traced to z80 adjustments.
No.13446 Osso Moderator Dec 1, 2016, 13:03	Putting back the PRVPC = -1 line in the void z80_device::take_interrupt() method "fixes" the problems. It was removed with this commit: https://git.redump.net/mame/commit/?id=2a6cbb70c4a8b2cdd6dbf6488cee92bd60fcc55e . I tested spaceod since it's the one which reproduces consistently letting attract run 3 times I guess it doesn't like the way the offs_t segag80r_state::decrypt_offset(address_space &space, offs_t offset) method is implemented.
No.13453 Tafoid Administrator Dec 1, 2016, 20:36	Actually it is from that code, but not that commit, as it was reverted here, The update was posted just prior to release with unknown consequences, so reverted. https://github.com/mamedev/mame/commit/7f1bb873d19bc81907e800b2ab6ed00e82a79f28 The commit I mentioned earlier is the recommit after the release of 0.178 was done. https://github.com/mamedev/mame/commit/00f6256e4a761b07d31b76ef30dca5e4deecc98a Your observations still seem valid, though. I'll try to let smf know to see if he has any ideas.