07855: roadblst: Random occasional graphics corruption

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
07855	Graphics	Major	Random	Jan 13, 2021, 21:20	Sep 1, 2022, 09:17

Tester	Miles	View Status	Public	Platform	MAME (Official Binary)
Assigned To	hap	Resolution	Fixed	OS	Windows 10 (64-bit)
Status [?]	Resolved			Driver	atari/atarisy1.cpp
Version	0.227	Fixed in Version	0.248	Build	64-bit
		Fixed in Git Commit	f6942dd	Github Pull Request #

Summary	07855: roadblst: Random occasional graphics corruption
Description	The status bar at the bottom of the screen starts to move up and to the right, revealing a "stripy" road underneath. Video attached.
Steps To Reproduce	Seems to be random, but may be related to the appearance of the drone that drops auxiliary weapons.
Additional Information
Github Commit

Flags
Regression Version
Affected Sets / Systems	roadblst

Attached Files	roadblst-bug.mp4 (316,908 bytes) Jan 13, 2021, 21:20 Uploaded by Miles
	roadblst227.zip (252,675 bytes) Jan 14, 2021, 20:33 Uploaded by Tafoid roadblst - 0.227 - Recording
	roadblst200421.zip (449,145 bytes) Jan 14, 2021, 23:36 Uploaded by Tafoid roadblst - 200421 - Recording
	roadblst200422.zip (240,965 bytes) Jan 14, 2021, 23:36 Uploaded by Tafoid roadblst - 200422 - Recording
	rb2.zip (14,661 bytes) Aug 30, 2022, 21:35 Uploaded by Haze MAME 0.246 / just before 0.247 release repro

No.18371 Tafoid Administrator Jan 14, 2021, 01:30	I've had this happened to me but I had not had the time to whip together an .inp or video illustrating it. Thanks!
No.18380 Haze Senior Tester Jan 14, 2021, 19:53 edited on: Jan 14, 2021, 19:53	is this another slapstic related issue? from the 227 rewrite, or did it occur before?
No.18382 Tafoid Administrator Jan 14, 2021, 20:08	I'll have to do some testing. Likely prior to slapstic stuff because (unless I'm misremembering), it happened to me many months ago as well. I only checked it on 0.227 because of the other failures and surely encountered it.
No.18383 Tafoid Administrator Jan 14, 2021, 20:35	I'd uploaded an .inp which has it the video issues showing at least 3 times I noticed. 2nd Round after the checkpoint and before the end. 4th Round after the checkpoint (near round end), after continuing. I may take some time to try various versions to see if I can ID a breaking one.
No.18384 Tafoid Administrator Jan 14, 2021, 23:35	I've done some testing and I think I've narrowed down the commit that seems to be causing this: https://github.com/mamedev/mame/commit/15386236cf26603d17717a130b9ae61b2f1b3864 I have 2 playbacks from binaries made on April 21 and April 22 (uploaded). A REMINDER Delete .nvram at each attempt to playback/record The prior one (roadblst200421.inp) can be played with 0.220 release and even after 8 levels and a few continues, I could not get the issue to show. The other one (roadblst200422.inp) can be played with my 200422 binary, at least, and shows malfunction. I've made the binaries for each day available for testing and linked below in case you want to use playbacks or do further confirmation. Malfunction after checkpoint 2nd Round as well as 2 instances back to back in 5th Round before end of round. https://mameload.mameworld.info/mame64-200421-121aa2fe370fa0df17f0f24827db4f0ff0409b78.7z https://mameload.mameworld.info/mame64-200422-996ec14b8395fb34bc861551777513b5f99c25db.7z
No.18443 AJR Developer Feb 8, 2021, 14:24	I'm having trouble reproducing this issue on my current build.
No.20493 Haze Senior Tester Aug 30, 2022, 20:19	It's confirmed to still happen in 0.246 https://www.reddit.com/r/MAME/comments/x1sdka/video_problem_with_the_one_game_i_play/
No.20494 Haze Senior Tester Aug 30, 2022, 21:36	you can repro it almost immediately anyway, I just uploaded another inp created with current, rb2.zip
No.20495 hap Developer Aug 30, 2022, 22:01	The commit Tafoid links, it changed some drivers from using an Atari sound comms glue device to generic latch devices. The sound comms device adds a boost interleave after writing to the latch, the generic one does not. Try again: https://github.com/mamedev/mame/commit/f6942dd17a67d7db3c2eb3a2bc43552c6b8bca0d
No.20498 Haze Senior Tester Aug 30, 2022, 22:12 edited on: Aug 30, 2022, 22:14	ah, yeah, just saw your comment hap, it's definitely going to be that then, as anything that increases the interleave level seems to fix the issue. I'm not sure why code to boost the interleave was marked as a 'hack' though, as a real board would run in perfect interleave, MAME not running in lock-step by default is the hack for obvious performance reasons.
No.20500 Haze Senior Tester Aug 31, 2022, 13:42	anyway, I'm 99% sure this is fixed now