Viewing Issue Advanced Details
ID Category [?] Severity [?] Reproducibility Date Submitted Last Update
07357 Crash/Freeze Critical (emulation) Always Jun 12, 2019, 01:18 21 days ago
Tester Hydreigon View Status Public Platform MAME (Official Binary)
Assigned To AJR Resolution Fixed OS Windows 10 (64-bit)
Status [?] Resolved Driver taitojc.cpp
Version 0.210 Fixed in Version 0.212GIT Build 64-bit
Summary 07357: All sets in taitojc.cpp: Reset loop after startup screen
Description The games will watchdog reset right after the startup screen completes in an infinite loop. All games would at least start up ingame in earlier versions (last tested with 0.201).

Tested with sidebs2 and dangcurv.
Steps To Reproduce
Additional Information
Flags
Regression Version 0.210
Affected Sets / Systems All sets in taitojc.cpp
Attached Files
 
Relationships
There are no relationship linked to this issue.
Notes
5
User avatar
No.16527
Hydreigon
Tester
Jun 12, 2019, 01:33
edited on: 22 days ago
Link to (incorrect) commit removed per Tafoid's statement.
User avatar
No.16528
Tafoid
Administrator
Jun 12, 2019, 02:21
Your assumption is incorrect, unfortunately. My research shows it works in 0.209 (tested dendego), so your commit linked would have already happened.
Actual breakage seems to stem, in testing, around the big core M68000 reconstruction:
https://github.com/mamedev/mame/commit/f759f41db963169ac0da9bf472efbb647ee9baad

Unless for some reason there are multiple issues, I only chose one set because you've claimed "all sets in".
User avatar
No.16532
Hydreigon
Tester
Jun 12, 2019, 13:15
I updated my report stating the two games I tested that did this behavior. I'm kind of curious if other games that used an m68k cpu in different drivers broke from the big rewrite?
User avatar
No.16533
Tafoid
Administrator
Jun 12, 2019, 13:51
I'd like to hope most issues were caught in regression testing shortly after the initial commit. That's my main task for MAME - performing all sorts of regression testing largely looking for changes of video/audio behavior of the first 10 seconds of an emulated run where most issues tend to pop up, logging snapshot and wavdata hashes for comparisons to previous tests. Sadly, stuff like this which is longer into the emulation can only be found through manual checking.
User avatar
No.16593
AJR
Developer
22 days ago
The point at which dangcurv runs off the rails (not dendego's rails) in 0.210 and later appears to be the CINVA DATA instruction at PC = 00082724. This results in an F-line exception which shouldn't occur here.